--- _id: '17411' abstract: - lang: eng text: "We present PeFLL, a new personalized federated learning algorithm that improves\r\nover the state-of-the-art in three aspects: 1) it produces more accurate models,\r\nespecially in the low-data regime, and not only for clients present during its\r\ntraining phase, but also for any that may emerge in the future; 2) it reduces the\r\namount of on-client computation and client-server communication by providing\r\nfuture clients with ready-to-use personalized models that require no additional\r\nfinetuning or optimization; 3) it comes with theoretical guarantees that establish\r\ngeneralization from the observed clients to future ones.\r\nAt the core of PeFLL lies a learning-to-learn approach that jointly trains an\r\nembedding network and a hypernetwork. The embedding network is used to\r\nrepresent clients in a latent descriptor space in a way that reflects their similarity\r\nto each other. The hypernetwork takes as input such descriptors and outputs the\r\nparameters of fully personalized client models. In combination, both networks\r\nconstitute a learning algorithm that achieves state-of-the-art performance in several\r\npersonalized federated learning benchmarks" acknowledged_ssus: - _id: ScienComp acknowledgement: "This research was supported by the Scientific Service Units (SSU) of ISTA through resources provided by Scientific Computing (SciComp).\r\n" article_processing_charge: No arxiv: 1 author: - first_name: Jonathan A full_name: Scott, Jonathan A id: e499926b-f6e0-11ea-865d-9c63db0031e8 last_name: Scott - first_name: Hossein full_name: Zakerinia, Hossein id: 653bd8b6-f394-11eb-9cf6-c0bbf6cd78d4 last_name: Zakerinia orcid: 0009-0007-3977-6462 - first_name: Christoph full_name: Lampert, Christoph id: 40C20FD2-F248-11E8-B48F-1D18A9856A87 last_name: Lampert orcid: 0000-0001-8622-7887 citation: ama: 'Scott JA, Zakerinia H, Lampert C. PEFLL: Personalized federated learning by learning to learn. In: 12th International Conference on Learning Representations. OpenReview; 2024.' apa: 'Scott, J. A., Zakerinia, H., & Lampert, C. (2024). PEFLL: Personalized federated learning by learning to learn. In 12th International Conference on Learning Representations. Vienna, Austria: OpenReview.' chicago: 'Scott, Jonathan A, Hossein Zakerinia, and Christoph Lampert. “PEFLL: Personalized Federated Learning by Learning to Learn.” In 12th International Conference on Learning Representations. OpenReview, 2024.' ieee: 'J. A. Scott, H. Zakerinia, and C. Lampert, “PEFLL: Personalized federated learning by learning to learn,” in 12th International Conference on Learning Representations, Vienna, Austria, 2024.' ista: 'Scott JA, Zakerinia H, Lampert C. 2024. PEFLL: Personalized federated learning by learning to learn. 12th International Conference on Learning Representations. ICLR: International Conference on Learning Representations.' mla: 'Scott, Jonathan A., et al. “PEFLL: Personalized Federated Learning by Learning to Learn.” 12th International Conference on Learning Representations, OpenReview, 2024.' short: J.A. Scott, H. Zakerinia, C. Lampert, in:, 12th International Conference on Learning Representations, OpenReview, 2024. conference: end_date: 2024-03-07 location: Vienna, Austria name: 'ICLR: International Conference on Learning Representations' start_date: 2024-03-07 corr_author: '1' date_created: 2024-08-11T22:01:12Z date_published: 2024-03-07T00:00:00Z date_updated: 2026-04-07T11:46:11Z day: '07' ddc: - '000' department: - _id: ChLa external_id: arxiv: - '2306.05515' file: - access_level: open_access checksum: 81b7ea2e667adaf9c7a7b6b376b1f251 content_type: application/pdf creator: dernst date_created: 2024-08-12T07:38:06Z date_updated: 2024-08-12T07:38:06Z file_id: '17415' file_name: 2024_ICLR_Scott.pdf file_size: 1029219 relation: main_file success: 1 file_date_updated: 2024-08-12T07:38:06Z has_accepted_license: '1' language: - iso: eng month: '03' oa: 1 oa_version: Published Version publication: 12th International Conference on Learning Representations publication_status: published publisher: OpenReview quality_controlled: '1' related_material: record: - id: '21198' relation: dissertation_contains status: public scopus_import: '1' status: public title: 'PEFLL: Personalized federated learning by learning to learn' type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2024' ... --- OA_place: publisher _id: '17490' abstract: - lang: eng text: "Deep learning is essential in numerous applications nowadays, with many recent advancements made possible by training very large models. Despite their broad applicability, training neural networks is often time-intensive, and it is usually impractical to manage large models and datasets on a single machine. To address these issues, distributed deep learning training has become increasingly important. However, distributed training requires synchronization among nodes, and the mini-batch stochastic gradient descent algorithm places a significant load on network connections. A possible solution to tackle the synchronization bottleneck is to reduce a message size by lossy compression.\r\n\r\nIn this thesis, we investigate systems and algorithmic approaches to communication compression during training. From the systems perspective, we demonstrate that a common approach of expensive hardware overprovisioning can be replaced through a thorough system design. We introduce a framework that introduces efficient software support for compressed communication in machine learning applications, applicable to both multi-GPU single-node training and larger-scale multi-node training. Our framework integrates with popular ML frameworks, providing up to 3x speedups for multi-GPU nodes based on commodity hardware and order-of-magnitude improvements in the multi-node setting, with negligible impact on accuracy.\r\n\r\nAlso, we consider an application of our framework to different communication schemes, such as Fully Sharded Data Parallel. We provide strong convergence guarantees for the compression in such a setup. Empirical validation shows that our method preserves model accuracy for GPT-family models with up to 1.3 billion parameters, while completely removing the communication bottlenecks of non-compressed alternatives, providing up to 2.2x speedups end-to-end.\r\n\r\nFrom the algorithmic side, we propose a general framework that dynamically adjusts the degree of compression across a model's layers during training. This approach enhances overall compression and results in significant speedups without compromising accuracy. Our algorithm utilizes an adaptive algorithm that automatically selects the optimal compression parameters for model layers, ensuring the best compression ratio while adhering to an error constraint. Our method is effective across all existing families of compression methods. It achieves up to 2.5x faster training and up to a 5x improvement in compression compared to efficient implementations of current approaches. Additionally, LGreCo can complement existing adaptive algorithms.\r\n" acknowledged_ssus: - _id: ScienComp alternative_title: - ISTA Thesis article_processing_charge: No author: - first_name: Ilia full_name: Markov, Ilia id: D0CF4148-C985-11E9-8066-0BDEE5697425 last_name: Markov citation: ama: 'Markov I. Communication-efficient distributed training of deep neural networks : An algorithms and systems perspective. 2024. doi:10.15479/at:ista:17490' apa: 'Markov, I. (2024). Communication-efficient distributed training of deep neural networks : An algorithms and systems perspective. Institute of Science and Technology Austria. https://doi.org/10.15479/at:ista:17490' chicago: 'Markov, Ilia. “Communication-Efficient Distributed Training of Deep Neural Networks : An Algorithms and Systems Perspective.” Institute of Science and Technology Austria, 2024. https://doi.org/10.15479/at:ista:17490.' ieee: 'I. Markov, “Communication-efficient distributed training of deep neural networks : An algorithms and systems perspective,” Institute of Science and Technology Austria, 2024.' ista: 'Markov I. 2024. Communication-efficient distributed training of deep neural networks : An algorithms and systems perspective. Institute of Science and Technology Austria.' mla: 'Markov, Ilia. Communication-Efficient Distributed Training of Deep Neural Networks : An Algorithms and Systems Perspective. Institute of Science and Technology Austria, 2024, doi:10.15479/at:ista:17490.' short: 'I. Markov, Communication-Efficient Distributed Training of Deep Neural Networks : An Algorithms and Systems Perspective, Institute of Science and Technology Austria, 2024.' corr_author: '1' date_created: 2024-09-04T08:51:11Z date_published: 2024-09-04T00:00:00Z date_updated: 2026-06-18T17:55:23Z day: '04' ddc: - '000' degree_awarded: PhD department: - _id: GradSch - _id: DaAl doi: 10.15479/at:ista:17490 ec_funded: 1 file: - access_level: closed checksum: 77609f4835d2730e46fa0d42d9134ed9 content_type: application/x-zip-compressed creator: imarkov date_created: 2024-09-04T08:35:35Z date_updated: 2024-09-04T08:35:35Z file_id: '17491' file_name: Thesis.zip file_size: 43327753 relation: source_file - access_level: open_access checksum: 9e68f7217570f756ceb8f70b980938cd content_type: application/pdf creator: imarkov date_created: 2024-09-04T08:36:06Z date_updated: 2024-09-04T08:36:06Z file_id: '17492' file_name: Thesis_final_version_pdfa2.pdf file_size: 2756082 relation: main_file success: 1 file_date_updated: 2024-09-04T08:36:06Z has_accepted_license: '1' language: - iso: eng license: https://creativecommons.org/licenses/by-nc-sa/4.0/ month: '09' oa: 1 oa_version: Published Version page: '102' project: - _id: 268A44D6-B435-11E9-9278-68D0E5697425 call_identifier: H2020 grant_number: '805223' name: Elastic Coordination for Scalable Machine Learning publication_identifier: issn: - 2663-337X publication_status: published publisher: Institute of Science and Technology Austria related_material: record: - id: '14461' relation: part_of_dissertation status: public - id: '12780' relation: part_of_dissertation status: public - id: '17456' relation: part_of_dissertation status: public status: public supervisor: - first_name: Dan-Adrian full_name: Alistarh, Dan-Adrian id: 4A899BFC-F248-11E8-B48F-1D18A9856A87 last_name: Alistarh orcid: 0000-0003-3650-940X title: 'Communication-efficient distributed training of deep neural networks : An algorithms and systems perspective' tmp: image: /images/cc_by_nc_sa.png legal_code_url: https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode name: Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) short: CC BY-NC-SA (4.0) type: dissertation user_id: ba8df636-2132-11f1-aed0-ed93e2281fdd year: '2024' ... --- OA_place: publisher _id: '17850' abstract: - lang: eng text: "Understanding the relationship between a given phenotype and its underlying genotype or genotypes is one of the most pressing challenges of biology, as it lies at the heart of not only basic understanding of evolutionary theory, but also of practical applications in medicine and bioengineering. Understanding this relationship is complicated by the ubiquitous phenomenon of epistasis, wherein mutation effects are dependent on their genetic context. Fitness landscapes — representations of phenotype as a function of genotype — are being increasingly used as a tool to study the effects and interactions of thousands of mutations, but are experimentally limited to exploring a small fraction of a protein’s theoretical sequence space. Furthermore, not all regions of said sequence space are necessarily equally informative. Thus, gene selection for landscape surveys should be carefully considered in order to maximize the usable output of necessarily limited data.\r\n\r\nIn this work, we analyzed the fitness landscapes of orthologous green fluorescent proteins from four different species, by systematically measuring the phenotype, fluorescence, of tens of thousands of mutant genotypes from each protein. These landscapes were highly heterogeneous, with some genes being mutationally robust and displaying epistasis only rarely, and others being highly epistatic and mutationally fragile. We used this data to train machine learning models to predict fluorescence from genotype. Although the training data contained almost exclusively genotypes with less than 3% sequence divergence from the original wild-type sequences, we were able to create novel, functional genotypes with up to 20% sequence divergence. Counterintuitively however, genes with high mutational robustness and rare epistasis were more difficult to introduce large numbers of mutations into, not less. This represents the first study of large-scale fitness landscapes of a protein family, and provides insights into how to approach future landscape surveys and their applications in novel protein design." acknowledged_ssus: - _id: Bio - _id: LifeSc - _id: ScienComp alternative_title: - ISTA Thesis article_processing_charge: No author: - first_name: Louisa full_name: Gonzalez Somermeyer, Louisa id: 4720D23C-F248-11E8-B48F-1D18A9856A87 last_name: Gonzalez Somermeyer orcid: 0000-0001-9139-5383 citation: ama: Gonzalez Somermeyer L. Fitness landscapes of orthologous green fluorescent proteins. 2024. doi:10.15479/at:ista:17850 apa: Gonzalez Somermeyer, L. (2024). Fitness landscapes of orthologous green fluorescent proteins. Institute of Science and Technology Austria. https://doi.org/10.15479/at:ista:17850 chicago: Gonzalez Somermeyer, Louisa. “Fitness Landscapes of Orthologous Green Fluorescent Proteins.” Institute of Science and Technology Austria, 2024. https://doi.org/10.15479/at:ista:17850. ieee: L. Gonzalez Somermeyer, “Fitness landscapes of orthologous green fluorescent proteins,” Institute of Science and Technology Austria, 2024. ista: Gonzalez Somermeyer L. 2024. Fitness landscapes of orthologous green fluorescent proteins. Institute of Science and Technology Austria. mla: Gonzalez Somermeyer, Louisa. Fitness Landscapes of Orthologous Green Fluorescent Proteins. Institute of Science and Technology Austria, 2024, doi:10.15479/at:ista:17850. short: L. Gonzalez Somermeyer, Fitness Landscapes of Orthologous Green Fluorescent Proteins, Institute of Science and Technology Austria, 2024. corr_author: '1' date_created: 2024-09-06T12:57:44Z date_published: 2024-09-06T00:00:00Z date_updated: 2026-04-07T13:25:01Z day: '06' ddc: - '570' degree_awarded: PhD department: - _id: GradSch - _id: FyKo doi: 10.15479/at:ista:17850 ec_funded: 1 file: - access_level: open_access checksum: d3303724e8d3c91321d71bbad4062048 content_type: application/pdf creator: lgonzale date_created: 2024-09-27T10:32:33Z date_updated: 2024-09-27T10:32:33Z file_id: '18151' file_name: louisa_thesis_draft__240904b.pdf file_size: 11219837 relation: main_file - access_level: closed checksum: 22e63f7f9014dffde2af7a47e7d1d014 content_type: application/vnd.openxmlformats-officedocument.wordprocessingml.document creator: lgonzale date_created: 2024-09-27T10:34:34Z date_updated: 2024-09-27T10:34:34Z file_id: '18152' file_name: louisa_thesis_draft__240904b.docx file_size: 43338677 relation: source_file file_date_updated: 2024-09-27T10:34:34Z has_accepted_license: '1' language: - iso: eng month: '09' oa: 1 oa_version: Published Version page: '89' project: - _id: 2564DBCA-B435-11E9-9278-68D0E5697425 call_identifier: H2020 grant_number: '665385' name: International IST Doctoral Program - _id: 26580278-B435-11E9-9278-68D0E5697425 call_identifier: H2020 grant_number: '771209' name: Characterizing the fitness landscape on population and global scales publication_identifier: issn: - 2663-337X publication_status: published publisher: Institute of Science and Technology Austria related_material: link: - relation: software url: https://github.com/aequorea238/Orthologous_GFP_Fitness_Peaks record: - id: '11448' relation: part_of_dissertation status: public status: public supervisor: - first_name: Fyodor full_name: Kondrashov, Fyodor id: 44FDEF62-F248-11E8-B48F-1D18A9856A87 last_name: Kondrashov orcid: 0000-0001-8243-4694 title: Fitness landscapes of orthologous green fluorescent proteins tmp: image: /images/cc_by_nc_nd.png legal_code_url: https://creativecommons.org/licenses/by-nc-nd/4.0/legalcode name: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) short: CC BY-NC-ND (4.0) type: dissertation user_id: ba8df636-2132-11f1-aed0-ed93e2281fdd year: '2024' ... --- OA_place: publisher _id: '17485' abstract: - lang: eng text: "Large language models (LLMs) have made tremendous progress in the past few years, from being able to generate coherent text to matching or surpassing humans in a wide variety of creative, knowledge or reasoning tasks. Much of this can be attributed to massively increased scale, both in the size of the model as well as the amount of training data, from 100s of millions to 100s of billions, or even trillions. This trend is expected to continue, which, although exciting, also raises major practical concerns. Already today's 100+ billion parameter LLMs require top-of-the-line hardware just to run. Hence, it is clear that sustaining these developments will require significant efficiency advances.