--- _id: '14219' abstract: - lang: eng text: "In this paper, we show that recent advances in self-supervised feature\r\nlearning enable unsupervised object discovery and semantic segmentation with a\r\nperformance that matches the state of the field on supervised semantic\r\nsegmentation 10 years ago. We propose a methodology based on unsupervised\r\nsaliency masks and self-supervised feature clustering to kickstart object\r\ndiscovery followed by training a semantic segmentation network on pseudo-labels\r\nto bootstrap the system on images with multiple objects. We present results on\r\nPASCAL VOC that go far beyond the current state of the art (50.0 mIoU), and we\r\nreport for the first time results on MS COCO for the whole set of 81 classes:\r\nour method discovers 34 categories with more than $20\\%$ IoU, while obtaining\r\nan average IoU of 19.6 for all 81 categories." article_processing_charge: No arxiv: 1 author: - first_name: Andrii full_name: Zadaianchuk, Andrii last_name: Zadaianchuk - first_name: Matthaeus full_name: Kleindessner, Matthaeus last_name: Kleindessner - first_name: Yi full_name: Zhu, Yi last_name: Zhu - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Thomas full_name: Brox, Thomas last_name: Brox citation: ama: 'Zadaianchuk A, Kleindessner M, Zhu Y, Locatello F, Brox T. Unsupervised semantic segmentation with self-supervised object-centric representations. In: The 11th International Conference on Learning Representations. ; 2023.' apa: Zadaianchuk, A., Kleindessner, M., Zhu, Y., Locatello, F., & Brox, T. (2023). Unsupervised semantic segmentation with self-supervised object-centric representations. In The 11th International Conference on Learning Representations. Kigali, Rwanda. chicago: Zadaianchuk, Andrii, Matthaeus Kleindessner, Yi Zhu, Francesco Locatello, and Thomas Brox. “Unsupervised Semantic Segmentation with Self-Supervised Object-Centric Representations.” In The 11th International Conference on Learning Representations, 2023. ieee: A. Zadaianchuk, M. Kleindessner, Y. Zhu, F. Locatello, and T. Brox, “Unsupervised semantic segmentation with self-supervised object-centric representations,” in The 11th International Conference on Learning Representations, Kigali, Rwanda, 2023. ista: 'Zadaianchuk A, Kleindessner M, Zhu Y, Locatello F, Brox T. 2023. Unsupervised semantic segmentation with self-supervised object-centric representations. The 11th International Conference on Learning Representations. ICLR: International Conference on Learning Representations.' mla: Zadaianchuk, Andrii, et al. “Unsupervised Semantic Segmentation with Self-Supervised Object-Centric Representations.” The 11th International Conference on Learning Representations, 2023. short: A. Zadaianchuk, M. Kleindessner, Y. Zhu, F. Locatello, T. Brox, in:, The 11th International Conference on Learning Representations, 2023. conference: end_date: 2023-05-05 location: Kigali, Rwanda name: 'ICLR: International Conference on Learning Representations' start_date: 2023-05-01 date_created: 2023-08-22T14:22:58Z date_published: 2023-05-01T00:00:00Z date_updated: 2023-09-13T11:25:43Z day: '01' department: - _id: FrLo extern: '1' external_id: arxiv: - '2207.05027' language: - iso: eng main_file_link: - open_access: '1' url: https://arxiv.org/abs/2207.05027 month: '05' oa: 1 oa_version: Preprint publication: The 11th International Conference on Learning Representations publication_status: published quality_controlled: '1' status: public title: Unsupervised semantic segmentation with self-supervised object-centric representations type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2023' ... --- _id: '14222' abstract: - lang: eng text: Learning generative object models from unlabelled videos is a long standing problem and required for causal scene modeling. We decompose this problem into three easier subtasks, and provide candidate solutions for each of them. Inspired by the Common Fate Principle of Gestalt Psychology, we first extract (noisy) masks of moving objects via unsupervised motion segmentation. Second, generative models are trained on the masks of the background and the moving objects, respectively. Third, background and foreground models are combined in a conditional "dead leaves" scene model to sample novel scene configurations where occlusions and depth layering arise naturally. To evaluate the individual stages, we introduce the Fishbowl dataset positioned between complex real-world scenes and common object-centric benchmarks of simplistic objects. We show that our approach allows learning generative models that generalize beyond the occlusions present in the input videos, and represent scenes in a modular fashion that allows sampling