\r\n\r\nHistorically, one of the most practical ways of improving model efficiency has been compression, especially in the form of sparsity or quantization. While this has been studied extensively in the past, existing accurate methods are all designed for models around 100 million parameters; scaling them up to ones literally 1000x larger is highly challenging. In this thesis, we introduce a new unified sparsification and quantization approach OBC, which through additional algorithmic enhancements leads to GPTQ and SparseGPT, the first techniques fast and accurate enough to compress 100+ billion parameter models to 4- or even 3-bit precision and 50% weight-sparsity, respectively. Additionally, we show how weight-only quantizion does not just bring space savings but also up to 4.5x faster generation speed, via custom GPU kernels.\r\n\r\nIn fact, we show for the first time that it is possible to develop an FP16 times INT4 mixed-precision matrix multiplication kernel, called Marlin, which comes close to simultaneously maximizing both memory and compute utilization, making weight-only quantization highly practical even for multi-user serving. Further, we demonstrate that GPTQ can be scaled to widely overparametrized trillion-parameter models, where extreme sub-1-bit compression rates can be achieved without any inference slow-down, by co-designing a bespoke entropy coding scheme together with an efficient kernel.\r\n\r\nFinally, we also study compression from the perspective of someone with access to massive amounts of compute resources for training large models completely from scratch. Here the key questions evolve around the joint scaling behavior between compression, model size, and amount of training data used. Based on extensive experimental results for both vision and text models, we introduce the first scaling law which accurately captures the relationship between weight-sparsity, number of non-zero weights and data. This further allows us to characterize the optimal sparsity, which we find to increase the longer a fixed cost model is being trained.\r\n\r\nOverall, this thesis presents contributions to three different angles of large model efficiency: affordable but accurate algorithms, highly efficient systems implementations, and fundamental scaling laws for compressed training." acknowledged_ssus: - _id: ScienComp alternative_title: - ISTA Thesis article_processing_charge: No author: - first_name: Elias full_name: Frantar, Elias id: 09a8f98d-ec99-11ea-ae11-c063a7b7fe5f last_name: Frantar citation: ama: 'Frantar E. Compressing large neural networks : Algorithms, systems and scaling laws. 2024. doi:10.15479/at:ista:17485' apa: 'Frantar, E. (2024). Compressing large neural networks : Algorithms, systems and scaling laws. Institute of Science and Technology Austria. https://doi.org/10.15479/at:ista:17485' chicago: 'Frantar, Elias. “Compressing Large Neural Networks : Algorithms, Systems and Scaling Laws.” Institute of Science and Technology Austria, 2024. https://doi.org/10.15479/at:ista:17485.' ieee: 'E. Frantar, “Compressing large neural networks : Algorithms, systems and scaling laws,” Institute of Science and Technology Austria, 2024.' ista: 'Frantar E. 2024. Compressing large neural networks : Algorithms, systems and scaling laws. Institute of Science and Technology Austria.' mla: 'Frantar, Elias. Compressing Large Neural Networks : Algorithms, Systems and Scaling Laws. Institute of Science and Technology Austria, 2024, doi:10.15479/at:ista:17485.' short: 'E. Frantar, Compressing Large Neural Networks : Algorithms, Systems and Scaling Laws, Institute of Science and Technology Austria, 2024.' corr_author: '1' date_created: 2024-09-02T11:01:48Z date_published: 2024-09-05T00:00:00Z date_updated: 2026-07-07T13:22:38Z day: '05' ddc: - '000' degree_awarded: PhD department: - _id: GradSch - _id: DaAl doi: 10.15479/at:ista:17485 ec_funded: 1 file: - access_level: closed checksum: 5d785645805a78c5b4ce7cc3df557b09 content_type: application/zip creator: efrantar date_created: 2024-09-05T12:04:11Z date_updated: 2024-09-05T12:04:11Z file_id: '17570' file_name: thesis-final.zip file_size: 1615167 relation: source_file - access_level: open_access checksum: a9dd1c2d23734986924eb44ebb55fd8f content_type: application/pdf creator: efrantar date_created: 2024-09-06T16:24:59Z date_updated: 2024-09-06T16:24:59Z file_id: '17880' file_name: frantar_thesis_final.pdf file_size: 2376611 relation: main_file success: 1 file_date_updated: 2024-09-06T16:24:59Z has_accepted_license: '1' language: - iso: eng month: '09' oa: 1 oa_version: Published Version page: '129' project: - _id: 268A44D6-B435-11E9-9278-68D0E5697425 call_identifier: H2020 grant_number: '805223' name: Elastic Coordination for Scalable Machine Learning publication_identifier: issn: - 2663-337X publication_status: published publisher: Institute of Science and Technology Austria related_material: record: - id: '17378' relation: part_of_dissertation status: public - id: '17087' relation: part_of_dissertation status: public - id: '14458' relation: part_of_dissertation status: public - id: '18062' relation: part_of_dissertation status: public - id: '18061' relation: part_of_dissertation status: public status: public supervisor: - first_name: Dan-Adrian full_name: Alistarh, Dan-Adrian id: 4A899BFC-F248-11E8-B48F-1D18A9856A87 last_name: Alistarh orcid: 0000-0003-3650-940X title: 'Compressing large neural networks : Algorithms, systems and scaling laws' type: dissertation user_id: ba8df636-2132-11f1-aed0-ed93e2281fdd year: '2024' ... --- OA_place: repository OA_type: green _id: '17147' abstract: - lang: eng text: Efficient utilization of large-scale biobank data is crucial for inferring the genetic basis of disease and predicting health outcomes from the DNA. Yet we lack efficient, accurate methods that scale to data where electronic health records are linked to whole genome sequence information. To address this issue, our paper develops a new algorithmic paradigm based on Approximate Message Passing (AMP), which is specifically tailored for genomic prediction and association testing. Our method yields comparable out-of-sample prediction accuracy to the state of the art on UK Biobank traits, whilst dramatically improving computational complexity, with a 8x-speed up in the run time. In addition, AMP theory provides a joint association testing framework, which outperforms the currently used REGENIE method, in roughly a third of the compute time. This first, truly large-scale application of the AMP framework lays the foundations for a far wider range of statistical analyses for hundreds of millions of variables measured on millions of people. acknowledged_ssus: - _id: ScienComp acknowledgement: "This work was supported by a Lopez-Loreta Prize to MM, an SNSF Eccellenza Grant to MRR (PCEGP3-181181), and core funding from ISTA. The authors thank Philip Schniter, Matthew Stephens and Pragya Sur for valuable suggestions on an early version of the work. The authors acknowledge the participants and investigators of the UK Biobank study. High-performance\r\ncomputing was supported by the Scientific Service Units (SSU) of IST Austria through resources provided by Scientific Computing (SciComp)." article_processing_charge: No author: - first_name: Al full_name: Depope, Al id: 0b77531d-dbcd-11ea-9d1d-a8eee0bf3830 last_name: Depope - first_name: Marco full_name: Mondelli, Marco id: 27EB676C-8706-11E9-9510-7717E6697425 last_name: Mondelli orcid: 0000-0002-3242-7020 - first_name: Matthew Richard full_name: Robinson, Matthew Richard id: E5D42276-F5DA-11E9-8E24-6303E6697425 last_name: Robinson orcid: 0000-0001-8982-8813 citation: ama: 'Depope A, Mondelli M, Robinson MR. Inference of genetic effects via approximate message passing. In: 2024 IEEE International Conference on Acoustics, Speech, and Signal Processing. IEEE; 2024:13151-13155. doi:10.1109/ICASSP48485.2024.10447198' apa: 'Depope, A., Mondelli, M., & Robinson, M. R. (2024). Inference of genetic effects via approximate message passing. In 2024 IEEE International Conference on Acoustics, Speech, and Signal Processing (pp. 13151–13155). Seoul, Korea: IEEE. https://doi.org/10.1109/ICASSP48485.2024.10447198' chicago: Depope, Al, Marco Mondelli, and Matthew Richard Robinson. “Inference of Genetic Effects via Approximate Message Passing.” In 2024 IEEE International Conference on Acoustics, Speech, and Signal Processing, 13151–55. IEEE, 2024. https://doi.org/10.1109/ICASSP48485.2024.10447198. ieee: A. Depope, M. Mondelli, and M. R. Robinson, “Inference of genetic effects via approximate message passing,” in 2024 IEEE International Conference on Acoustics, Speech, and Signal Processing, Seoul, Korea, 2024, pp. 13151–13155. ista: 'Depope A, Mondelli M, Robinson MR. 2024. Inference of genetic effects via approximate message passing. 2024 IEEE International Conference on Acoustics, Speech, and Signal Processing. ICASSP: International Conference on Acoustics, Speech and Signal Processing, 13151–13155.' mla: Depope, Al, et al. “Inference of Genetic Effects via Approximate Message Passing.” 2024 IEEE International Conference on Acoustics, Speech, and Signal Processing, IEEE, 2024, pp. 13151–55, doi:10.1109/ICASSP48485.2024.10447198. short: A. Depope, M. Mondelli, M.R. Robinson, in:, 2024 IEEE International Conference on Acoustics, Speech, and Signal Processing, IEEE, 2024, pp. 13151–13155. conference: end_date: 2024-04-19 location: Seoul, Korea name: 'ICASSP: International Conference on Acoustics, Speech and Signal Processing' start_date: 2024-04-14 corr_author: '1' date_created: 2024-06-16T22:01:07Z date_published: 2024-04-19T00:00:00Z date_updated: 2026-07-13T14:57:55Z day: '19' department: - _id: MaMo - _id: MaRo doi: 10.1109/ICASSP48485.2024.10447198 external_id: isi: - '001396233806078' isi: 1 language: - iso: eng main_file_link: - open_access: '1' url: https://openreview.net/forum?id=aQYCDxfZV0 month: '04' oa: 1 oa_version: Submitted Version page: 13151-13155 project: - _id: 059876FA-7A3F-11EA-A408-12923DDC885E name: Prix Lopez-Loretta 2019 - Marco Mondelli - _id: 9B8D11D6-BA93-11EA-9121-9846C619BF3A grant_number: PCEGP3_181181 name: Improving estimation and prediction of common complex disease risk publication: 2024 IEEE International Conference on Acoustics, Speech, and Signal Processing publication_identifier: isbn: - '9798350344851' issn: - 1520-6149 publication_status: published publisher: IEEE quality_controlled: '1' scopus_import: '1' status: public title: Inference of genetic effects via approximate message passing type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2024' ... --- OA_place: publisher _id: '18471' abstract: - lang: eng text: "Spatial omics technologies are enriching our understanding of complex biological samples, by\r\nallowing us to study their molecular composition while preserving the spatial relationships\r\nbetween molecules in their native context. As the field continues to advance, there are\r\ntechnical challenges that need to be addressed in order to take full advantage of the spatial\r\ncapabilities of these methods. In this work, I present two technical developments that I\r\nestablished for multiplexed error robust FISH (MERFISH) throughout my PhD: (1) pushing the\r\nspatial resolution limits to the nanoscale, and (2) adding rich tissue context to the mouse brain\r\ntranscriptome. To achieve nanoscale resolution with MERFISH in cultured cells, I combined it\r\nwith stimulated emission depletion (STED) and expansion microscopy (ExM) to achieve a\r\nspatial resolution as low as ~20 nm, and explored the compatibility of MERFISH with singlemolecule localization microscopy (SMLM) techniques. To visualize targeted mRNAs in mouse\r\nbrain tissue, I applied the comprehensive analysis of tissues across scales (CATS) toolbox, which\r\nprovides an unbiased morphological readout by labeling the extracellular domain. I\r\nsuccessfully established this method, which we call CATS-MERFISH-ExM, to work with thick\r\nmouse brain slices, being able to extract transcriptomics information with 3D tissue context.\r\nCATS-MERFISH-ExM enabled us to identify cell types and further visualize the subcellular\r\ndistribution of transcripts in mouse brain tissue, shedding light on the neuropil-specific\r\ntranscriptome. This method provides integrated information on cellular structure and\r\ntranscriptomes in situ, and could potentially be applied with other modalities, opening new\r\navenues for scientific discovery. " acknowledged_ssus: - _id: Bio - _id: LifeSc - _id: PreCl - _id: M-Shop - _id: ScienComp alternative_title: - ISTA Thesis article_processing_charge: No author: - first_name: Nathalie full_name: Agudelo Duenas, Nathalie id: 40E7F008-F248-11E8-B48F-1D18A9856A87 last_name: Agudelo Duenas citation: ama: Agudelo Duenas N. Visualizing the neuronal transcriptional landscape with tissue context. 2024. doi:10.15479/at:ista:18471 apa: Agudelo Duenas, N. (2024). Visualizing the neuronal transcriptional landscape with tissue context. Institute of Science and Technology Austria. https://doi.org/10.15479/at:ista:18471 chicago: Agudelo Duenas, Nathalie. “Visualizing the Neuronal Transcriptional Landscape with Tissue Context.” Institute of Science and Technology Austria, 2024. https://doi.org/10.15479/at:ista:18471. ieee: N. Agudelo Duenas, “Visualizing the neuronal transcriptional landscape with tissue context,” Institute of Science and Technology Austria, 2024. ista: Agudelo Duenas N. 2024. Visualizing the neuronal transcriptional landscape with tissue context. Institute of Science and Technology Austria. mla: Agudelo Duenas, Nathalie. Visualizing the Neuronal Transcriptional Landscape with Tissue Context. Institute of Science and Technology Austria, 2024, doi:10.15479/at:ista:18471. short: N. Agudelo Duenas, Visualizing the Neuronal Transcriptional Landscape with Tissue Context, Institute of Science and Technology Austria, 2024. corr_author: '1' date_created: 2024-10-26T20:02:42Z date_published: 2024-10-28T00:00:00Z date_updated: 2026-04-14T08:34:37Z day: '28' ddc: - '570' degree_awarded: PhD department: - _id: GradSch - _id: JoDa doi: 10.15479/at:ista:18471 ec_funded: 1 file: - access_level: closed checksum: 6d7c7725d040d8debc070dcb35ac965d content_type: application/vnd.openxmlformats-officedocument.wordprocessingml.document creator: nagudelo date_created: 2024-10-26T22:29:06Z date_updated: 2025-05-05T22:30:04Z embargo_to: open_access file_id: '18475' file_name: PhD_thesis_Nathalie_Agudelo_Duenas_ISTA_final.docx file_size: 183077763 relation: source_file - access_level: open_access checksum: 52f9c0bf2bdafa3baf827b73814a53ff content_type: application/pdf creator: nagudelo date_created: 2024-10-26T23:13:33Z date_updated: 2025-05-05T22:30:04Z embargo: 2025-05-05 file_id: '18476' file_name: PhD_thesis_Nathalie_Agudelo_Duenas_ISTA_final.pdf file_size: 47027710 relation: main_file file_date_updated: 2025-05-05T22:30:04Z has_accepted_license: '1' language: - iso: eng month: '10' oa: 1 oa_version: Published Version page: '97' project: - _id: 2564DBCA-B435-11E9-9278-68D0E5697425 call_identifier: H2020 grant_number: '665385' name: International IST Doctoral Program - _id: 2548AE96-B435-11E9-9278-68D0E5697425 call_identifier: FWF grant_number: W1232 name: Molecular Drug Targets publication_identifier: isbn: - 978-3-99078-044-2 issn: - 2663-337X publication_status: published publisher: Institute of Science and Technology Austria status: public supervisor: - first_name: Johann G full_name: Danzl, Johann G id: 42EFD3B6-F248-11E8-B48F-1D18A9856A87 last_name: Danzl orcid: 0000-0001-8559-3973 title: Visualizing the neuronal transcriptional landscape with tissue context tmp: image: /images/cc_by.png legal_code_url: https://creativecommons.org/licenses/by/4.0/legalcode name: Creative Commons Attribution 4.0 International Public License (CC-BY 4.0) short: CC BY (4.0) type: dissertation user_id: ba8df636-2132-11f1-aed0-ed93e2281fdd year: '2024' ... --- OA_embargo: '12' OA_place: publisher _id: '18574' abstract: - lang: eng text: "Biological vision is unlike a camera; rather than transmitting light information faithfully, early\r\nvisual circuits process the visual scene to convey only the relevant information in an efficient\r\nmanner. Consequentially, the nature of this visual processing then depends on what is the\r\nrelevant information in a scene and on the notion of efficiency. In this work, I study how visual\r\nprocessing is modulated by two different variations in the visual scene. First, I discovered that\r\nin the mouse (Mus musculus) retina, Retinal Ganglion Cells in the upper and lower visual\r\nfield have differences in the center surround structure of their receptive fields. Comparison\r\nwith models of efficient coding show that this adaptation likely evolved to cope with the\r\nbrightness gradient from the sky to the ground that is pervasive in natural scenes. In the\r\nsecond project, I study how the downstream neurons in the Superior Colliculus dynamically\r\nchange their temporal selectivity depending on the ambient luminance and behavioral state.\r\nAs the scene gets darker or when the animal is is less aroused, the neuronal responses get\r\nlaggier, while still maintaining their relative timing with respect to the population. Overall, this\r\nwork emphasises the need to understand visual processing in the context of specific demands\r\nof the animal in its the environment. The adaptive changes in the visual system, from the\r\nretinal ganglion cells to the superior colliculus, highlight the intricate ways in which biological\r\nvision optimizes the processing of visual information.\r\n" acknowledged_ssus: - _id: Bio - _id: ScienComp - _id: PreCl - _id: LifeSc - _id: M-Shop - _id: E-Lib acknowledgement: "This work would have been impossible without the Scientific Service Units of IST Austria. The resources and expertise provided by Scientific Computing (especially Alois Schlögl), the MIBA Machine Shop (especially Todor Asenov), the Preclinical Facility (especially Freyja Langer), the Library, the Lab Support Facility and the Imaging and Optics Facility were the essential bedrock I could build upon. I would also like to thank IT support at ISTA for powering through remote work and a cyberattack.\r\nI am grateful for having been funded initially by the European Union Horizon 2020 Marie Skłodowska-Curie grant 665385 and later by Prof. Maximilian Joesch's the European Research Council Starting (756502) and Consolidator (101086580) Grants." alternative_title: - ISTA Thesis article_processing_charge: No author: - first_name: Divyansh full_name: Gupta, Divyansh id: 2A485EBE-F248-11E8-B48F-1D18A9856A87 last_name: Gupta orcid: 0000-0001-7400-6665 citation: ama: Gupta D. Visual adaptations to natural statistics. 2024. doi:10.15479/at:ista:18574 apa: Gupta, D. (2024). Visual adaptations to natural statistics. Institute of Science and Technology Austria. https://doi.org/10.15479/at:ista:18574 chicago: Gupta, Divyansh. “Visual Adaptations to Natural Statistics.” Institute of Science and Technology Austria, 2024. https://doi.org/10.15479/at:ista:18574. ieee: D. Gupta, “Visual adaptations to natural statistics,” Institute of Science and Technology Austria, 2024. ista: Gupta D. 2024. Visual adaptations to natural statistics. Institute of Science and Technology Austria. mla: Gupta, Divyansh. Visual Adaptations to Natural Statistics. Institute of Science and Technology Austria, 2024, doi:10.15479/at:ista:18574. short: D. Gupta, Visual Adaptations to Natural Statistics, Institute of Science and Technology Austria, 2024. corr_author: '1' date_created: 2024-11-20T21:30:44Z date_published: 2024-11-22T00:00:00Z date_updated: 2026-04-07T13:24:48Z day: '22' ddc: - '573' degree_awarded: PhD department: - _id: GradSch - _id: MaJö doi: 10.15479/at:ista:18574 ec_funded: 1 file: - access_level: closed checksum: ebb000d361c36b22ed6e639a931c6b7c content_type: application/zip creator: dgupta date_created: 2024-11-25T14:44:03Z date_updated: 2025-11-11T23:30:02Z embargo_to: open_access file_id: '18589' file_name: PhD Thesis - Divyansh Gupta.zip file_size: 75512262 relation: source_file - access_level: open_access checksum: 1282401eb71598bc311058b0fcefc6a1 content_type: application/pdf creator: dgupta date_created: 2024-11-26T11:43:19Z date_updated: 2025-11-11T23:30:02Z embargo: 2025-11-11 file_id: '18591' file_name: PDFA_PhD_Thesis___Divyansh_Gupta-26_11_24.pdf file_size: 6412619 relation: main_file file_date_updated: 2025-11-11T23:30:02Z has_accepted_license: '1' language: - iso: eng month: '11' oa: 1 oa_version: Published Version page: '86' project: - _id: bdaf81a8-d553-11ed-ba76-c95961984540 grant_number: '101086580' name: 'Action Selection in the Midbrain: Neuromodulation of Visuomotor Senses' - _id: 2564DBCA-B435-11E9-9278-68D0E5697425 call_identifier: H2020 grant_number: '665385' name: International IST Doctoral Program - _id: 2634E9D2-B435-11E9-9278-68D0E5697425 call_identifier: H2020 grant_number: '756502' name: Circuits of Visual Attention publication_identifier: isbn: - 978-3-99078-050-3 issn: - 2663-337X publication_status: published publisher: Institute of Science and Technology Austria related_material: record: - id: '12349' relation: part_of_dissertation status: public - id: '12370' relation: research_data status: public status: public supervisor: - first_name: Maximilian A full_name: Jösch, Maximilian A id: 2BD278E6-F248-11E8-B48F-1D18A9856A87 last_name: Jösch orcid: 0000-0002-3937-1330 title: Visual adaptations to natural statistics tmp: image: /images/cc_by_nc_sa.png legal_code_url: https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode name: Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) short: CC BY-NC-SA (4.0) type: dissertation user_id: ba8df636-2132-11f1-aed0-ed93e2281fdd year: '2024' ... --- _id: '15323' abstract: - lang: eng text: Supercomplexes of the respiratory chain are established constituents of the oxidative phosphorylation system, but their role in mammalian metabolism has been hotly debated. Although recent studies have shown that different tissues/organs are equipped with specific sets of supercomplexes, depending on their metabolic needs, the notion that supercomplexes have a role in the regulation of metabolism has been challenged. However, irrespective of the mechanistic conclusions, the composition of various high molecular weight supercomplexes remains uncertain. Here, using cryogenic electron microscopy, we demonstrate that mammalian (mouse) tissues contain three defined types of ‘respirasome’, supercomplexes made of CI, CIII2 and CIV. The stoichiometry and position of CIV differs in the three respirasomes, of which only one contains the supercomplex-associated factor SCAF1, whose involvement in respirasome formation has long been contended. Our structures confirm that the ‘canonical’ respirasome (the C-respirasome, CICIII2CIV) does not contain SCAF1, which is instead associated to a different respirasome (the CS-respirasome), containing a second copy of CIV. We also identify an alternative respirasome (A-respirasome), with CIV bound to the ‘back’ of CI, instead of the ‘toe’. This structural characterization of mouse mitochondrial supercomplexes allows us to hypothesize a mechanistic basis for their specific role in different metabolic conditions. acknowledged_ssus: - _id: EM-Fac - _id: LifeSc - _id: PreCl - _id: ScienComp acknowledgement: Supercomplexes of the respiratory chain are established constituents of the oxidative phosphorylation system, but their role in mammalian metabolism has been hotly debated. Although recent studies have shown that different tissues/organs are equipped with specific sets of supercomplexes, depending on their metabolic needs, the notion that supercomplexes have a role in the regulation of metabolism has been challenged. However, irrespective of the mechanistic conclusions, the composition of various high molecular weight supercomplexes remains uncertain. Here, using cryogenic electron microscopy, we demonstrate that mammalian (mouse) tissues contain three defined types of ‘respirasome’, supercomplexes made of CI, CIII2 and CIV. The stoichiometry and position of CIV differs in the three respirasomes, of which only one contains the supercomplex-associated factor SCAF1, whose involvement in respirasome formation has long been contended. Our structures confirm that the ‘canonical’ respirasome (the C-respirasome, CICIII2CIV) does not contain SCAF1, which is instead associated to a different respirasome (the CS-respirasome), containing a second copy of CIV. We also identify an alternative respirasome (A-respirasome), with CIV bound to the ‘back’ of CI, instead of the ‘toe’. This structural characterization of mouse mitochondrial supercomplexes allows us to hypothesize a mechanistic basis for their specific role in different metabolic conditions. article_processing_charge: No article_type: original author: - first_name: Irene full_name: Vercellino, Irene id: 3ED6AF16-F248-11E8-B48F-1D18A9856A87 last_name: Vercellino orcid: 0000-0001-5618-3449 - first_name: Leonid A full_name: Sazanov, Leonid A id: 338D39FE-F248-11E8-B48F-1D18A9856A87 last_name: Sazanov orcid: 0000-0002-0977-7989 citation: ama: Vercellino I, Sazanov LA. SCAF1 drives the compositional diversity of mammalian respirasomes. Nature Structural and Molecular Biology. 2024;31:1061-1071. doi:10.1038/s41594-024-01255-0 apa: Vercellino, I., & Sazanov, L. A. (2024). SCAF1 drives the compositional diversity of mammalian respirasomes. Nature Structural and Molecular Biology. Springer Nature. https://doi.org/10.1038/s41594-024-01255-0 chicago: Vercellino, Irene, and Leonid A Sazanov. “SCAF1 Drives the Compositional Diversity of Mammalian Respirasomes.” Nature Structural and Molecular Biology. Springer Nature, 2024. https://doi.org/10.1038/s41594-024-01255-0. ieee: I. Vercellino and L. A. Sazanov, “SCAF1 drives the compositional diversity of mammalian respirasomes,” Nature Structural and Molecular Biology, vol. 31. Springer Nature, pp. 1061–1071, 2024. ista: Vercellino I, Sazanov LA. 2024. SCAF1 drives the compositional diversity of mammalian respirasomes. Nature Structural and Molecular Biology. 31, 1061–1071. mla: Vercellino, Irene, and Leonid A. Sazanov. “SCAF1 Drives the Compositional Diversity of Mammalian Respirasomes.” Nature Structural and Molecular Biology, vol. 31, Springer Nature, 2024, pp. 1061–71, doi:10.1038/s41594-024-01255-0. short: I. Vercellino, L.A. Sazanov, Nature Structural and Molecular Biology 31 (2024) 1061–1071. corr_author: '1' date_created: 2024-04-14T22:01:03Z date_published: 2024-07-01T00:00:00Z date_updated: 2025-11-24T08:35:04Z day: '01' ddc: - '572' department: - _id: LeSa doi: 10.1038/s41594-024-01255-0 ec_funded: 1 external_id: isi: - '001196897300001' pmid: - '38575788' file: - access_level: open_access checksum: 21f05d188762acd7f49a97f3d09c8d9f content_type: application/pdf creator: lsazanov date_created: 2024-05-14T11:57:56Z date_updated: 2025-01-01T23:30:03Z embargo: 2025-01-01 file_id: '15392' file_name: megacomplex_submit_NSMB_withFigures.pdf file_size: 24424729 relation: main_file file_date_updated: 2025-01-01T23:30:03Z has_accepted_license: '1' intvolume: ' 31' isi: 1 language: - iso: eng month: '07' oa: 1 oa_version: Submitted Version page: 1061-1071 pmid: 1 project: - _id: 627abdeb-2b32-11ec-9570-ec31a97243d3 call_identifier: H2020 grant_number: '101020697' name: Structure and mechanism of respiratory chain molecular machines publication: Nature Structural and Molecular Biology publication_identifier: eissn: - 1545-9985 issn: - 1545-9993 publication_status: published publisher: Springer Nature quality_controlled: '1' related_material: link: - relation: erratum url: https://doi.org/10.1038/s41594-025-01721-3 scopus_import: '1' status: public title: SCAF1 drives the compositional diversity of mammalian respirasomes tmp: image: /images/cc_by.png legal_code_url: https://creativecommons.org/licenses/by/4.0/legalcode name: Creative Commons Attribution 4.0 International Public License (CC-BY 4.0) short: CC BY (4.0) type: journal_article user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 volume: 31 year: '2024' ... --- OA_embargo: '6' OA_place: publisher _id: '18531' abstract: - lang: eng text: "Sex chromosomes and autosomes exhibit very different evolutionary dynamics.\r\nThe Y chromosome usually degenerates, leaving many X-linked loci hemizygous in\r\nmales. Since recessive X-linked mutations are always exposed to selection in males,\r\nselection is more efficient on the X chromosome than on autosomes on recessive\r\nmutations, leading to faster adaptation on the X chromosome than other genomic\r\nregions, if beneficial mutations are on average recessive (known as the Faster-X\r\neffect). In the presence of the functional, but non-recombining gametolog on the Y (as\r\nis often the case in young non-recombining regions), recessive mutations are\r\nsheltered from selection on the X chromosome. We model this scenario and show that\r\nthe efficiency of selection is reduced on diploid X loci due to sheltering by the Y\r\nchromosome. Reduced efficiency of selection leads to slower adaptation and\r\nincreased accumulation of deleterious mutations (Slower-X effect). We extended this\r\nmodel to explore the effect of sex-specific selection on degeneration of sex\r\nchromosomes, showing theoretically that male-limited genes degenerate on the X\r\nchromosome and female-biased genes degenerate on the Y chromosome. This\r\nprediction depends on the effective population size and the mutation rate, explaining\r\nthe variety of sex chromosome degeneration patterns observed in nature.\r\nTo test for direct evidence of a Slower-X (or Slower-Z) effect, we analyzed the\r\nZW sex chromosomes of the flatworm Schistosoma japonicum, which have a very\r\nyoung non-recombining region with non-degenerated W. Diploid Z-linked genes have\r\nhigher ratios of non-synonymous to synonymous polymorphisms than autosomal\r\ngenes, supporting reduced efficiency of selection on the diploid Z region. These results\r\nprovide evidence of sheltering by the W chromosome, a mechanism that could\r\ncontribute to Z (X) chromosome degeneration, and illustrate contrasting evolutionary\r\npatterns in old and young sex chromosome regions. In addition, genes with sexspecific patterns of expression show opposite patterns of selection in the young\r\n(diploid) and old (hemizygous) Z, showing the complex manner in which sex-specific selection shapes the evolutionary patterns of sex chromosomes. " acknowledged_ssus: - _id: ScienComp - _id: CampIT alternative_title: - ISTA Thesis article_processing_charge: No author: - first_name: Andrea full_name: Mrnjavac, Andrea id: 353FAC84-AE61-11E9-8BFC-00D3E5697425 last_name: Mrnjavac citation: ama: Mrnjavac A. Early stages of sex chromosome evolution. 2024. doi:10.15479/at:ista:18531 apa: Mrnjavac, A. (2024). Early stages of sex chromosome evolution. Institute of Science and Technology Austria. https://doi.org/10.15479/at:ista:18531 chicago: Mrnjavac, Andrea. “Early Stages of Sex Chromosome Evolution.” Institute of Science and Technology Austria, 2024. https://doi.org/10.15479/at:ista:18531. ieee: A. Mrnjavac, “Early stages of sex chromosome evolution,” Institute of Science and Technology Austria, 2024. ista: Mrnjavac A. 2024. Early stages of sex chromosome evolution. Institute of Science and Technology Austria. mla: Mrnjavac, Andrea. Early Stages of Sex Chromosome Evolution. Institute of Science and Technology Austria, 2024, doi:10.15479/at:ista:18531. short: A. Mrnjavac, Early Stages of Sex Chromosome Evolution, Institute of Science and Technology Austria, 2024. corr_author: '1' date_created: 2024-11-11T08:40:45Z date_published: 2024-11-11T00:00:00Z date_updated: 2026-04-07T13:22:45Z day: '11' ddc: - '576' degree_awarded: PhD department: - _id: GradSch - _id: BeVi doi: 10.15479/at:ista:18531 file: - access_level: closed checksum: 3e48b163c22114ef5d5371f758668289 content_type: application/vnd.openxmlformats-officedocument.wordprocessingml.document creator: amrnjava date_created: 2024-11-13T12:15:28Z date_updated: 2025-05-11T22:30:04Z embargo_to: open_access file_id: '18551' file_name: AMrnjavac_thesis_library.docx file_size: 26870629 relation: source_file title: Early stages of sex chromosome evolution - access_level: open_access checksum: 3ead60c1b678e7dcf018043aef3b5db2 content_type: application/pdf creator: amrnjava date_created: 2024-11-13T12:15:54Z date_updated: 2025-05-11T22:30:04Z embargo: 2025-05-11 file_id: '18552' file_name: AMrnjavac_thesis_library.pdf file_size: 4228766 relation: main_file title: Early stages of sex chromosome evolution file_date_updated: 2025-05-11T22:30:04Z has_accepted_license: '1' keyword: - Sex chromosomes - evolution - selection - sheltering language: - iso: eng month: '11' oa: 1 oa_version: Published Version page: '181' publication_identifier: issn: - 2663-337X publication_status: published publisher: Institute of Science and Technology Austria related_material: record: - id: '12521' relation: part_of_dissertation status: public - id: '18549' relation: part_of_dissertation status: public status: public supervisor: - first_name: Beatriz full_name: Vicoso, Beatriz id: 49E1C5C6-F248-11E8-B48F-1D18A9856A87 last_name: Vicoso orcid: 0000-0002-4579-8306 title: Early stages of sex chromosome evolution tmp: image: /images/cc_by_nc_nd.png legal_code_url: https://creativecommons.org/licenses/by-nc-nd/4.0/legalcode name: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) short: CC BY-NC-ND (4.0) type: dissertation user_id: ba8df636-2132-11f1-aed0-ed93e2281fdd year: '2024' ... --- OA_place: repository OA_type: free access _id: '18648' abstract: - lang: eng text: "Statistical causal learning in genomics relies on the instrumental variable method of\r\nMendelian Randomization (MR). Currently, an overwhelming number of MR studies\r\npurport to show causal relationships among a wide range of risk factors and outcomes.