plausible scenes outside the training distribution by permitting, for instance, object numbers or densities not observed in the training set. article_number: '2110.06562' article_processing_charge: No arxiv: 1 author: - first_name: Matthias full_name: Tangemann, Matthias last_name: Tangemann - first_name: Steffen full_name: Schneider, Steffen last_name: Schneider - first_name: Julius von full_name: Kügelgen, Julius von last_name: Kügelgen - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Peter full_name: Gehler, Peter last_name: Gehler - first_name: Thomas full_name: Brox, Thomas last_name: Brox - first_name: Matthias full_name: Kümmerer, Matthias last_name: Kümmerer - first_name: Matthias full_name: Bethge, Matthias last_name: Bethge - first_name: Bernhard full_name: Schölkopf, Bernhard last_name: Schölkopf citation: ama: 'Tangemann M, Schneider S, Kügelgen J von, et al. Unsupervised object learning via common fate. In: 2nd Conference on Causal Learning and Reasoning. ; 2023.' apa: Tangemann, M., Schneider, S., Kügelgen, J. von, Locatello, F., Gehler, P., Brox, T., … Schölkopf, B. (2023). Unsupervised object learning via common fate. In 2nd Conference on Causal Learning and Reasoning. Tübingen, Germany. chicago: Tangemann, Matthias, Steffen Schneider, Julius von Kügelgen, Francesco Locatello, Peter Gehler, Thomas Brox, Matthias Kümmerer, Matthias Bethge, and Bernhard Schölkopf. “Unsupervised Object Learning via Common Fate.” In 2nd Conference on Causal Learning and Reasoning, 2023. ieee: M. Tangemann et al., “Unsupervised object learning via common fate,” in 2nd Conference on Causal Learning and Reasoning, Tübingen, Germany, 2023. ista: 'Tangemann M, Schneider S, Kügelgen J von, Locatello F, Gehler P, Brox T, Kümmerer M, Bethge M, Schölkopf B. 2023. Unsupervised object learning via common fate. 2nd Conference on Causal Learning and Reasoning. CLeaR: Conference on Causal Learning and Reasoning, 2110.06562.' mla: Tangemann, Matthias, et al. “Unsupervised Object Learning via Common Fate.” 2nd Conference on Causal Learning and Reasoning, 2110.06562, 2023. short: M. Tangemann, S. Schneider, J. von Kügelgen, F. Locatello, P. Gehler, T. Brox, M. Kümmerer, M. Bethge, B. Schölkopf, in:, 2nd Conference on Causal Learning and Reasoning, 2023. conference: end_date: 2023-04-14 location: Tübingen, Germany name: 'CLeaR: Conference on Causal Learning and Reasoning' start_date: 2023-04-11 date_created: 2023-08-22T14:23:54Z date_published: 2023-04-15T00:00:00Z date_updated: 2023-09-13T11:31:14Z day: '15' department: - _id: FrLo extern: '1' external_id: arxiv: - '2110.06562' language: - iso: eng main_file_link: - open_access: '1' url: https://arxiv.org/abs/2110.06562 month: '04' oa: 1 oa_version: Preprint publication: 2nd Conference on Causal Learning and Reasoning publication_status: published quality_controlled: '1' status: public title: Unsupervised object learning via common fate type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2023' ... --- _id: '14333' abstract: - lang: eng text: "As causal ground truth is incredibly rare, causal discovery algorithms are\r\ncommonly only evaluated on simulated data. This is concerning, given that\r\nsimulations reflect common preconceptions about generating processes regarding\r\nnoise distributions, model classes, and more. In this work, we propose a novel\r\nmethod for falsifying the output of a causal discovery algorithm in the absence\r\nof ground truth. Our key insight is that while statistical learning seeks\r\nstability across subsets of data points, causal learning should seek stability\r\nacross subsets of variables. Motivated by this insight, our method relies on a\r\nnotion of compatibility between causal graphs learned on different subsets of\r\nvariables. We prove that detecting incompatibilities can falsify wrongly\r\ninferred causal relations due to violation of assumptions or errors from finite\r\nsample effects. Although passing such compatibility tests is only a necessary\r\ncriterion for good performance, we argue that it provides strong evidence for\r\nthe causal models whenever compatibility entails strong implications for the\r\njoint distribution. We also demonstrate experimentally that detection of\r\nincompatibilities can aid in causal model selection." article_number: '2307.09552' article_processing_charge: No arxiv: 1 author: - first_name: Philipp M. full_name: Faller, Philipp M. last_name: Faller - first_name: Leena Chennuru full_name: Vankadara, Leena Chennuru last_name: Vankadara - first_name: Atalanti A. full_name: Mastakouri, Atalanti A. last_name: Mastakouri - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Dominik full_name: Janzing, Dominik last_name: Janzing citation: ama: 'Faller PM, Vankadara LC, Mastakouri AA, Locatello F, Janzing D. Self-compatibility: Evaluating causal discovery without ground truth. arXiv. doi:10.48550/arXiv.2307.09552' apa: 'Faller, P. M., Vankadara, L. C., Mastakouri, A. A., Locatello, F., & Janzing, D. (n.d.). Self-compatibility: Evaluating causal discovery without ground truth. arXiv. https://doi.org/10.48550/arXiv.2307.09552' chicago: 'Faller, Philipp M., Leena Chennuru Vankadara, Atalanti A. Mastakouri, Francesco Locatello, and Dominik Janzing. “Self-Compatibility: Evaluating Causal Discovery without Ground Truth.” ArXiv, n.d. https://doi.org/10.48550/arXiv.2307.09552.' ieee: 'P. M. Faller, L. C. Vankadara, A. A. Mastakouri, F. Locatello, and D. Janzing, “Self-compatibility: Evaluating causal discovery without ground truth,” arXiv. .' ista: 'Faller PM, Vankadara LC, Mastakouri AA, Locatello F, Janzing D. Self-compatibility: Evaluating causal discovery without ground truth. arXiv, 2307.09552.' mla: 'Faller, Philipp M., et al. “Self-Compatibility: Evaluating Causal Discovery without Ground Truth.” ArXiv, 2307.09552, doi:10.48550/arXiv.2307.09552.' short: P.M. Faller, L.C. Vankadara, A.A. Mastakouri, F. Locatello, D. Janzing, ArXiv (n.d.). date_created: 2023-09-13T12:44:59Z date_published: 2023-07-18T00:00:00Z date_updated: 2024-10-14T12:31:20Z day: '18' department: - _id: FrLo doi: 10.48550/arXiv.2307.09552 extern: '1' external_id: arxiv: - '2307.09552' fulldoi: https://doi.org/10.48550/arXiv.2307.09552 language: - iso: eng main_file_link: - open_access: '1' url: https://doi.org/10.48550/arXiv.2307.09552 month: '07' oa: 1 oa_version: Preprint publication: arXiv publication_status: submitted status: public title: 'Self-compatibility: Evaluating causal discovery without ground truth' type: preprint user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2023' ... --- _id: '14949' abstract: - lang: eng text: Many approaches have been proposed to use diffusion models to augment training datasets for downstream tasks, such as classification. However, diffusion models are themselves trained on large datasets, often with noisy annotations, and it remains an open question to which extent these models contribute to downstream classification performance. In particular, it remains unclear if they generalize enough to improve over directly using the additional data of their pre-training process for augmentation. We systematically evaluate a range of existing methods to generate images from diffusion models and study new extensions to assess their benefit for data augmentation. Personalizing diffusion models towards the target data outperforms simpler prompting strategies. However, using the pre-training data of the diffusion model alone, via a simple nearest-neighbor retrieval procedure, leads to even stronger downstream performance. Our study explores the potential of diffusion models in generating new training data, and surprisingly finds that these sophisticated models are not yet able to beat a simple and strong image retrieval baseline on simple downstream vision tasks. acknowledgement: The authors would like to thank Varad Gunjal and Vishaal Udandarao. MFB thanks the International Max Planck Research School for Intelligent Systems (IMPRS-IS). alternative_title: - TMLR article_processing_charge: No article_type: original author: - first_name: Max full_name: Burg, Max last_name: Burg - first_name: Florian full_name: Wenzel, Florian last_name: Wenzel - first_name: Dominik full_name: Zietlow, Dominik last_name: Zietlow - first_name: Max full_name: Horn, Max last_name: Horn - first_name: Osama full_name: Makansi, Osama last_name: Makansi - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Chris full_name: Russell, Chris last_name: Russell citation: ama: Burg M, Wenzel F, Zietlow D, et al. Image retrieval outperforms diffusion models on data augmentation. Journal of Machine Learning Research. 2023. apa: Burg, M., Wenzel, F., Zietlow, D., Horn, M., Makansi, O., Locatello, F., & Russell, C. (2023). Image retrieval outperforms diffusion models on data augmentation. Journal of Machine Learning Research. ML Research Press. chicago: Burg, Max, Florian Wenzel, Dominik Zietlow, Max Horn, Osama Makansi, Francesco Locatello, and Chris Russell. “Image Retrieval Outperforms Diffusion Models on Data Augmentation.” Journal of Machine Learning Research. ML Research Press, 2023. ieee: M. Burg et al., “Image retrieval outperforms diffusion models on data augmentation,” Journal of Machine Learning Research. ML Research Press, 2023. ista: Burg M, Wenzel F, Zietlow D, Horn M, Makansi O, Locatello F, Russell C. 2023. Image retrieval outperforms diffusion models on data augmentation. Journal of Machine