\r\nHere, we show that selecting instrument variables from genome-wide association study\r\nestimates leads to high false discovery rates for many MR approaches, which can be\r\ngreatly reduced by employing a graphical inference approach which: (i) explicitly tests\r\ninstrumental variable assumptions; (ii) distinguishes direct from indirect factors in very\r\nhigh-dimensional data; (iii) discriminates pleiotropic from trait-specific markers, controlling for LD genome-wide; (iv) accommodates rare variants and binary outcomes in a\r\nprincipled way; and (v) identifies potential unobserved latent confounding. For 17 traits\r\nand 8.4M variants recorded for 458,747 individuals in the UK Biobank, we show that\r\nstandard MR analysis gives an abundance of findings that disappear under stringent\r\nassumption checks, with many relationships reflecting potential unmeasured confounding. This implies that mixtures of temporal precedence and potential for reverse-causality\r\nprohibit understanding the underlying nature of phenotypic and genetic correlations in\r\nbiobank data. We propose that well-curated longitudinal records are likely needed and\r\nthat our approach provides a first-step toward robust principled screening for potential\r\ncausal links.\r\n" acknowledged_ssus: - _id: ScienComp acknowledgement: "We thank Zoltan Kutalik and members of the Robinson group \r\nat ISTA for their comments, which improved this manuscript. This work was funded \r\nby a research collaboration agreement between Boehringer Ingelheim and the research \r\ngroup of MRR at the Institute of Science and Technology Austria. Additional funding \r\nwas also provided by an SNSF Eccellenza Grant to MRR (PCEGP3-181181), and by \r\ncore funding from the Institute of Science and Technology Austria. We would like \r\nto acknowledge the participants and investigators of the UK Biobank study. High- \r\nperformance computing was supported by the Scientific Service Units (SSU) of IST \r\nAustria through resources provided by Scientific Computing (SciComp). " article_processing_charge: No author: - first_name: Nick N full_name: Machnik, Nick N id: 3591A0AA-F248-11E8-B48F-1D18A9856A87 last_name: Machnik orcid: 0000-0001-6617-9742 - first_name: Seyed Mahdi full_name: Mahmoudi, Seyed Mahdi id: b9f6d5ef-7774-11eb-a47f-df2c75c02ee7 last_name: Mahmoudi - first_name: Malgorzata full_name: Borczyk, Malgorzata last_name: Borczyk - first_name: Ilse full_name: Krätschmer, Ilse id: 30d4014e-7753-11eb-b44b-db6d61112e73 last_name: Krätschmer orcid: 0000-0002-5636-9259 - first_name: Markus J. full_name: Bauer, Markus J. last_name: Bauer - first_name: Matthew Richard full_name: Robinson, Matthew Richard id: E5D42276-F5DA-11E9-8E24-6303E6697425 last_name: Robinson orcid: 0000-0001-8982-8813 citation: ama: Machnik NN, Mahmoudi SM, Borczyk M, Krätschmer I, Bauer MJ, Robinson MR. Causal inference for multiple risk factors and diseases from genomics data. bioRxiv. 2024. doi:10.1101/2023.12.06.570392 apa: Machnik, N. N., Mahmoudi, S. M., Borczyk, M., Krätschmer, I., Bauer, M. J., & Robinson, M. R. (2024). Causal inference for multiple risk factors and diseases from genomics data. bioRxiv. https://doi.org/10.1101/2023.12.06.570392 chicago: Machnik, Nick N, Seyed Mahdi Mahmoudi, Malgorzata Borczyk, Ilse Krätschmer, Markus J. Bauer, and Matthew Richard Robinson. “Causal Inference for Multiple Risk Factors and Diseases from Genomics Data.” BioRxiv, 2024. https://doi.org/10.1101/2023.12.06.570392. ieee: N. N. Machnik, S. M. Mahmoudi, M. Borczyk, I. Krätschmer, M. J. Bauer, and M. R. Robinson, “Causal inference for multiple risk factors and diseases from genomics data,” bioRxiv. 2024. ista: Machnik NN, Mahmoudi SM, Borczyk M, Krätschmer I, Bauer MJ, Robinson MR. 2024. Causal inference for multiple risk factors and diseases from genomics data. bioRxiv, 10.1101/2023.12.06.570392. mla: Machnik, Nick N., et al. “Causal Inference for Multiple Risk Factors and Diseases from Genomics Data.” BioRxiv, 2024, doi:10.1101/2023.12.06.570392. short: N.N. Machnik, S.M. Mahmoudi, M. Borczyk, I. Krätschmer, M.J. Bauer, M.R. Robinson, BioRxiv (2024). corr_author: '1' date_created: 2024-12-11T10:42:59Z date_published: 2024-08-10T00:00:00Z date_updated: 2026-07-26T22:30:20Z day: '10' department: - _id: MaRo doi: 10.1101/2023.12.06.570392 language: - iso: eng license: https://creativecommons.org/licenses/by-nc/4.0/ main_file_link: - open_access: '1' url: https://doi.org/10.1101/2023.12.06.570392 month: '08' oa: 1 oa_version: Preprint project: - _id: 9B8D11D6-BA93-11EA-9121-9846C619BF3A grant_number: PCEGP3_181181 name: Improving estimation and prediction of common complex disease risk - _id: bd936e6f-d553-11ed-ba76-a82299f63e8c grant_number: '590359' name: Advanced statistical modelling to facilitate more accurate characterisation of disease phenotypes, improved genetic mapping, and effective therapeutic hypothesis generation publication: bioRxiv publication_status: published related_material: record: - id: '18642' relation: dissertation_contains status: public status: public title: Causal inference for multiple risk factors and diseases from genomics data tmp: image: /images/cc_by_nc.png legal_code_url: https://creativecommons.org/licenses/by-nc/4.0/legalcode name: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) short: CC BY-NC (4.0) type: preprint user_id: 8b945eb4-e2f2-11eb-945a-df72226e66a9 year: '2024' ... --- OA_place: publisher _id: '17206' abstract: - lang: eng text: "Males and females exhibit numerous differences, from the initial stages of sex determination to the\r\ndevelopment of secondary sexual characteristics. In Drosophila, these differences have been\r\nthoroughly studied. Extensive research has been performed to understand the role and molecular\r\nmode of action of central sex in determining switch genes, such as transformer (tra) and Sex-lethal\r\n(Sxl). Furthermore, studies have highlighted differential gene expression as an essential mechanism to\r\ncreate sexual dimorphism. An alternative path to sexual dimorphism that has been less explored is\r\nalternative splicing, the mechanism through which genes can produce multiple transcripts with\r\ndistinct properties and functions. The primary switch sex-determining gene Sxl is a good example of\r\nthe role of alternative splicing for sex-specific functions: the inclusion of a specific exon determines\r\nthe male or female form of the protein, which in turn switches on either the male or female\r\ndevelopmental pathway. The genes that act upstream of Sxl and determine which form is expressed -\r\nthe counter genes - have received less attention. This thesis addresses two critical questions about\r\nthe molecular encoding of sexes in the Drosophila melanogaster genome: First, the use of splice forms\r\nin male and female tissues in D. melanogaster is examined, inferring the molecular and evolutionary\r\nparameters shaping the diversity of the splicing landscape. Second, the behaviour of counter genes in\r\nDrosophila-related species is investigated, shedding light on potential changes leading to their\r\nincorporation into the sex-determination pathway.\r\nFor the alternative splicing analyses, long-read RNA sequencing of testes, ovaries, female and male\r\nmidguts, heads, and whole bodies was performed. A novel pipeline was developed to assign unique\r\ntranscript identifiers for each sequence of exons and introns in the read, enabling detailed\r\ncomparisons of splicing variants in each tissue/sex. Alternative splicing was found to be more\r\npervasive in females than males (22,201 exclusive splice forms in females versus 12,631 in males),\r\nespecially when comparing ovaries to other tissues. The ovaries alone displayed 15,299 exclusive\r\nsplice forms, suggesting most female exclusive splice forms originate there. Genome location and gene\r\nage were also correlated with the number of splice forms per gene. In particular, the X and 4th\r\nchromosomes (Muller elements A and F) showed more splice forms per gene than other\r\nchromosomes. Additionally, genes older than 63 million years exhibited more splice forms per gene\r\nthan younger genes. Our results suggest that alternative splicing is more prevalent than previously\r\nbelieved, with numerous female-exclusive forms, age, and location playing significant roles in shaping\r\nits prevalence.\r\nFor the counter genes analyses, we combined published gene expression, genomic, and gene\r\ninteraction data from various clades (Bactrocera jarvisi, B. oleae, Ceratitis capitata, Mus musculus,\r\nCaenorhabditis elegans, Homo sapiens, and D. melanogaster). The counter genes scute (sc), extra\r\nmacrochaetae (emc), groucho (gro), deadpan (dpn), daughterless (da), runt (run), Sxl, hermaphrodite\r\n(her), and tra maintain conserved Muller element locations between C. capitata and D. melanogaster,\r\nwhich are most of the counter genes identified in the C. capitata genome. Their expression patterns\r\nduring early embryogenesis in B. jarvisi and D. melanogaster are also similar for counter genes dpn,\r\ngro, da, and emc. However, Sxl and sc are also found to have more extreme expression ratios between\r\nthe species. Lastly, gene interactions within the counter genes are conserved, with da-sc and gro-dpn\r\ninteractions occurring in Drosophila, worms, humans, and mice. Interactions such as dpn-sc, dpn-da,\r\nda-emc, and gro-run are present in Drosophila, mice, and humans, suggesting these genes were\r\nrecruited by ancestral characteristics, primarily during embryogenesis. The conserved expression,\r\nlocation, and interactions of counter genes suggest serendipitous recruitment of such genes instead\r\nof a change in those characteristics as they were recruited for this function. " acknowledged_ssus: - _id: ScienComp alternative_title: - ISTA Thesis article_processing_charge: No author: - first_name: Julia full_name: Raices, Julia id: 3EE67F22-F248-11E8-B48F-1D18A9856A87 last_name: Raices citation: ama: 'Raices J. Novel approaches to studying alternative splicing in Drosophila Melanogaster : Insights into sex-specific gene expression and the evolution of sex determination. 2024. doi:10.15479/at:ista:17206' apa: 'Raices, J. (2024). Novel approaches to studying alternative splicing in Drosophila Melanogaster : Insights into sex-specific gene expression and the evolution of sex determination. Institute of Science and Technology Austria. https://doi.org/10.15479/at:ista:17206' chicago: 'Raices, Julia. “Novel Approaches to Studying Alternative Splicing in Drosophila Melanogaster : Insights into Sex-Specific Gene Expression and the Evolution of Sex Determination.” Institute of Science and Technology Austria, 2024. https://doi.org/10.15479/at:ista:17206.' ieee: 'J. Raices, “Novel approaches to studying alternative splicing in Drosophila Melanogaster : Insights into sex-specific gene expression and the evolution of sex determination,” Institute of Science and Technology Austria, 2024.' ista: 'Raices J. 2024. Novel approaches to studying alternative splicing in Drosophila Melanogaster : Insights into sex-specific gene expression and the evolution of sex determination. Institute of Science and Technology Austria.' mla: 'Raices, Julia. Novel Approaches to Studying Alternative Splicing in Drosophila Melanogaster : Insights into Sex-Specific Gene Expression and the Evolution of Sex Determination. Institute of Science and Technology Austria, 2024, doi:10.15479/at:ista:17206.' short: 'J. Raices, Novel Approaches to Studying Alternative Splicing in Drosophila Melanogaster : Insights into Sex-Specific Gene Expression and the Evolution of Sex Determination, Institute of Science and Technology Austria, 2024.' corr_author: '1' date_created: 2024-07-05T14:15:29Z date_published: 2024-07-05T00:00:00Z date_updated: 2026-04-07T13:03:22Z day: '05' ddc: - '570' degree_awarded: PhD department: - _id: BeVi - _id: GradSch doi: 10.15479/at:ista:17206 ec_funded: 1 file: - access_level: closed checksum: d5e9234bde8667b005a8cfe18bb467d3 content_type: application/vnd.openxmlformats-officedocument.wordprocessingml.document creator: cchlebak date_created: 2024-07-11T07:18:01Z date_updated: 2025-01-11T23:30:04Z embargo_to: open_access file_id: '17223' file_name: ThesisRaices2024_postDefense.docx file_size: 13788479 relation: source_file - access_level: open_access checksum: f5ed0139aa3e11ce58369f0915647c5c content_type: application/pdf creator: cchlebak date_created: 2024-07-11T07:22:32Z date_updated: 2025-01-11T23:30:04Z embargo: 2025-01-11 file_id: '17224' file_name: ThesisRaices2024_nosignature.pdf file_size: 5580296 relation: main_file file_date_updated: 2025-01-11T23:30:04Z has_accepted_license: '1' language: - iso: eng license: https://creativecommons.org/licenses/by-sa/4.0/ month: '07' oa: 1 oa_version: Published Version page: '82' project: - _id: 250BDE62-B435-11E9-9278-68D0E5697425 call_identifier: H2020 grant_number: '715257' name: Prevalence and Influence of Sexual Antagonism on Genome Evolution publication_identifier: issn: - 2663-337X publication_status: published publisher: Institute of Science and Technology Austria status: public supervisor: - first_name: Beatriz full_name: Vicoso, Beatriz id: 49E1C5C6-F248-11E8-B48F-1D18A9856A87 last_name: Vicoso orcid: 0000-0002-4579-8306 title: 'Novel approaches to studying alternative splicing in Drosophila Melanogaster : Insights into sex-specific gene expression and the evolution of sex determination' tmp: image: /images/cc_by_sa.png legal_code_url: https://creativecommons.org/licenses/by-sa/4.0/legalcode name: Creative Commons Attribution-ShareAlike 4.0 International Public License (CC BY-SA 4.0) short: CC BY-SA (4.0) type: dissertation user_id: ba8df636-2132-11f1-aed0-ed93e2281fdd year: '2024' ... --- OA_place: publisher _id: '18101' abstract: - lang: eng text: "The Retroviridae family consists of two sub-families, the Orthoretrovirinae and the\r\nSpumaretrovirinae. The Orthoretroviruses contain important human pathogens, such as the\r\nhuman immunodeficiency virus 1 (HIV-1). They also harbor other retrovirus species which\r\nare regularly used as model systems to study the retroviral life cycle. The main structural\r\ncomponent of the retroviruses, is the Gag protein and its truncation derivatives occurring\r\nduring viral maturation. Orthoretroviral Gag assemblies have been extensively studied to\r\nunderstand the interactions that confer stability and morphology to viral particles.\r\nThe Spumaretrovirinae subfamily represent an early diverging branch of the Retroviridae.\r\nIts members, the Foamy viruses (FV), share most of the conventional features found in\r\nretroviruses. However, they also possess multiple characteristics that make them unique. In\r\nparticular, FV Gag does not get extensively cleaved as in orthoretroviruses. Hence, the Gag\r\narchitecture deviates from the canonical domain arrangement in FV. They also exhibit a\r\npeculiar particle morphology, having no apparent immature state and a seemingly\r\nicosahedral mature particle. Due to this, many fundamental questions on FV structural\r\nassembly mechanisms remain open. To answer these questions, was the main focus of this\r\nthesis.\r\nMainly, it is not known how FV assemble their core in a virus particle and what are the\r\nimportant assembly interaction sites within said core. What is the minimum assembly\r\ncompetent domain of FV Gag? Is there a morphological change in the assembly type of FVGag lattices? If so, what is defining these morphological shifts? Finally, it would be\r\ninteresting to know what is the evolutionary relationship between FV and the rest of the\r\nretrotranscribing elements, from a structural point of view?\r\nTo answer these questions, membrane-enveloped mammalian cell-derived FV virus-like\r\nparticles (VLPs) were produced. Cryo-electron tomography (cryo-ET) analysis suggested\r\nthese FV VLPs do not form a canonical retroviral Gag lattice structure, which is in line with\r\nearlier observations. To further evaluate FV Gag assembly competence and morphology,\r\nthe first bacterial cell-derived in vitro VLP assembly system was designed and optimized.\r\nUsing this system with different truncation variants, the minimum assembly competent\r\ndomain of FV Gag was found to be the putative CA300-477 domain. Varying VLP\r\nmorphologies were also observed and strongly suggested residues upstream of CA300-477\r\nplay a role in morphology determination. Finally, a combined cryo-electron microscopy (cryoEM) and cryo-ET approach was taken to analyze tubular assemblies from the minimal\r\nassembly competent domain. This revealed an unexpectedly unique non-canonical\r\nassembly architecture. Three novel lattice stabilizing interfaces were described which\r\nproved to be as unique as the lattice arrangement. Comparison to a newly published FV CA\r\ncore structure revealed the CA-CA interactions in the atypical assembly do not recapitulate\r\nwhat is described for the FV core lattice. However, the new in vitro VLP assembly system\r\nobtained in this thesis also provides an exciting opportunity to study still unresolved FV\r\nassembly features in a potentially facilitated approach compared to conventional methods.\r\nIn summary, this work provided a deeper understanding of the basic FV Gag assembly unit,\r\nas well as presenting the first FV Gag-derived in vitro VLP assembly system. This system\r\nreveals a novel and unique assembly architecture among retroviral in vitro assemblies." acknowledged_ssus: - _id: EM-Fac - _id: LifeSc - _id: ScienComp alternative_title: - ISTA Thesis article_processing_charge: No author: - first_name: Dario J full_name: Porley, Dario J id: 2FD6EA6C-F248-11E8-B48F-1D18A9856A87 last_name: Porley citation: ama: Porley Esteves D. Structural characterization of spumavirus capsid assemblies. 2024. doi:10.15479/at:ista:18101 apa: Porley Esteves, D. (2024). Structural characterization of spumavirus capsid assemblies. Institute of Science and Technology Austria. https://doi.org/10.15479/at:ista:18101 chicago: Porley Esteves, Darío. “Structural Characterization of Spumavirus Capsid Assemblies.” Institute of Science and Technology Austria, 2024. https://doi.org/10.15479/at:ista:18101. ieee: D. Porley Esteves, “Structural characterization of spumavirus capsid assemblies,” Institute of Science and Technology Austria, 2024. ista: Porley Esteves D. 2024. Structural characterization of spumavirus capsid assemblies. Institute of Science and Technology Austria. mla: Porley Esteves, Darío. Structural Characterization of Spumavirus Capsid Assemblies. Institute of Science and Technology Austria, 2024, doi:10.15479/at:ista:18101. short: D. Porley Esteves, Structural Characterization of Spumavirus Capsid Assemblies, Institute of Science and Technology Austria, 2024. corr_author: '1' date_created: 2024-09-20T10:21:03Z date_published: 2024-09-26T00:00:00Z date_updated: 2026-04-07T13:21:01Z day: '26' ddc: - '570' degree_awarded: PhD department: - _id: GradSch - _id: FlSc doi: 10.15479/at:ista:18101 ec_funded: 1 file: - access_level: closed checksum: 3b8b0bacfe61112f3852744f3170e468 content_type: application/vnd.openxmlformats-officedocument.wordprocessingml.document creator: dporley date_created: 2024-09-26T13:40:33Z date_updated: 2025-03-25T23:30:03Z embargo_to: open_access file_id: '18149' file_name: PhD_thesis_DPorley_final_20240919.docx file_size: 14213128 relation: source_file - access_level: open_access checksum: 6c3a652a8eede874118e11d66a63652f content_type: application/pdf creator: dporley date_created: 2024-09-26T13:41:39Z date_updated: 2025-03-25T23:30:03Z embargo: 2025-03-25 file_id: '18150' file_name: PhD_thesis_DPorley_final_20240926_pdfa1.pdf file_size: 18583031 relation: main_file file_date_updated: 2025-03-25T23:30:03Z has_accepted_license: '1' language: - iso: eng month: '09' oa: 1 oa_version: Published Version page: '131' project: - _id: 2564DBCA-B435-11E9-9278-68D0E5697425 call_identifier: H2020 grant_number: '665385' name: International IST Doctoral Program - _id: 9B9C98E0-BA93-11EA-9121-9846C619BF3A grant_number: '25762' name: Structural characterization of spumavirus capsid assemblies to understand conserved Ortervirales assembly mechanisms publication_identifier: isbn: - 978-3-99078-041-1 issn: - 2663-337X publication_status: published publisher: Institute of Science and Technology Austria status: public supervisor: - first_name: Florian KM full_name: Schur, Florian KM id: 48AD8942-F248-11E8-B48F-1D18A9856A87 last_name: Schur orcid: 0000-0003-4790-8078 title: Structural characterization of spumavirus capsid assemblies type: dissertation user_id: ba8df636-2132-11f1-aed0-ed93e2281fdd year: '2024' ... --- OA_place: repository _id: '17465' abstract: - lang: eng text: "In the modern age of machine learning, artificial neural networks have become an integral part\r\nof many practical systems. One of the key ingredients of the success of the deep learning\r\napproach is recent computational advances which allowed the training of models with billions\r\nof parameters on large-scale data. Such over-parameterized and data-hungry regimes pose a\r\nchallenge for the theoretical analysis of modern models since “classical” statistical wisdom\r\nis no longer applicable. In this view, it is paramount to extend or develop new machinery\r\nthat will allow tackling the neural network analysis under new challenging asymptotic regimes,\r\nwhich is the focus of this thesis.\r\nLarge neural network systems are usually optimized via “local” search algorithms, such\r\nas stochastic gradient descent (SGD). However, given the high-dimensional nature of the\r\nparameter space, it is a priori not clear why such a crude “local” approach works so remarkably\r\nwell in practice. We take a step towards demystifying this phenomenon by showing that\r\nthe landscape of the SGD training dynamics exhibits a few beneficial properties for the\r\noptimization. First, we show that along the SGD trajectory an over-parameterized network\r\nis dropout stable. The emergence of dropout stability allows to conclude that the minima\r\nfound by SGD are connected via a continuous path of small loss. This in turn means that\r\nthe high-dimensional landscape of the neural network optimization problem is provably not so\r\nunfavourable to gradient-based training, due to mode connectivity. Next, we show that SGD\r\nfor an over-parameterized network tends to find solutions that are functionally more “simple”.\r\nThis in turn means that the SGD minima are more robust, since a less complicated solution\r\nwill less likely overfit the data. More formally, for a prototypical example of a wide two-layer\r\nReLU network on a 1d regression task we show that the SGD algorithm is implicitly selective in\r\nits choice of an interpolating solution. Namely, at convergence the neural network implements\r\na piece-wise linear function with the number of linear regions depending only on the amount\r\nof training data. This is in contrast to a “smooth”-like behaviour which one would expect\r\ngiven such a severe over-parameterization of the model.\r\nDiverging from the generic supervised setting of classification and regression problems, we\r\nanalyze an auto-encoder model that is commonly used for representation learning and data\r\ncompression. Despite the wide applicability of the auto-encoding paradigm, the theoretical\r\nunderstanding of their behaviour is limited even in the simplistic shallow case. The related\r\nwork is restricted to extreme asymptotic regimes in which the auto-encoder is either severely\r\nover-parameterized or under-parameterized. In contrast, we provide a tight characterization\r\nfor the 1-bit compression of Gaussian signals in the challenging proportional regime, i.e., the\r\ninput dimension and the size of the compressed representation obey the same asymptotics.\r\nWe also show that gradient-based methods are able to find a globally optimal solution and\r\nthat the predictions made for Gaussian data extrapolate beyond - to the case of compression\r\nof natural images. Next, we relax the Gaussian assumption and study more structured input\r\nsources. We show that the shallow model is sometimes agnostic to the structure of the data\r\nvii\r\nwhich results in a Gaussian-like behaviour. We prove that making the decoding component\r\nslightly less shallow is already enough to escape the “curse” of Gaussian performance.\r\n" acknowledged_ssus: - _id: ScienComp alternative_title: - ISTA Thesis article_processing_charge: No author: - first_name: Aleksandr full_name: Shevchenko, Aleksandr id: F2B06EC2-C99E-11E9-89F0-752EE6697425 last_name: Shevchenko citation: ama: Shevchenko A. High-dimensional limits in artificial neural networks. 2024. doi:10.15479/at:ista:17465 apa: Shevchenko, A. (2024). High-dimensional limits in artificial neural networks. Institute of Science and Technology Austria. https://doi.org/10.15479/at:ista:17465 chicago: Shevchenko, Alexander. “High-Dimensional Limits in Artificial Neural Networks.” Institute of Science and Technology Austria, 2024. https://doi.org/10.15479/at:ista:17465. ieee: A. Shevchenko, “High-dimensional limits in artificial neural networks,” Institute of Science and Technology Austria, 2024. ista: Shevchenko A. 2024. High-dimensional limits in artificial neural networks. Institute of Science and Technology Austria. mla: Shevchenko, Alexander. High-Dimensional Limits in Artificial Neural Networks. Institute of Science and Technology Austria, 2024, doi:10.15479/at:ista:17465. short: A. Shevchenko, High-Dimensional Limits in Artificial Neural Networks, Institute of Science and Technology Austria, 2024. corr_author: '1' date_created: 2024-08-28T15:14:25Z date_published: 2024-08-29T00:00:00Z date_updated: 2026-06-18T17:55:53Z day: '29' ddc: - '519' degree_awarded: PhD department: - _id: GradSch - _id: DaAl - _id: MaMo doi: 10.15479/at:ista:17465 file: - access_level: open_access checksum: da6dd3166078934577f6af93d27000e2 content_type: application/pdf creator: ashevche date_created: 2024-09-02T09:23:32Z date_updated: 2024-10-05T22:30:05Z embargo: 2024-10-04 file_id: '17482' file_name: thesis_a2b.pdf file_size: 4468610 relation: main_file - access_level: closed checksum: 76a39ef252239560923cdda4ce0a31a4 content_type: application/zip creator: ashevche date_created: 2024-09-02T09:23:46Z date_updated: 2024-10-05T22:30:05Z embargo_to: open_access file_id: '17483' file_name: Thesis Alex - ISTA.zip file_size: 15930999 relation: source_file file_date_updated: 2024-10-05T22:30:05Z has_accepted_license: '1' language: - iso: eng month: '08' oa: 1 oa_version: Published Version page: '232' project: - _id: 059876FA-7A3F-11EA-A408-12923DDC885E name: Prix Lopez-Loretta 2019 - Marco Mondelli - _id: 9B9290DE-BA93-11EA-9121-9846C619BF3A grant_number: W1260-N35 name: Vienna Graduate School on Computational Optimization publication_identifier: issn: - 2663-337X publication_status: published publisher: Institute of Science and Technology Austria related_material: record: - id: '11420' relation: part_of_dissertation status: public - id: '14459' relation: part_of_dissertation status: public - id: '9198' relation: part_of_dissertation status: public - id: '17469' relation: part_of_dissertation status: public status: public supervisor: - first_name: Marco full_name: Mondelli, Marco id: 27EB676C-8706-11E9-9510-7717E6697425 last_name: Mondelli orcid: 0000-0002-3242-7020 - first_name: Dan-Adrian full_name: Alistarh, Dan-Adrian id: 4A899BFC-F248-11E8-B48F-1D18A9856A87 last_name: Alistarh orcid: 0000-0003-3650-940X title: High-dimensional limits in artificial neural networks type: dissertation user_id: 8b945eb4-e2f2-11eb-945a-df72226e66a9 year: '2024' ... --- APC_amount: 3145,39 EUR DOAJ_listed: '1' OA_place: publisher OA_type: gold _id: '17890' abstract: - lang: eng text: Our understanding of the molecular pathways that regulate oogenesis and define cellular identity in the Arthropod female reproductive system and the extent of their conservation is currently very limited. This is due to the focus on model systems, including Drosophila and Daphnia, which do not reflect the observed diversity of morphologies, reproductive modes, and sex chromosome systems. We use single-nucleus RNA and ATAC sequencing to produce a comprehensive single nucleus atlas of the adult Artemia franciscana female reproductive system. We map our data to the Fly Cell Atlas single-nucleus dataset of the Drosophila melanogaster ovary, shedding light on the conserved regulatory programs between the two distantly related Arthropod species. We identify the major cell types known to be present in the Artemia ovary, including germ cells, follicle cells, and ovarian muscle cells. Additionally, we use the germ cells to explore gene regulation and expression of the Z chromosome during meiosis, highlighting its unique regulatory dynamics and allowing us to explore the presence of meiotic sex chromosome silencing in this group. acknowledged_ssus: - _id: ScienComp acknowledgement: "We thank the Vicoso group for their valuable comments on the earlier draft of the manuscript. We would also like to thank the Vienna BioCenter Next Generation Sequencing (NGS) facility staff, and in particular, Thomas Grentzinger for his support with the handling and sequencing of the samples, the scientific computing unit at ISTA for the computational resources, Brittney Wick for the help with hosting our data on the UCSC Cell Browser, and Lora B. Sweeney for her valuable input at the different stages of the project.\r\nThis research was funded by the Austrian science fund (FWF), as part of the SFB Meiosis consortium https://sfbmeiosis.org/, grant ID FWF SFB F88-10) to BV. " article_number: e1011376 article_processing_charge: Yes article_type: original author: - first_name: Marwan N full_name: Elkrewi, Marwan N id: 0B46FACA-A8E1-11E9-9BD3-79D1E5697425 last_name: Elkrewi orcid: 0000-0002-5328-7231 - first_name: Beatriz full_name: Vicoso, Beatriz id: 49E1C5C6-F248-11E8-B48F-1D18A9856A87 last_name: Vicoso orcid: 0000-0002-4579-8306 citation: ama: Elkrewi MN, Vicoso B. Single-nucleus atlas of the Artemia female reproductive system suggests germline repression of the Z chromosome. PLoS Genetics. 2024;20(8). doi:10.1371/journal.pgen.1011376 apa: Elkrewi, M. N., & Vicoso, B. (2024). Single-nucleus atlas of the Artemia female reproductive system suggests germline repression of the Z chromosome. PLoS Genetics. Public Library of Science. https://doi.org/10.1371/journal.pgen.1011376 chicago: Elkrewi, Marwan N, and Beatriz Vicoso. “Single-Nucleus Atlas of the Artemia Female Reproductive System Suggests Germline Repression of the Z Chromosome.” PLoS Genetics. Public Library of Science, 2024. https://doi.org/10.1371/journal.pgen.1011376. ieee: M. N. Elkrewi and B. Vicoso, “Single-nucleus atlas of the Artemia female reproductive system suggests germline repression of the Z chromosome,” PLoS Genetics, vol. 20, no. 8. Public Library of Science, 2024. ista: Elkrewi MN, Vicoso B. 2024. Single-nucleus atlas of the Artemia female reproductive system suggests germline repression of the Z chromosome. PLoS Genetics. 