Learning Research. mla: Burg, Max, et al. “Image Retrieval Outperforms Diffusion Models on Data Augmentation.” Journal of Machine Learning Research, ML Research Press, 2023. short: M. Burg, F. Wenzel, D. Zietlow, M. Horn, O. Makansi, F. Locatello, C. Russell, Journal of Machine Learning Research (2023). date_created: 2024-02-07T14:57:39Z date_published: 2023-12-10T00:00:00Z date_updated: 2024-02-12T08:30:21Z day: '10' ddc: - '000' department: - _id: FrLo file: - access_level: open_access checksum: af87ddea7908923426365347b9c87ba7 content_type: application/pdf creator: ptazenko date_created: 2024-02-07T14:57:32Z date_updated: 2024-02-07T14:57:32Z file_id: '14950' file_name: Burg_et_al_2023_Image_retrieval_outperforms.pdf file_size: 27325153 relation: main_file file_date_updated: 2024-02-07T14:57:32Z has_accepted_license: '1' language: - iso: eng main_file_link: - open_access: '1' url: https://openreview.net/forum?id=xflYdGZMpv month: '12' oa: 1 oa_version: Published Version publication: Journal of Machine Learning Research publication_identifier: eissn: - 2835-8856 publication_status: published publisher: ML Research Press quality_controlled: '1' status: public title: Image retrieval outperforms diffusion models on data augmentation 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 year: '2023' ... --- OA_place: repository OA_type: green _id: '14958' abstract: - lang: eng text: Causal representation learning (CRL) aims at identifying high-level causal variables from low-level data, e.g. images. Current methods usually assume that all causal variables are captured in the high-dimensional observations. In this work, we focus on learning causal representations from data under partial observability, i.e., when some of the causal variables are not observed in the measurements, and the set of masked variables changes across the different samples. We introduce some initial theoretical results for identifying causal variables under partial observability by exploiting a sparsity regularizer, focusing in particular on the linear and piecewise linear mixing function case. We provide a theorem that allows us to identify the causal variables up to permutation and element-wise linear transformations in the linear case and a lemma that allows us to identify causal variables up to linear transformation in the piecewise case. Finally, we provide a conjecture that would allow us to identify the causal variables up to permutation and element-wise linear transformations also in the piecewise linear case. We test the theorem and conjecture on simulated data, showing the effectiveness of our method. acknowledgement: "This work was initiated at the Second Bellairs Workshop on Causality held at the Bellairs Research Institute, January 6–13, 2022; we thank all workshop participants for providing a stimulating research environment. The research of DX and SM was supported by the Air Force Office of Scientific Research under award number FA8655-22-1-7155. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the United States Air Force. We also thank SURF for the support in using the Dutch National Supercomputer Snellius. DY was supported by an Amazon fellowship and the International Max Planck Research School for Intelligent Systems (IMPRS-IS). Work done outside of Amazon. SL was supported by an IVADO excellence PhD scholarship and by Samsung Electronics Co., Ldt. JvK acknowledges support from the German Federal Ministry of Education and Research (BMBF)\r\nthrough the Tübingen AI Center (FKZ: 01IS18039B).\r\n" article_number: '54' article_processing_charge: No author: - first_name: Danru full_name: Xu, Danru last_name: Xu - first_name: Dingling full_name: Yao, Dingling id: d3e02e50-48a8-11ee-8f62-c108061797fa last_name: Yao - first_name: Sebastien full_name: Lachapelle, Sebastien last_name: Lachapelle - first_name: Perouz full_name: Taslakian, Perouz last_name: Taslakian - first_name: Julius full_name: von Kügelgen, Julius last_name: von Kügelgen - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Sara full_name: Magliacane, Sara last_name: Magliacane citation: ama: 'Xu D, Yao D, Lachapelle S, et al. A sparsity principle for partially observable causal representation learning. In: Causal Representation Learning Workshop at NeurIPS 2023. OpenReview; 2023.' apa: 'Xu, D., Yao, D., Lachapelle, S., Taslakian, P., von Kügelgen, J., Locatello, F., & Magliacane, S. (2023). A sparsity principle for partially observable causal representation learning. In Causal Representation Learning Workshop at NeurIPS 2023. New Orleans, LA, United States: OpenReview.' chicago: Xu, Danru, Dingling Yao, Sebastien Lachapelle, Perouz Taslakian, Julius von Kügelgen, Francesco Locatello, and