20(8), e1011376. mla: Elkrewi, Marwan N., and Beatriz Vicoso. “Single-Nucleus Atlas of the Artemia Female Reproductive System Suggests Germline Repression of the Z Chromosome.” PLoS Genetics, vol. 20, no. 8, e1011376, Public Library of Science, 2024, doi:10.1371/journal.pgen.1011376. short: M.N. Elkrewi, B. Vicoso, PLoS Genetics 20 (2024). corr_author: '1' date_created: 2024-09-08T22:01:11Z date_published: 2024-08-30T00:00:00Z date_updated: 2026-07-26T22:31:06Z day: '30' ddc: - '570' department: - _id: BeVi doi: 10.1371/journal.pgen.1011376 external_id: isi: - '001304090200001' pmid: - '39213449' file: - access_level: open_access checksum: f5d96b9af57126fc1063e951440477d6 content_type: application/pdf creator: dernst date_created: 2024-09-11T07:54:12Z date_updated: 2024-09-11T07:54:12Z file_id: '18056' file_name: 2024_PloSGenetics_Elkrewi.pdf file_size: 8962687 relation: main_file success: 1 file_date_updated: 2024-09-11T07:54:12Z has_accepted_license: '1' intvolume: ' 20' isi: 1 issue: '8' language: - iso: eng month: '08' oa: 1 oa_version: Published Version pmid: 1 project: - _id: 3AC91DDA-15DF-11EA-824D-93A3E7B544D1 call_identifier: FWF name: FWF Open Access Fund - _id: 34ae1506-11ca-11ed-8bc3-c14f4c474396 grant_number: F8810 name: The highjacking of meiosis for asexual reproduction publication: PLoS Genetics publication_identifier: eissn: - 1553-7404 issn: - 1553-7390 publication_status: published publisher: Public Library of Science quality_controlled: '1' related_material: link: - relation: software url: https://github.com/Melkrewi/Artemia-snRNAseq-Project record: - id: '17362' relation: research_data status: public - id: '19386' relation: dissertation_contains status: public scopus_import: '1' status: public title: Single-nucleus atlas of the Artemia female reproductive system suggests germline repression of the Z chromosome tmp: image: /images/cc_by.png legal_code_url: https://creativecommons.org/licenses/by/4.0/legalcode name: Creative Commons Attribution 4.0 International Public License (CC-BY 4.0) short: CC BY (4.0) type: journal_article user_id: 317138e5-6ab7-11ef-aa6d-ffef3953e345 volume: 20 year: '2024' ... --- _id: '18121' abstract: - lang: eng text: It is known that sparsity can improve interpretability for deep neural networks. However, existing methods in the area either require networks that are pre-trained with sparsity constraints, or impose sparsity after the fact, altering the network’s general behavior. In this paper, we demonstrate, for the first time, that sparsity can instead be incorporated into the interpretation process itself, as a sample-specific preprocessing step. Unlike previous work, this approach, which we call SPADE, does not place constraints on the trained model and does not affect its behavior during inference on the sample. Given a trained model and a target sample, SPADE uses sample-targeted pruning to provide a "trace" of the network’s execution on the sample, reducing the network to the most important connections prior to computing an interpretation. We demonstrate that preprocessing with SPADE significantly increases the accuracy of image saliency maps across several interpretability methods. Additionally, SPADE improves the usefulness of neuron visualizations, aiding humans in reasoning about network behavior. Our code is available at https://github.com/IST-DASLab/SPADE. acknowledged_ssus: - _id: ScienComp acknowledgement: The authors would like to thank Stephen Casper and Tony Wang for their feedback on this work, and Eldar Kurtic for his advice on aspects of the project. This research was supported by the Scientific Service Units (SSU) of IST Austria through resources provided by Scientific Computing (SciComp). EI was supported in part by the FWF DK VGSCO, grant agreement number W1260-N35. alternative_title: - PMLR article_processing_charge: No arxiv: 1 author: - first_name: Arshia Soltani full_name: Moakhar, Arshia Soltani last_name: Moakhar - first_name: Eugenia B full_name: Iofinova, Eugenia B id: f9a17499-f6e0-11ea-865d-fdf9a3f77117 last_name: Iofinova orcid: 0000-0002-7778-3221 - first_name: Elias full_name: Frantar, Elias id: 09a8f98d-ec99-11ea-ae11-c063a7b7fe5f last_name: Frantar - first_name: Dan-Adrian full_name: Alistarh, Dan-Adrian id: 4A899BFC-F248-11E8-B48F-1D18A9856A87 last_name: Alistarh orcid: 0000-0003-3650-940X citation: ama: 'Moakhar AS, Iofinova EB, Frantar E, Alistarh D-A. SPADE: Sparsity-guided debugging for deep neural networks. In: Proceedings of the 41st International Conference on Machine Learning. Vol 235. ML Research Press; 2024:45955-45987.' apa: 'Moakhar, A. S., Iofinova, E. B., Frantar, E., & Alistarh, D.-A. (2024). SPADE: Sparsity-guided debugging for deep neural networks. In Proceedings of the 41st International Conference on Machine Learning (Vol. 235, pp. 45955–45987). Vienna, Austria: ML Research Press.' chicago: 'Moakhar, Arshia Soltani, Eugenia B Iofinova, Elias Frantar, and Dan-Adrian Alistarh. “SPADE: Sparsity-Guided Debugging for Deep Neural Networks.” In Proceedings of the 41st International Conference on Machine Learning, 235:45955–87. ML Research Press, 2024.' ieee: 'A. S. Moakhar, E. B. Iofinova, E. Frantar, and D.-A. Alistarh, “SPADE: Sparsity-guided debugging for deep neural networks,” in Proceedings of the 41st International Conference on Machine Learning, Vienna, Austria, 2024, vol. 235, pp. 45955–45987.' ista: 'Moakhar AS, Iofinova EB, Frantar E, Alistarh D-A. 2024. SPADE: Sparsity-guided debugging for deep neural networks. Proceedings of the 41st International Conference on Machine Learning. ICML: International Conference on Machine Learning, PMLR, vol. 235, 45955–45987.' mla: 'Moakhar, Arshia Soltani, et al. “SPADE: Sparsity-Guided Debugging for Deep Neural Networks.” Proceedings of the 41st International Conference on Machine Learning, vol. 235, ML Research Press, 2024, pp. 45955–87.' short: A.S. Moakhar, E.B. Iofinova, E. Frantar, D.-A. Alistarh, in:, Proceedings of the 41st International Conference on Machine Learning, ML Research Press, 2024, pp. 45955–45987. conference: end_date: 2024-07-27 location: Vienna, Austria name: 'ICML: International Conference on Machine Learning' start_date: 2024-07-21 corr_author: '1' date_created: 2024-09-22T22:01:46Z date_published: 2024-09-01T00:00:00Z date_updated: 2026-07-27T12:50:03Z day: '01' department: - _id: DaAl external_id: arxiv: - '2310.04519' intvolume: ' 235' language: - iso: eng main_file_link: - open_access: '1' url: https://doi.org/10.48550/arXiv.2310.04519 month: '09' oa: 1 oa_version: Preprint page: 45955-45987 project: - _id: 9B9290DE-BA93-11EA-9121-9846C619BF3A grant_number: W1260-N35 name: Vienna Graduate School on Computational Optimization publication: Proceedings of the 41st International Conference on Machine Learning publication_identifier: eissn: - 2640-3498 publication_status: published publisher: ML Research Press quality_controlled: '1' related_material: link: - relation: software url: https://github.com/IST-DASLab/SPADE record: - id: '21854' relation: dissertation_contains status: public scopus_import: '1' status: public title: 'SPADE: Sparsity-guided debugging for deep neural networks' type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 volume: 235 year: '2024' ... --- _id: '12334' abstract: - lang: eng text: Regulation of the Arp2/3 complex is required for productive nucleation of branched actin networks. An emerging aspect of regulation is the incorporation of subunit isoforms into the Arp2/3 complex. Specifically, both ArpC5 subunit isoforms, ArpC5 and ArpC5L, have been reported to fine-tune nucleation activity and branch junction stability. We have combined reverse genetics and cellular structural biology to describe how ArpC5 and ArpC5L differentially affect cell migration. Both define the structural stability of ArpC1 in branch junctions and, in turn, by determining protrusion characteristics, affect protein dynamics and actin network ultrastructure. ArpC5 isoforms also affect the positioning of members of the Ena/Vasodilator-stimulated phosphoprotein (VASP) family of actin filament elongators, which mediate ArpC5 isoform–specific effects on the actin assembly level. Our results suggest that ArpC5 and Ena/VASP proteins are part of a signaling pathway enhancing cell migration. acknowledged_ssus: - _id: ScienComp - _id: LifeSc - _id: Bio - _id: EM-Fac acknowledgement: "We would like to thank K. von Peinen and B. Denker (Helmholtz Centre for Infection Research, Braunschweig, Germany) for experimental and technical assistance, respectively.\r\nThis research was supported by the Scientific Service Units (SSUs) of ISTA through resources provided by Scientific Computing (SciComp), the Life Science Facility (LSF), the Imaging and Optics facility (IOF), and the Electron Microscopy Facility (EMF). We acknowledge support from ISTA and from the Austrian Science Fund (FWF) (P33367) to F.K.M.S., from the Research Training Group GRK2223 and the Helmholtz Society to K.R,. and from the Deutsche Forschungsgemeinschaft (DFG) to J.F. and K.R." article_number: add6495 article_processing_charge: No article_type: original author: - first_name: Florian full_name: Fäßler, Florian id: 404F5528-F248-11E8-B48F-1D18A9856A87 last_name: Fäßler orcid: 0000-0001-7149-769X - first_name: Manjunath full_name: Javoor, Manjunath id: 305ab18b-dc7d-11ea-9b2f-b58195228ea2 last_name: Javoor orcid: 0000-0003-2311-2112 - first_name: Julia full_name: Datler, Julia id: 3B12E2E6-F248-11E8-B48F-1D18A9856A87 last_name: Datler orcid: 0000-0002-3616-8580 - first_name: Hermann full_name: Döring, Hermann last_name: Döring - first_name: Florian full_name: Hofer, Florian id: b9d234ba-9e33-11ed-95b6-cd561df280e6 last_name: Hofer - first_name: Georgi A full_name: Dimchev, Georgi A id: 38C393BE-F248-11E8-B48F-1D18A9856A87 last_name: Dimchev orcid: 0000-0001-8370-6161 - first_name: Victor-Valentin full_name: Hodirnau, Victor-Valentin id: 3661B498-F248-11E8-B48F-1D18A9856A87 last_name: Hodirnau orcid: 0000-0003-3904-947X - first_name: Jan full_name: Faix, Jan last_name: Faix - first_name: Klemens full_name: Rottner, Klemens last_name: Rottner - first_name: Florian KM full_name: Schur, Florian KM id: 48AD8942-F248-11E8-B48F-1D18A9856A87 last_name: Schur orcid: 0000-0003-4790-8078 citation: ama: Fäßler F, Javoor M, Datler J, et al. ArpC5 isoforms regulate Arp2/3 complex–dependent protrusion through differential Ena/VASP positioning. Science Advances. 2023;9(3). doi:10.1126/sciadv.add6495 apa: Fäßler, F., Javoor, M., Datler, J., Döring, H., Hofer, F., Dimchev, G. A., … Schur, F. K. (2023). ArpC5 isoforms regulate Arp2/3 complex–dependent protrusion through differential Ena/VASP positioning. Science Advances. American Association for the Advancement of Science. https://doi.org/10.1126/sciadv.add6495 chicago: Fäßler, Florian, Manjunath Javoor, Julia Datler, Hermann Döring, Florian Hofer, Georgi A Dimchev, Victor-Valentin Hodirnau, Jan Faix, Klemens Rottner, and Florian KM Schur. “ArpC5 Isoforms Regulate Arp2/3 Complex–Dependent Protrusion through Differential Ena/VASP Positioning.” Science Advances. American Association for the Advancement of Science, 2023. https://doi.org/10.1126/sciadv.add6495. ieee: F. Fäßler et al., “ArpC5 isoforms regulate Arp2/3 complex–dependent protrusion through differential Ena/VASP positioning,” Science Advances, vol. 9, no. 3. American Association for the Advancement of Science, 2023. ista: Fäßler F, Javoor M, Datler J, Döring H, Hofer F, Dimchev GA, Hodirnau V-V, Faix J, Rottner K, Schur FK. 2023. ArpC5 isoforms regulate Arp2/3 complex–dependent protrusion through differential Ena/VASP positioning. Science Advances. 9(3), add6495. mla: Fäßler, Florian, et al. “ArpC5 Isoforms Regulate Arp2/3 Complex–Dependent Protrusion through Differential Ena/VASP Positioning.” Science Advances, vol. 9, no. 3, add6495, American Association for the Advancement of Science, 2023, doi:10.1126/sciadv.add6495. short: F. Fäßler, M. Javoor, J. Datler, H. Döring, F. Hofer, G.A. Dimchev, V.-V. Hodirnau, J. Faix, K. Rottner, F.K. Schur, Science Advances 9 (2023). corr_author: '1' date_created: 2023-01-23T07:26:42Z date_published: 2023-01-20T00:00:00Z date_updated: 2026-04-07T12:59:44Z day: '20' ddc: - '570' department: - _id: FlSc - _id: EM-Fac doi: 10.1126/sciadv.add6495 external_id: isi: - '000964550100015' pmid: - '36662867' file: - access_level: open_access checksum: ce81a6d0b84170e5e8c62f6acfa15d9e content_type: application/pdf creator: dernst date_created: 2023-01-23T07:45:54Z date_updated: 2023-01-23T07:45:54Z file_id: '12335' file_name: 2023_ScienceAdvances_Faessler.pdf file_size: 1756234 relation: main_file success: 1 file_date_updated: 2023-01-23T07:45:54Z has_accepted_license: '1' intvolume: ' 9' isi: 1 issue: '3' keyword: - Multidisciplinary language: - iso: eng month: '01' oa: 1 oa_version: Published Version pmid: 1 project: - _id: 9B954C5C-BA93-11EA-9121-9846C619BF3A grant_number: P33367 name: Structure and isoform diversity of the Arp2/3 complex publication: Science Advances publication_identifier: issn: - 2375-2548 publication_status: published publisher: American Association for the Advancement of Science quality_controlled: '1' related_material: record: - id: '14562' relation: research_data status: public - id: '18766' relation: dissertation_contains status: public scopus_import: '1' status: public title: ArpC5 isoforms regulate Arp2/3 complex–dependent protrusion through differential Ena/VASP positioning tmp: image: /images/cc_by.png legal_code_url: https://creativecommons.org/licenses/by/4.0/legalcode name: Creative Commons Attribution 4.0 International Public License (CC-BY 4.0) short: CC BY (4.0) type: journal_article user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 volume: 9 year: '2023' ... --- OA_place: publisher _id: '12732' abstract: - lang: eng text: "Nonergodic systems, whose out-of-equilibrium dynamics fail to thermalize, provide a fascinating research direction both for fundamental reasons and for application in state of the art quantum devices.\r\nGoing beyond the description of statistical mechanics, ergodicity breaking yields a new paradigm in quantum many-body physics, introducing novel phases of matter with no counterpart at equilibrium.\r\nIn this Thesis, we address different open questions in the field, focusing on disorder-induced many-body localization (MBL) and on weak ergodicity breaking in kinetically constrained models.\r\nIn particular, we contribute to the debate about transport in kinetically constrained models, studying the effect of $U(1)$ conservation and inversion-symmetry breaking in a family of quantum East models.\r\nUsing tensor network techniques, we analyze the dynamics of large MBL systems beyond the limit of exact numerical methods.\r\nIn this setting, we approach the debated topic of the coexistence of localized and thermal eigenstates separated by energy thresholds known as many-body mobility edges.\r\nInspired by recent experiments, our work further investigates the localization of a small bath induced by the coupling to a large localized chain, the so-called MBL proximity effect.\r\n\r\nIn the first Chapter, we introduce a family of particle-conserving kinetically constrained models, inspired by the quantum East model.\r\nThe system we study features strong inversion-symmetry breaking, due to the nature of the correlated hopping.\r\nWe show that these models host so-called quantum Hilbert space fragmentation, consisting of disconnected subsectors in an entangled basis, and further provide an analytical description of this phenomenon.\r\nWe further probe its effect on dynamics of simple product states, showing revivals in fidelity and local observalbes.\r\nThe study of dynamics within the largest subsector reveals an anomalous transient superdiffusive behavior crossing over to slow logarithmic dynamics at later times.\r\nThis work suggests that particle conserving constrained models with inversion-symmetry breaking realize new universality classes of dynamics and invite their further theoretical and experimental studies.\r\n\r\nNext, we use kinetic constraints and disorder to design a model with many-body mobility edges in particle density.\r\nThis feature allows to study the dynamics of localized and thermal states in large systems beyond the limitations of previous studies.\r\nThe time-evolution shows typical signatures of localization at small densities, replaced by thermal behavior at larger densities.\r\nOur results provide evidence in favor of the stability of many-body mobility edges, which was recently challenged by a theoretical argument.\r\nTo support our findings, we probe the mechanism proposed as a cause of delocalization in many-body localized systems with mobility edges suggesting its ineffectiveness in the model studied.\r\n\r\nIn the last Chapter of this Thesis, we address the topic of many-body localization proximity effect.\r\nWe study a model inspired by recent experiments, featuring Anderson localized coupled to a small bath of free hard-core bosons.\r\nThe interaction among the two particle species results in non-trivial dynamics, which we probe using tensor network techniques.\r\nOur simulations show convincing evidence of many-body localization proximity effect when the bath is composed by a single free particle and interactions are strong.\r\nWe furthter observe an anomalous entanglement dynamics, which we explain through a phenomenological theory.\r\nFinally, we extract highly excited eigenstates of large systems, providing supplementary evidence in favor of our findings." acknowledged_ssus: - _id: ScienComp alternative_title: - ISTA Thesis article_processing_charge: No author: - first_name: Pietro full_name: Brighi, Pietro id: 4115AF5C-F248-11E8-B48F-1D18A9856A87 last_name: Brighi orcid: 0000-0002-7969-2729 citation: ama: Brighi P. Ergodicity breaking in disordered and kinetically constrained quantum many-body systems. 2023. doi:10.15479/at:ista:12732 apa: Brighi, P. (2023). Ergodicity breaking in disordered and kinetically constrained quantum many-body systems. Institute of Science and Technology Austria. https://doi.org/10.15479/at:ista:12732 chicago: Brighi, Pietro. “Ergodicity Breaking in Disordered and Kinetically Constrained Quantum Many-Body Systems.” Institute of Science and Technology Austria, 2023. https://doi.org/10.15479/at:ista:12732. ieee: P. Brighi, “Ergodicity breaking in disordered and kinetically constrained quantum many-body systems,” Institute of Science and Technology Austria, 2023. ista: Brighi P. 2023. Ergodicity breaking in disordered and kinetically constrained quantum many-body systems. Institute of Science and Technology Austria. mla: Brighi, Pietro. Ergodicity Breaking in Disordered and Kinetically Constrained Quantum Many-Body Systems. Institute of Science and Technology Austria, 2023, doi:10.15479/at:ista:12732. short: P. Brighi, Ergodicity Breaking in Disordered and Kinetically Constrained Quantum Many-Body Systems, Institute of Science and Technology Austria, 2023. corr_author: '1' date_created: 2023-03-17T13:30:48Z date_published: 2023-03-21T00:00:00Z date_updated: 2026-04-07T13:26:32Z day: '21' ddc: - '530' degree_awarded: PhD department: - _id: GradSch - _id: MaSe doi: 10.15479/at:ista:12732 ec_funded: 1 file: - access_level: closed checksum: 5d2de651ef9449c1b8dc27148ca74777 content_type: application/zip creator: pbrighi date_created: 2023-03-23T16:42:56Z date_updated: 2023-03-23T16:42:56Z file_id: '12753' file_name: Thesis_sub_PBrighi.zip file_size: 42167561 relation: source_file - access_level: open_access checksum: 7caa153d4a5b0873a79358787d2dfe1e content_type: application/pdf creator: pbrighi date_created: 2023-03-23T16:43:14Z date_updated: 2023-03-23T16:43:14Z file_id: '12754' file_name: Thesis_PBrighi.pdf file_size: 13977000 relation: main_file success: 1 file_date_updated: 2023-03-23T16:43:14Z has_accepted_license: '1' language: - iso: eng month: '03' oa: 1 oa_version: Published Version page: '158' project: - _id: 23841C26-32DE-11EA-91FC-C7463DDC885E call_identifier: H2020 grant_number: '850899' name: 'Non-Ergodic Quantum Matter: Universality, Dynamics and Control' publication_identifier: issn: - 2663-337X publication_status: published publisher: Institute of Science and Technology Austria related_material: record: - id: '12750' relation: part_of_dissertation status: public - id: '11470' relation: part_of_dissertation status: public - id: '8308' relation: part_of_dissertation status: public - id: '11469' relation: part_of_dissertation status: public status: public supervisor: - first_name: Maksym full_name: Serbyn, Maksym id: 47809E7E-F248-11E8-B48F-1D18A9856A87 last_name: Serbyn orcid: 0000-0002-2399-5827 title: Ergodicity breaking in disordered and kinetically constrained quantum many-body systems tmp: image: /images/cc_by_nc_sa.png legal_code_url: https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode name: Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) short: CC BY-NC-SA (4.0) type: dissertation user_id: ba8df636-2132-11f1-aed0-ed93e2281fdd year: '2023' ... --- _id: '12817' abstract: - lang: eng text: 3D-reconstruction of living brain tissue down to individual synapse level would create opportunities for decoding the dynamics and structure-function relationships of the brain’s complex and dense information processing network. However, it has been hindered by insufficient 3D-resolution, inadequate signal-to-noise-ratio, and prohibitive light burden in optical imaging, whereas electron microscopy is inherently static. Here we solved these challenges by developing an integrated optical/machine learning technology, LIONESS (Live Information-Optimized Nanoscopy Enabling Saturated Segmentation). It leverages optical modifications to stimulated emission depletion (STED) microscopy in comprehensively, extracellularly labelled tissue and prior information on sample structure via machine learning to simultaneously achieve isotropic super-resolution, high signal-to-noise-ratio, and compatibility with living tissue. This allows dense deep-learning-based instance segmentation and 3D-reconstruction at synapse level incorporating molecular, activity, and morphodynamic information. LIONESS opens up avenues for studying the dynamic functional (nano-)architecture of living brain tissue. acknowledged_ssus: - _id: ScienComp - _id: Bio - _id: PreCl - _id: LifeSc - _id: M-Shop - _id: E-Lib acknowledgement: 'We thank J. Vorlaufer, N. Agudelo, A. Wartak for microscope maintenance and troubleshooting, C. Kreuzinger and A. Freeman for technical assistance, and M. Šuplata for hardware control support, and Márcia Cunha dos Santos for initial exploration of software. We thank Paul Henderson for advice on deep-learning training and Michael Sixt, Scott Boyd, and Tamara Weiss for discussions and critical reading of the manuscript. Luke Lavis (Janelia Research Campus) generously provided JF585-HaloTag ligand. ' article_processing_charge: No author: - first_name: Johann G full_name: Danzl, Johann G id: 42EFD3B6-F248-11E8-B48F-1D18A9856A87 last_name: Danzl orcid: 0000-0001-8559-3973 citation: ama: Danzl JG. Research data for the publication “Dense 4D nanoscale reconstruction of living brain tissue.” 2023. doi:10.15479/AT:ISTA:12817 apa: Danzl, J. G. (2023). Research data for the publication “Dense 4D nanoscale reconstruction of living brain tissue.” Institute of Science and Technology Austria. https://doi.org/10.15479/AT:ISTA:12817 chicago: Danzl, Johann G. “Research Data for the Publication ‘Dense 4D Nanoscale Reconstruction of Living Brain Tissue.’” Institute of Science and Technology Austria, 2023. https://doi.org/10.15479/AT:ISTA:12817. ieee: J. G. Danzl, “Research data for the publication ‘Dense 4D nanoscale reconstruction of living brain tissue.’” Institute of Science and Technology Austria, 2023. ista: Danzl JG. 2023. Research data for the publication ‘Dense 4D nanoscale reconstruction of living brain tissue’, Institute of Science and Technology Austria, 10.15479/AT:ISTA:12817. mla: Danzl, Johann G. Research Data for the Publication “Dense 4D Nanoscale Reconstruction of Living Brain Tissue.” Institute of Science and Technology Austria, 2023, doi:10.15479/AT:ISTA:12817. short: J.G. Danzl, (2023). contributor: - first_name: Philipp id: 39BDC62C-F248-11E8-B48F-1D18A9856A87 last_name: Velicky orcid: 0000-0002-2340-7431 - first_name: Eder id: 3FB91342-F248-11E8-B48F-1D18A9856A87 last_name: Miguel Villalba - first_name: Julia M id: 443DB6DE-F248-11E8-B48F-1D18A9856A87 last_name: Michalska - first_name: Julia id: 46E28B80-F248-11E8-B48F-1D18A9856A87 last_name: Lyudchik - first_name: Donglai last_name: Wei - first_name: Zudi last_name: Lin - first_name: Jake id: 63836096-4690-11EA-BD4E-32803DDC885E last_name: Watson orcid: 0000-0002-8698-3823 - first_name: Jakob last_name: Troidl - first_name: Johanna last_name: Beyer - first_name: Yoav id: 43DF3136-F248-11E8-B48F-1D18A9856A87 last_name: Ben Simon - first_name: Christoph M id: 4DF26D8C-F248-11E8-B48F-1D18A9856A87 last_name: Sommer orcid: 0000-0003-1216-9105 - first_name: Wiebke id: 425C1CE8-F248-11E8-B48F-1D18A9856A87 last_name: Jahr - first_name: Alban id: 9ac8f577-2357-11eb-997a-e566c5550886 last_name: Cenameri - first_name: Johannes last_name: Broichhagen - first_name: 'Seth G. N. ' last_name: Grant - first_name: Peter M id: 353C1B58-F248-11E8-B48F-1D18A9856A87 last_name: Jonas orcid: 0000-0001-5001-4804 - first_name: Gaia id: 3E57A680-F248-11E8-B48F-1D18A9856A87 last_name: Novarino orcid: 0000-0002-7673-7178 - first_name: Hanspeter last_name: Pfister - first_name: Bernd id: 49876194-F248-11E8-B48F-1D18A9856A87 last_name: Bickel orcid: 0000-0001-6511-9385 corr_author: '1' date_created: 2023-04-07T11:37:40Z date_published: 2023-05-19T00:00:00Z date_updated: 2026-04-14T08:34:34Z day: '19' ddc: - '570' department: - _id: JoDa doi: 10.15479/AT:ISTA:12817 file: - access_level: open_access checksum: c1819889e72ec86ee1bba809e19739e0 content_type: text/plain creator: jdanzl date_created: 2023-05-18T17:06:12Z date_updated: 2023-05-18T17:06:12Z file_id: '13030' file_name: Readme.txt file_size: 651 relation: main_file success: 1 - access_level: open_access checksum: 8f6259fc5128ffcc0cd89d25d51995c1 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T19:51:52Z date_updated: 2023-05-18T19:51:52Z file_id: '13031' file_name: Fig1a_LIONESS.tif file_size: 347448884 relation: main_file success: 1 - access_level: open_access checksum: 204e8ef763b5a122fa95a594f5217d83 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T16:48:26Z date_updated: 2023-05-18T16:48:26Z file_id: '13029' file_name: Fig_1b.tif file_size: 91626 relation: main_file success: 1 - access_level: open_access checksum: dc521a4e6e80a57c100a06ba1e87c766 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T16:48:26Z date_updated: 2023-05-18T16:48:26Z file_id: '13028' file_name: Fig_1c.tif file_size: 90300 relation: main_file success: 1 - access_level: open_access checksum: 56d3d5dbe7e1f54dbecc32ad2d2cc6df content_type: image/tiff creator: jdanzl date_created: 2023-05-18T16:00:06Z date_updated: 2023-05-18T16:00:06Z file_id: '13022' file_name: Figure2a_low_exposure_input.tif file_size: 18391928 relation: main_file success: 1 - access_level: open_access checksum: 2c30dbd6c4db079f7fdc2eaeea5f9710 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:59:30Z date_updated: 2023-05-18T15:59:30Z file_id: '13021' file_name: Figure2a_LIONESS.tif file_size: 18392017 relation: main_file success: 1 - access_level: open_access checksum: 799868024bf734fecaf36f4ffd843ed2 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:58:54Z date_updated: 2023-05-18T15:58:54Z file_id: '13020' file_name: Figure2a_merge_LIONESS_EGFP.tif file_size: 36785440 relation: main_file success: 1 - access_level: open_access checksum: 522fe0fe505e437d6f637f9b8059dfaa content_type: image/tiff creator: jdanzl date_created: 2023-05-18T16:44:45Z date_updated: 2023-05-18T16:44:45Z file_id: '13025' file_name: Fig2c_LIONESS.tif file_size: 670051576 relation: main_file success: 1 - access_level: open_access checksum: c9c0d873341d2f974ba799fa71a1b789 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T16:23:10Z date_updated: 2023-05-18T16:23:10Z file_id: '13024' file_name: Fig2de_LIONESS_1.tif file_size: 412478217 relation: main_file success: 1 - access_level: open_access checksum: ff86ad8642c56ba1c67c814e20925a85 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T19:48:02Z date_updated: 2023-05-18T19:48:02Z file_id: '13032' file_name: Fig2de_LIONESS_2.tif file_size: 637360217 relation: main_file success: 1 - access_level: open_access checksum: 32681bd33673088193fad9f9c7b79090 content_type: application/zip creator: jdanzl date_created: 2023-05-18T14:39:05Z date_updated: 2023-05-18T14:39:05Z file_id: '12995' file_name: Fig_3_LIONESS.zip file_size: 46563159 relation: main_file success: 1 - access_level: open_access checksum: 17239ed0f54ac7495e2b1daa5d0864ef content_type: application/zip creator: jdanzl date_created: 2023-05-18T15:04:40Z date_updated: 2023-05-18T15:04:40Z file_id: '13001' file_name: Fig_4_LIONESS_synaptic_markers.zip file_size: 29159397 relation: main_file success: 1 - access_level: open_access checksum: c060c898f2bf93312a4c07dfdcb745cb content_type: application/zip creator: jdanzl date_created: 2023-05-18T14:58:09Z date_updated: 2023-05-18T14:58:09Z file_id: '12997' file_name: Fig_5a_time_series_Ca2+_imaging.zip file_size: 16563855 relation: main_file success: 1 - access_level: open_access checksum: 85b4c0cd099212ecf0beec3a78bb35c5 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:55:44Z date_updated: 2023-05-18T15:55:44Z file_id: '13017' file_name: Fig5b_LIONESS_T1.tif file_size: 30632526 relation: main_file success: 1 - access_level: open_access checksum: eb83649a0ebc2f5ea7a0308324ebf308 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:57:40Z date_updated: 2023-05-18T15:57:40Z file_id: '13018' file_name: Fig5b_LIONESS_T2.tif file_size: 57634426 relation: main_file success: 1 - access_level: open_access checksum: 40a2d93a34f83ce37043d8414d981b3d content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:14:33Z date_updated: 2023-05-18T15:14:33Z file_id: '13005' file_name: Extended_Data_Figure 3a_low_exposure.tif file_size: 12512676 relation: main_file success: 1 - access_level: open_access checksum: 516521bb400f5c058184994b57165f8b content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:15:01Z date_updated: 2023-05-18T15:15:01Z file_id: '13006' file_name: Extended_Data_Figure 3a_high_exposure.tif file_size: 12512678 relation: main_file success: 1 - access_level: open_access checksum: 1d3a21970307de2d84b614985ec0706d content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:17:08Z date_updated: 2023-05-18T15:17:08Z file_id: '13007' file_name: Extended_Data_Figure 3a_restored.tif file_size: 50038242 relation: main_file success: 1 - access_level: open_access checksum: 441575853bf667ed89cc218079cc397b content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:56:57Z date_updated: 2023-05-18T15:56:57Z file_id: '13016' file_name: Extended_Data_Figure 3b_example dataset_low_exposure_input.tif file_size: 68298045 relation: main_file success: 1 - access_level: open_access checksum: e499355760270e14bcf12bdc7960bd7a content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:54:43Z date_updated: 2023-05-18T15:54:43Z file_id: '13015' file_name: Extended_Data_Figure 3b_example dataset_merge_LIONESS_EGFP.tif file_size: 136597524 relation: main_file success: 1 - access_level: open_access checksum: add9f1adc6763091233dbe8504f1c396 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:50:19Z date_updated: 2023-05-18T15:50:19Z file_id: '13014' file_name: Extended_Data_Figure 3b_dataset_1_merge_LIONESS_EGFP.tif file_size: 76103800 relation: main_file success: 1 - access_level: open_access checksum: 311719efb1c68d7243be557b779173f2 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:51:20Z date_updated: 2023-05-18T15:51:20Z file_id: '13013' file_name: Extended_Data_Figure 3b_dataset_2_merge_LIONESS_EGFP.tif file_size: 49810007 relation: main_file success: 1 - access_level: open_access checksum: 8d5d9d859f8d099bb3b877c4e455f699 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:32:55Z date_updated: 2023-05-18T15:32:55Z file_id: '13010' file_name: Extended_Data_Figure 3b_dataset_3_merge_LIONESS_EGFP.tif file_size: 169060658 relation: main_file success: 1 - access_level: open_access checksum: 79eca23cc5e2fc1fe26e03e1bba2e90f content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:47:03Z date_updated: 2023-05-18T15:47:03Z file_id: '13011' file_name: Extended_Data_Figure 3b_dataset_4_merge_LIONESS_EGFP.tif file_size: 169060658 relation: main_file success: 1 - access_level: open_access checksum: 107bf2e0f8a14b4bd0169c9104765171 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:47:47Z date_updated: 2023-05-18T15:47:47Z file_id: '13012' file_name: Extended_Data_Figure 3b_dataset_5_merge_LIONESS_EGFP.tif file_size: 244861000 relation: main_file success: 1 - access_level: open_access checksum: b71e4a090dd9a8048aee847ae47624f0 content_type: application/zip creator: jdanzl date_created: 2023-05-18T14:36:34Z date_updated: 2023-05-18T14:36:34Z file_id: '12994' file_name: Extended_Data_Fig_4_LIONESS_TIF_stacks.zip file_size: 76795968 relation: main_file success: 1 - access_level: open_access checksum: ea011c3d1966ba79a62e8cf775e595c5 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T16:09:31Z date_updated: 2023-05-18T16:09:31Z file_id: '13023' file_name: Extended_Data_Fig5b.tif file_size: 288330201 relation: main_file success: 1 - access_level: open_access checksum: 1c56b5b324b6270b0df72a77e1e9d6f1 content_type: application/zip creator: jdanzl date_created: 2023-05-18T15:21:29Z date_updated: 2023-05-18T15:21:29Z file_id: '13003' file_name: Extended_Data_Fig_6_TIF_stacks.zip file_size: 255976297 relation: main_file success: 1 - access_level: open_access checksum: 