Sara Magliacane. “A Sparsity Principle for Partially Observable Causal Representation Learning.” In Causal Representation Learning Workshop at NeurIPS 2023. OpenReview, 2023. ieee: D. Xu et al., “A sparsity principle for partially observable causal representation learning,” in Causal Representation Learning Workshop at NeurIPS 2023, New Orleans, LA, United States, 2023. ista: 'Xu D, Yao D, Lachapelle S, Taslakian P, von Kügelgen J, Locatello F, Magliacane S. 2023. A sparsity principle for partially observable causal representation learning. Causal Representation Learning Workshop at NeurIPS 2023. CRL: Causal Representation Learning Workshop at NeurIPS, 54.' mla: Xu, Danru, et al. “A Sparsity Principle for Partially Observable Causal Representation Learning.” Causal Representation Learning Workshop at NeurIPS 2023, 54, OpenReview, 2023. short: D. Xu, D. Yao, S. Lachapelle, P. Taslakian, J. von Kügelgen, F. Locatello, S. Magliacane, in:, Causal Representation Learning Workshop at NeurIPS 2023, OpenReview, 2023. conference: end_date: 2023-12-15 location: New Orleans, LA, United States name: 'CRL: Causal Representation Learning Workshop at NeurIPS' start_date: 2023-12-15 date_created: 2024-02-07T15:17:51Z date_published: 2023-12-05T00:00:00Z date_updated: 2025-02-04T12:37:34Z day: '05' ddc: - '000' department: - _id: FrLo file: - access_level: open_access checksum: 484efc27bda75ed6666044989695d9b6 content_type: application/pdf creator: dernst date_created: 2024-02-13T08:50:53Z date_updated: 2024-02-13T08:50:53Z file_id: '14982' file_name: 2023_CRL_Xu.pdf file_size: 552357 relation: main_file success: 1 file_date_updated: 2024-02-13T08:50:53Z has_accepted_license: '1' language: - iso: eng main_file_link: - open_access: '1' url: https://openreview.net/forum?id=Whr6uobelR month: '12' oa: 1 oa_version: Published Version publication: Causal Representation Learning Workshop at NeurIPS 2023 publication_status: published publisher: OpenReview quality_controlled: '1' status: public title: A sparsity principle for partially observable causal representation learning 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: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2023' ... --- _id: '14961' abstract: - lang: eng text: "The use of simulated data in the field of causal discovery is ubiquitous due to the scarcity of annotated real data. Recently, Reisach et al., 2021 highlighted the emergence of patterns in simulated linear data, which displays increasing marginal variance in the casual direction. As an ablation in their experiments, Montagna et al., 2023 found that similar patterns may emerge in\r\nnonlinear models for the variance of the score vector $\\nabla \\log p_{\\mathbf{X}}$, and introduced the ScoreSort algorithm. In this work, we formally define and characterize this score-sortability pattern of nonlinear additive noise models. We find that it defines a class of identifiable (bivariate) causal models overlapping with nonlinear additive noise models. We\r\ntheoretically demonstrate the advantages of ScoreSort in terms of statistical efficiency compared to prior state-of-the-art score matching-based methods and empirically show the score-sortability of the most common synthetic benchmarks in the literature. Our findings remark (1) the lack of diversity in the data as an important limitation in the evaluation of nonlinear causal discovery approaches, (2) the importance of thoroughly testing different settings within a problem class, and (3) the importance of analyzing statistical properties in\r\ncausal discovery, where research is often limited to defining identifiability conditions of the model. " article_number: '2310.14246' article_processing_charge: No arxiv: 1 author: - first_name: Francesco full_name: Montagna, Francesco last_name: Montagna - first_name: Nicoletta full_name: Noceti, Nicoletta last_name: Noceti - first_name: Lorenzo full_name: Rosasco, Lorenzo last_name: Rosasco - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 citation: ama: Montagna F, Noceti N, Rosasco L, Locatello F. Shortcuts for causal discovery of nonlinear models by score matching. arXiv. doi:10.48550/arXiv.2310.14246 apa: Montagna, F., Noceti, N., Rosasco, L., & Locatello, F. (n.d.). Shortcuts for causal discovery of nonlinear models by score matching. arXiv. https://doi.org/10.48550/arXiv.2310.14246 chicago: Montagna, Francesco, Nicoletta Noceti, Lorenzo Rosasco, and Francesco Locatello. “Shortcuts for Causal Discovery of Nonlinear Models by Score Matching.” ArXiv, n.d. https://doi.org/10.48550/arXiv.2310.14246. ieee: F. Montagna, N. Noceti, L. Rosasco, and F. Locatello, “Shortcuts for causal discovery of nonlinear models by score matching,” arXiv. . ista: Montagna F, Noceti N, Rosasco L, Locatello F. Shortcuts for causal discovery of nonlinear models by score matching. arXiv, 2310.14246. mla: Montagna, Francesco, et al. “Shortcuts for Causal Discovery of Nonlinear Models by Score Matching.” ArXiv, 2310.14246, doi:10.48550/arXiv.2310.14246. short: F. Montagna, N. Noceti, L. Rosasco, F. Locatello, ArXiv (n.d.). corr_author: '1' date_created: 2024-02-08T15:31:46Z date_published: 2023-10-22T00:00:00Z date_updated: 2024-10-09T21:08:10Z day: '22' department: - _id: FrLo doi: 10.48550/arXiv.2310.14246 external_id: arxiv: - '2310.14246' fulldoi: https://doi.org/10.48550/arXiv.2310.14246 language: - iso: eng main_file_link: - open_access: '1' url: https://doi.org/10.48550/arXiv.2310.14246 month: '10' oa: 1 oa_version: Preprint publication: arXiv publication_status: submitted status: public title: Shortcuts for causal discovery of nonlinear models by score matching type: preprint user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2023' ... --- _id: '14962' abstract: - lang: eng text: "In this paper, we show that recent advances in video representation learning\r\nand pre-trained vision-language models allow for substantial improvements in\r\nself-supervised video object localization. We propose a method that first\r\nlocalizes objects in videos via a slot attention approach and then assigns text\r\nto the obtained slots. The latter is achieved by an unsupervised way to read\r\nlocalized semantic information from the pre-trained CLIP model. The resulting\r\nvideo object localization is entirely unsupervised apart from the implicit\r\nannotation contained in CLIP, and it is effectively the first unsupervised\r\napproach that yields good results on regular video benchmarks." article_number: '2309.09858' article_processing_charge: No arxiv: 1 author: - first_name: Ke full_name: Fan, Ke last_name: Fan - first_name: Zechen full_name: Bai, Zechen last_name: Bai - first_name: Tianjun full_name: Xiao, Tianjun last_name: Xiao - first_name: Dominik full_name: Zietlow, Dominik last_name: Zietlow - first_name: Max full_name: Horn, Max last_name: Horn - first_name: Zixu full_name: Zhao, Zixu last_name: Zhao - first_name: Carl-Johann Simon-Gabriel full_name: Carl-Johann Simon-Gabriel, Carl-Johann Simon-Gabriel last_name: Carl-Johann Simon-Gabriel - first_name: Mike Zheng full_name: Shou, Mike Zheng last_name: Shou - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Bernt full_name: Schiele, Bernt last_name: Schiele - first_name: Thomas full_name: Brox, Thomas last_name: Brox - first_name: Zheng full_name: Zhang, Zheng last_name: Zhang - first_name: Yanwei full_name: Fu, Yanwei last_name: Fu - first_name: Tong full_name: He, Tong last_name: He citation: ama: Fan K, Bai Z, Xiao T, et al. Unsupervised open-vocabulary object localization in videos. arXiv. doi:10.48550/arXiv.2309.09858 apa: Fan, K., Bai, Z., Xiao, T., Zietlow, D., Horn, M., Zhao, Z., … He, T. (n.d.). Unsupervised open-vocabulary object localization in videos. arXiv. https://doi.org/10.48550/arXiv.2309.09858 chicago: Fan, Ke, Zechen Bai, Tianjun Xiao, Dominik Zietlow, Max Horn, Zixu Zhao, Carl-Johann Simon-Gabriel Carl-Johann Simon-Gabriel, et al. “Unsupervised Open-Vocabulary Object Localization in Videos.” ArXiv, n.d. https://doi.org/10.48550/arXiv.2309.09858. ieee: K. Fan et al., “Unsupervised open-vocabulary object localization in videos,” arXiv. . ista: Fan K, Bai Z, Xiao T, Zietlow D, Horn M, Zhao Z, Carl-Johann Simon-Gabriel C-JS-G, Shou MZ, Locatello F, Schiele B, Brox T, Zhang Z, Fu Y, He T. Unsupervised open-vocabulary object localization in videos. arXiv, 2309.09858. mla: Fan, Ke, et al. “Unsupervised Open-Vocabulary Object Localization in Videos.” ArXiv, 2309.09858, doi:10.48550/arXiv.2309.09858. short: K. Fan, Z. Bai, T. Xiao, D. Zietlow, M. Horn, Z. Zhao, C.-J.S.-G. Carl-Johann Simon-Gabriel, M.Z. Shou, F. Locatello, B. Schiele, T. Brox, Z. Zhang, Y. Fu, T. He, ArXiv (n.d.). date_created: 2024-02-08T15:33:39Z date_published: 2023-09-18T00:00:00Z date_updated: 2024-02-12T10:12:22Z day: '18' department: - _id: FrLo doi: 10.48550/arXiv.2309.09858 extern: '1' external_id: arxiv: - '2309.09858' fulldoi: https://doi.org/10.48550/arXiv.2309.09858 language: - iso: eng main_file_link: - open_access: '1' url: https://doi.org/10.48550/arXiv.2309.09858 month: '09' oa: 1 oa_version: Preprint publication: arXiv publication_status: submitted status: public title: Unsupervised open-vocabulary object localization in videos type: preprint user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2023' ... --- _id: '14963' abstract: - lang: eng text: "Unsupervised object-centric learning methods allow the partitioning of scenes\r\ninto entities