89150ebb0044ead282b9048e55cd3961 content_type: application/zip creator: jdanzl date_created: 2023-05-18T15:08:20Z date_updated: 2023-05-18T15:08:20Z file_id: '13002' file_name: Extended_Data_Fig_7.zip file_size: 70694476 relation: main_file success: 1 - access_level: open_access checksum: 81e79a6d4e6745dce63cdad94bfcdca4 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T16:46:36Z date_updated: 2023-05-18T16:46:36Z file_id: '13026' file_name: Extended_Data_Fig_8b_T1.tif file_size: 57634426 relation: main_file success: 1 - access_level: open_access checksum: 53da2b10969a185239b317be0ffa61ef content_type: image/tiff creator: jdanzl date_created: 2023-05-18T16:48:26Z date_updated: 2023-05-18T16:48:26Z file_id: '13027' file_name: Extended_Data_Fig_8b_T2.tif file_size: 57634426 relation: main_file success: 1 - access_level: open_access checksum: 9a97ccd3eb0c75a5c4f142e5dadc0fb3 content_type: application/zip creator: jdanzl date_created: 2023-05-18T15:02:09Z date_updated: 2023-05-18T15:02:09Z file_id: '12999' file_name: Extended_Data_Fig_8c_png.zip file_size: 2942961 relation: main_file success: 1 - access_level: open_access checksum: baaee60e695160118ea2167c6cb91775 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:17:38Z date_updated: 2023-05-18T15:17:38Z file_id: '13008' file_name: Extended_Data_Fig9c_LIONESS.tif file_size: 11949362 relation: main_file success: 1 - access_level: open_access checksum: 095a980d0d746bc2277b0586dbb9d652 content_type: application/zip creator: jdanzl date_created: 2023-05-18T15:02:00Z date_updated: 2023-05-18T15:02:00Z file_id: '12998' file_name: Extended_Data_Fig_10_LIONESS_tif_xy_view.zip file_size: 72553526 relation: main_file success: 1 - access_level: open_access checksum: 343b87c4842eab9e5086f248668dec84 content_type: application/zip creator: jdanzl date_created: 2023-05-18T14:22:43Z date_updated: 2023-05-18T14:22:43Z file_id: '12992' file_name: Extended_Data_Fig_10_LIONESS_tif_xz_view.zip file_size: 72906090 relation: main_file success: 1 - access_level: open_access checksum: c3d4970e5185f72cb89f36005e83a556 content_type: application/zip creator: jdanzl date_created: 2023-05-18T15:14:01Z date_updated: 2023-05-18T15:14:01Z file_id: '13004' file_name: Supplementary_Fig_4.zip file_size: 428208 relation: main_file success: 1 - access_level: open_access checksum: f451852622f8ba8992582ab9b57c49e5 content_type: application/zip creator: jdanzl date_created: 2023-05-18T14:32:29Z date_updated: 2023-05-18T14:32:29Z file_id: '12991' file_name: Supplementary_Fig_8.zip file_size: 230506869 relation: main_file success: 1 - access_level: open_access checksum: 874b524f38fad6b86be14326bb5aa706 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T16:00:15Z date_updated: 2023-05-18T16:00:15Z file_id: '13019' file_name: Suppl_Fig_11_LIONESS.tif file_size: 40406386 relation: main_file success: 1 - access_level: open_access checksum: 7ab5cc6c581c463d1d7e430399b8b384 content_type: application/zip creator: jdanzl date_created: 2023-05-18T14:57:18Z date_updated: 2023-05-18T14:57:18Z file_id: '12996' file_name: Supplementary_Fig_12.zip file_size: 357705621 relation: main_file success: 1 - access_level: open_access checksum: 9763330288c2d544a3e1d7bc12657ee8 content_type: image/tiff creator: jdanzl date_created: 2023-05-18T15:39:45Z date_updated: 2023-05-18T15:39:45Z file_id: '13009' file_name: Supplementary_Fig_13.tif file_size: 617709326 relation: main_file success: 1 - access_level: open_access checksum: 1e83b64e220694942ccf0ca70f17e33b content_type: application/zip creator: jdanzl date_created: 2023-05-18T14:19:48Z date_updated: 2023-05-18T14:19:48Z file_id: '12990' file_name: GCaMP_recording_overview_time_series_Suppl_Video_5.zip file_size: 9387873 relation: main_file success: 1 - access_level: open_access checksum: fc7208299e6393e69b31a0051fa5ac2e content_type: application/zip creator: jdanzl date_created: 2023-05-18T15:13:55Z date_updated: 2023-05-18T15:13:55Z file_id: '12993' file_name: CARE model.zip file_size: 1218458059 relation: main_file success: 1 - access_level: open_access checksum: 47db2185ff2c6ba9b059c9b1a8882832 content_type: application/zip creator: jdanzl date_created: 2023-05-18T15:03:10Z date_updated: 2023-05-18T15:03:10Z file_id: '13000' file_name: Segmentation model.zip file_size: 19052980 relation: main_file success: 1 file_date_updated: 2023-05-18T19:51:52Z has_accepted_license: '1' month: '05' oa: 1 oa_version: Published Version publisher: Institute of Science and Technology Austria related_material: record: - id: '13267' relation: used_in_publication status: public status: public title: Research data for the publication "Dense 4D nanoscale reconstruction of living brain tissue" tmp: image: /images/cc_by_sa.png legal_code_url: https://creativecommons.org/licenses/by-sa/4.0/legalcode name: Creative Commons Attribution-ShareAlike 4.0 International Public License (CC BY-SA 4.0) short: CC BY-SA (4.0) type: research_data user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2023' ... --- _id: '13053' abstract: - lang: eng text: 'Deep neural networks (DNNs) often have to be compressed, via pruning and/or quantization, before they can be deployed in practical settings. In this work we propose a new compression-aware minimizer dubbed CrAM that modifies the optimization step in a principled way, in order to produce models whose local loss behavior is stable under compression operations such as pruning. Thus, dense models trained via CrAM should be compressible post-training, in a single step, without significant accuracy loss. Experimental results on standard benchmarks, such as residual networks for ImageNet classification and BERT models for language modelling, show that CrAM produces dense models that can be more accurate than the standard SGD/Adam-based baselines, but which are stable under weight pruning: specifically, we can prune models in one-shot to 70-80% sparsity with almost no accuracy loss, and to 90% with reasonable (∼1%) accuracy loss, which is competitive with gradual compression methods. Additionally, CrAM can produce sparse models which perform well for transfer learning, and it also works for semi-structured 2:4 pruning patterns supported by GPU hardware. The code for reproducing the results is available at this https URL .' acknowledged_ssus: - _id: ScienComp acknowledgement: "AP, EK, DA received funding from the European Research Council (ERC) under the European\r\nUnion’s Horizon 2020 research and innovation programme (grant agreement No 805223 ScaleML). AV acknowledges the support of the French Agence Nationale de la Recherche (ANR), under grant ANR-21-CE48-0016 (project COMCOPT). We further acknowledge the support from the Scientific Service Units (SSU) of ISTA through resources provided by Scientific Computing (SciComp)." article_processing_charge: No arxiv: 1 author: - first_name: Elena-Alexandra full_name: Peste, Elena-Alexandra id: 32D78294-F248-11E8-B48F-1D18A9856A87 last_name: Peste - first_name: Adrian full_name: Vladu, Adrian last_name: Vladu - first_name: Eldar full_name: Kurtic, Eldar id: 47beb3a5-07b5-11eb-9b87-b108ec578218 last_name: Kurtic - first_name: Christoph full_name: Lampert, Christoph id: 40C20FD2-F248-11E8-B48F-1D18A9856A87 last_name: Lampert orcid: 0000-0001-8622-7887 - first_name: Dan-Adrian full_name: Alistarh, Dan-Adrian id: 4A899BFC-F248-11E8-B48F-1D18A9856A87 last_name: Alistarh orcid: 0000-0003-3650-940X citation: ama: 'Krumes A, Vladu A, Kurtic E, Lampert C, Alistarh D-A. CrAM: A Compression-Aware Minimizer. In: 11th International Conference on Learning Representations . OpenReview; 2023.' apa: 'Krumes, A., Vladu, A., Kurtic, E., Lampert, C., & Alistarh, D.-A. (2023). CrAM: A Compression-Aware Minimizer. In 11th International Conference on Learning Representations . Kigali, Rwanda : OpenReview.' chicago: 'Krumes, Alexandra, Adrian Vladu, Eldar Kurtic, Christoph Lampert, and Dan-Adrian Alistarh. “CrAM: A Compression-Aware Minimizer.” In 11th International Conference on Learning Representations . OpenReview, 2023.' ieee: 'A. Krumes, A. Vladu, E. Kurtic, C. Lampert, and D.-A. Alistarh, “CrAM: A Compression-Aware Minimizer,” in 11th International Conference on Learning Representations , Kigali, Rwanda , 2023.' ista: 'Krumes A, Vladu A, Kurtic E, Lampert C, Alistarh D-A. 2023. CrAM: A Compression-Aware Minimizer. 11th International Conference on Learning Representations . ICLR: International Conference on Learning Representations.' mla: 'Krumes, Alexandra, et al. “CrAM: A Compression-Aware Minimizer.” 11th International Conference on Learning Representations , OpenReview, 2023.' short: A. Krumes, A. Vladu, E. Kurtic, C. Lampert, D.-A. Alistarh, in:, 11th International Conference on Learning Representations , OpenReview, 2023. conference: end_date: 2023-05-05 location: 'Kigali, Rwanda ' name: 'ICLR: International Conference on Learning Representations' start_date: 2023-05-01 corr_author: '1' date_created: 2023-05-23T11:36:18Z date_published: 2023-05-01T00:00:00Z date_updated: 2026-04-07T13:30:19Z day: '01' ddc: - '000' department: - _id: GradSch - _id: DaAl - _id: ChLa ec_funded: 1 external_id: arxiv: - '2207.14200' file: - access_level: open_access checksum: a6eec897e13a91cdc3eeaf309801752c content_type: application/pdf creator: dernst date_created: 2024-07-22T09:09:45Z date_updated: 2024-07-22T09:09:45Z file_id: '17294' file_name: 2023_ICLR_Peste.pdf file_size: 458201 relation: main_file success: 1 file_date_updated: 2024-07-22T09:09:45Z has_accepted_license: '1' language: - iso: eng main_file_link: - open_access: '1' url: https://openreview.net/pdf?id=_eTZBs-yedr month: '05' oa: 1 oa_version: Published Version project: - _id: 268A44D6-B435-11E9-9278-68D0E5697425 call_identifier: H2020 grant_number: '805223' name: Elastic Coordination for Scalable Machine Learning publication: '11th International Conference on Learning Representations ' publication_status: published publisher: OpenReview quality_controlled: '1' related_material: link: - relation: software url: https://github.com/IST-DASLab/CrAM record: - id: '13074' relation: dissertation_contains status: public status: public title: 'CrAM: A Compression-Aware Minimizer' type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2023' ... --- OA_place: publisher _id: '13074' abstract: - lang: eng text: "Deep learning has become an integral part of a large number of important applications, and many of the recent breakthroughs have been enabled by the ability to train very large models, capable to capture complex patterns and relationships from the data. At the same time, the massive sizes of modern deep learning models have made their deployment to smaller devices more challenging; this is particularly important, as in many applications the users rely on accurate deep learning predictions, but they only have access to devices with limited memory and compute power. One solution to this problem is to prune neural networks, by setting as many of their parameters as possible to zero, to obtain accurate sparse models with lower memory footprint. Despite the great research progress in obtaining sparse models that preserve accuracy, while satisfying memory and computational constraints, there are still many challenges associated with efficiently training sparse models, as well as understanding their generalization properties.\r\n\r\nThe focus of this thesis is to investigate how the training process of sparse models can be made more efficient, and to understand the differences between sparse and dense models in terms of how well they can generalize to changes in the data distribution. We first study a method for co-training sparse and dense models, at a lower cost compared to regular training. With our method we can obtain very accurate sparse networks, and dense models that can recover the baseline accuracy. Furthermore, we are able to more easily analyze the differences, at prediction level, between the sparse-dense model pairs. Next, we investigate the generalization properties of sparse neural networks in more detail, by studying how well different sparse models trained on a larger task can adapt to smaller, more specialized tasks, in a transfer learning scenario. Our analysis across multiple pruning methods and sparsity levels reveals that sparse models provide features that can transfer similarly to or better than the dense baseline. However, the choice of the pruning method plays an important role, and can influence the results when the features are fixed (linear finetuning), or when they are allowed to adapt to the new task (full finetuning). Using sparse models with fixed masks for finetuning on new tasks has an important practical advantage, as it enables training neural networks on smaller devices. However, one drawback of current pruning methods is that the entire training cycle has to be repeated to obtain the initial sparse model, for every sparsity target; in consequence, the entire training process is costly and also multiple models need to be stored. In the last part of the thesis we propose a method that can train accurate dense models that are compressible in a single step, to multiple sparsity levels, without additional finetuning. Our method results in sparse models that can be competitive with existing pruning methods, and which can also successfully generalize to new tasks." acknowledged_ssus: - _id: ScienComp alternative_title: - ISTA Thesis article_processing_charge: No author: - first_name: Elena-Alexandra full_name: Peste, Elena-Alexandra id: 32D78294-F248-11E8-B48F-1D18A9856A87 last_name: Peste citation: ama: Krumes A. Efficiency and generalization of sparse neural networks. 2023. doi:10.15479/at:ista:13074 apa: Krumes, A. (2023). Efficiency and generalization of sparse neural networks. Institute of Science and Technology Austria. https://doi.org/10.15479/at:ista:13074 chicago: Krumes, Alexandra. “Efficiency and Generalization of Sparse Neural Networks.” Institute of Science and Technology Austria, 2023. https://doi.org/10.15479/at:ista:13074. ieee: A. Krumes, “Efficiency and generalization of sparse neural networks,” Institute of Science and Technology Austria, 2023. ista: Krumes A. 2023. Efficiency and generalization of sparse neural networks. Institute of Science and Technology Austria. mla: Krumes, Alexandra. Efficiency and Generalization of Sparse Neural Networks. Institute of Science and Technology Austria, 2023, doi:10.15479/at:ista:13074. short: A. Krumes, Efficiency and Generalization of Sparse Neural Networks, Institute of Science and Technology Austria, 2023. corr_author: '1' date_created: 2023-05-23T17:07:53Z date_published: 2023-05-23T00:00:00Z date_updated: 2026-06-18T17:18:20Z day: '23' ddc: - '000' degree_awarded: PhD department: - _id: GradSch - _id: DaAl - _id: ChLa doi: 10.15479/at:ista:13074 ec_funded: 1 file: - access_level: open_access checksum: 6b3354968403cb9d48cc5a83611fb571 content_type: application/pdf creator: epeste date_created: 2023-05-24T16:11:16Z date_updated: 2023-05-24T16:11:16Z file_id: '13087' file_name: PhD_Thesis_Alexandra_Peste_final.pdf file_size: 2152072 relation: main_file success: 1 - access_level: closed checksum: 8d0df94bbcf4db72c991f22503b3fd60 content_type: application/zip creator: epeste date_created: 2023-05-24T16:12:59Z date_updated: 2023-05-24T16:12:59Z file_id: '13088' file_name: PhD_Thesis_APeste.zip file_size: 1658293 relation: source_file file_date_updated: 2023-05-24T16:12:59Z has_accepted_license: '1' language: - iso: eng month: '05' oa: 1 oa_version: Published Version page: '147' project: - _id: 2564DBCA-B435-11E9-9278-68D0E5697425 call_identifier: H2020 grant_number: '665385' name: International IST Doctoral Program - _id: 268A44D6-B435-11E9-9278-68D0E5697425 call_identifier: H2020 grant_number: '805223' name: Elastic Coordination for Scalable Machine Learning publication_identifier: issn: - 2663-337X publication_status: published publisher: Institute of Science and Technology Austria related_material: record: - id: '13053' relation: part_of_dissertation status: public - id: '12299' relation: part_of_dissertation status: public - id: '11458' relation: part_of_dissertation status: public status: public supervisor: - first_name: Christoph full_name: Lampert, Christoph id: 40C20FD2-F248-11E8-B48F-1D18A9856A87 last_name: Lampert orcid: 0000-0001-8622-7887 - first_name: Dan-Adrian full_name: Alistarh, Dan-Adrian id: 4A899BFC-F248-11E8-B48F-1D18A9856A87 last_name: Alistarh orcid: 0000-0003-3650-940X title: Efficiency and generalization of sparse neural networks type: dissertation user_id: ba8df636-2132-11f1-aed0-ed93e2281fdd year: '2023' ...