without additional localization information and are excellent\r\ncandidates for reducing the annotation burden of multiple-object tracking (MOT)\r\npipelines. Unfortunately, they lack two key properties: objects are often split\r\ninto parts and are not consistently tracked over time. In fact,\r\nstate-of-the-art models achieve pixel-level accuracy and temporal consistency\r\nby relying on supervised object detection with additional ID labels for the\r\nassociation through time. This paper proposes a video object-centric model for\r\nMOT. It consists of an index-merge module that adapts the object-centric slots\r\ninto detection outputs and an object memory module that builds complete object\r\nprototypes to handle occlusions. Benefited from object-centric learning, we\r\nonly require sparse detection labels (0%-6.25%) for object localization and\r\nfeature binding. Relying on our self-supervised\r\nExpectation-Maximization-inspired loss for object association, our approach\r\nrequires no ID labels. Our experiments significantly narrow the gap between the\r\nexisting object-centric model and the fully supervised state-of-the-art and\r\noutperform several unsupervised trackers." article_number: '2309.00233' article_processing_charge: No arxiv: 1 author: - first_name: Zixu full_name: Zhao, Zixu last_name: Zhao - first_name: Jiaze full_name: Wang, Jiaze last_name: Wang - first_name: Max full_name: Horn, Max last_name: Horn - first_name: Yizhuo full_name: Ding, Yizhuo last_name: Ding - first_name: Tong full_name: He, Tong last_name: He - first_name: Zechen full_name: Bai, Zechen last_name: Bai - first_name: Dominik full_name: Zietlow, Dominik last_name: Zietlow - first_name: Carl-Johann Simon-Gabriel full_name: Carl-Johann Simon-Gabriel, Carl-Johann Simon-Gabriel last_name: Carl-Johann Simon-Gabriel - first_name: Bing full_name: Shuai, Bing last_name: Shuai - first_name: Zhuowen full_name: Tu, Zhuowen last_name: Tu - first_name: Thomas full_name: Brox, Thomas last_name: Brox - first_name: Bernt full_name: Schiele, Bernt last_name: Schiele - first_name: Yanwei full_name: Fu, Yanwei last_name: Fu - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Zheng full_name: Zhang, Zheng last_name: Zhang - first_name: Tianjun full_name: Xiao, Tianjun last_name: Xiao citation: ama: Zhao Z, Wang J, Horn M, et al. Object-centric multiple object tracking. arXiv. doi:10.48550/arXiv.2309.00233 apa: Zhao, Z., Wang, J., Horn, M., Ding, Y., He, T., Bai, Z., … Xiao, T. (n.d.). Object-centric multiple object tracking. arXiv. https://doi.org/10.48550/arXiv.2309.00233 chicago: Zhao, Zixu, Jiaze Wang, Max Horn, Yizhuo Ding, Tong He, Zechen Bai, Dominik Zietlow, et al. “Object-Centric Multiple Object Tracking.” ArXiv, n.d. https://doi.org/10.48550/arXiv.2309.00233. ieee: Z. Zhao et al., “Object-centric multiple object tracking,” arXiv. . ista: Zhao Z, Wang J, Horn M, Ding Y, He T, Bai Z, Zietlow D, Carl-Johann Simon-Gabriel C-JS-G, Shuai B, Tu Z, Brox T, Schiele B, Fu Y, Locatello F, Zhang Z, Xiao T. Object-centric multiple object tracking. arXiv, 2309.00233. mla: Zhao, Zixu, et al. “Object-Centric Multiple Object Tracking.” ArXiv, 2309.00233, doi:10.48550/arXiv.2309.00233. short: Z. Zhao, J. Wang, M. Horn, Y. Ding, T. He, Z. Bai, D. Zietlow, C.-J.S.-G. Carl-Johann Simon-Gabriel, B. Shuai, Z. Tu, T. Brox, B. Schiele, Y. Fu, F. Locatello, Z. Zhang, T. Xiao, ArXiv (n.d.). date_created: 2024-02-08T15:34:43Z date_published: 2023-09-01T00:00:00Z date_updated: 2024-02-12T10:16:21Z day: '01' department: - _id: FrLo doi: 10.48550/arXiv.2309.00233 extern: '1' external_id: arxiv: - '2309.00233' fulldoi: https://doi.org/10.48550/arXiv.2309.00233 language: - iso: eng main_file_link: - open_access: '1' url: ' https://doi.org/10.48550/arXiv.2309.00233' month: '09' oa: 1 oa_version: Preprint publication: arXiv publication_status: submitted status: public title: Object-centric multiple object tracking type: preprint user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2023' ... --- OA_place: repository _id: '14965' abstract: - lang: eng text: 'A method of determining a correspondence between a first biological property of a cell and one or more further biological properties of cells is provided. The first biological property and the further biological properties are determined by different analysis techniques and each are contained in a respective one of a plurality of sets of biological properties. The method includes the steps of: converting the plurality of sets of biological properties into corresponding representations in a representation format which is invariant to the technologies used to derive the biological properties; determining, in said representation format, a representation from each of the converted sets of further biological properties which most closely matches the first representation of the first biological property; and re-converting the determined representations from the representation format back to the biological properties associated with the determined representations and thereby determining a correspondence between the first biological property and each of the further biological properties.' applicant: - ETH Zürich application_date: 2021-04-21 application_number: PCT/EP2021/060318 article_processing_charge: No author: - first_name: Joanna full_name: Ficek, Joanna last_name: Ficek - first_name: Kjong-Van full_name: Lehmann, Kjong-Van last_name: Lehmann - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: 'Gunnar ' full_name: 'Raetsch, Gunnar ' last_name: Raetsch - first_name: Stefan full_name: Stark, Stefan last_name: Stark citation: ama: Ficek J, Lehmann K-V, Locatello F, Raetsch G, Stark S. Methods of determining correspondences between biological properties of cells. 2023. apa: Ficek, J., Lehmann, K.-V., Locatello, F., Raetsch, G., & Stark, S. (2023). Methods of determining correspondences between biological properties of cells. chicago: Ficek, Joanna, Kjong-Van Lehmann, Francesco Locatello, Gunnar Raetsch, and Stefan Stark. “Methods of Determining Correspondences between Biological Properties of Cells,” 2023. ieee: J. Ficek, K.-V. Lehmann, F. Locatello, G. Raetsch, and S. Stark, “Methods of determining correspondences between biological properties of cells.” 2023. ista: Ficek J, Lehmann K-V, Locatello F, Raetsch G, Stark S. 2023. Methods of determining correspondences between biological properties of cells. mla: Ficek, Joanna, et al. Methods of Determining Correspondences between Biological Properties of Cells. 2023. short: J. Ficek, K.-V. Lehmann, F. Locatello, G. Raetsch, S. Stark, (2023). date_created: 2024-02-08T15:52:21Z date_published: 2023-05-25T00:00:00Z date_updated: 2025-01-29T10:53:48Z day: '25' ddc: - '540' department: - _id: FrLo extern: '1' file: - access_level: open_access checksum: 55ed444b176b48e4fb4d609ea895de36 content_type: application/pdf creator: ptazenko date_created: 2024-02-08T15:41:51Z date_updated: 2024-02-08T15:41:51Z file_id: '14966' file_name: Patent_FrLo_US20230162818A1.pdf file_size: 2893462 relation: main_file success: 1 file_date_updated: 2024-02-08T15:41:51Z has_accepted_license: '1' ipc: C12Q1/68 ; G06V10/82 ; G06V20/69 ; G16B40/30 ipn: US20230162818A1 month: '05' oa: 1 oa_version: Published Version page: '9' publication_date: 2023-05-25 status: public title: Methods of determining correspondences between biological properties of cells type: patent user_id: 8b945eb4-e2f2-11eb-945a-df72226e66a9 year: '2023' ... --- _id: '14974' abstract: - lang: eng text: "The field of machine learning and AI has witnessed remarkable breakthroughs with the emergence of LLMs, which have also sparked a lively debate in the causal community. As researchers in this field, we are interested in exploring how LLMs relate to causality research, and how we can leverage the technology to advance it. In the second conference of Causal Learning and Reasoning (CLeaR), 2023, we held a round table discussion to gather and integrate the diverse perspectives of the CLeaR community on this topic.\r\nThere is a general consensus that LLMs are not yet capable of causal reasoning at the current\r\nstage but has a lot of potential with public available information by CLeaR 2023. Enhancing causal machine learning is vital not only for its own sake but also to help LLMs improve their performance, especially regarding trustworthiness. In this document, we present both the summary and the raw outcome of the round table discussion. We acknowledge that with the progress of both fields, the opportunities and impact may rapidly change. We will repeat the same exercise in CLeaR 2024 to document the evolution." article_processing_charge: No author: - first_name: Cheng full_name: Zhang, Cheng last_name: Zhang - first_name: Dominik full_name: Janzing, Dominik last_name: Janzing - first_name: 'Mihaela ' full_name: 'van der Schaar, Mihaela ' last_name: van der Schaar - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Peter full_name: Spirtes, Peter last_name: Spirtes - first_name: Kun full_name: Zhang, Kun last_name: Zhang - first_name: Bernhard full_name: Schölkopf, Bernhard last_name: Schölkopf - first_name: Caroline full_name: Uhler, Caroline id: 49ADD78E-F248-11E8-B48F-1D18A9856A87 last_name: Uhler orcid: 0000-0002-7008-0216 citation: ama: 'Zhang C, Janzing D, van der Schaar M, et al. Causality in the time of LLMs: Round table discussion results of CLeaR 2023. In: 2nd Conference on Causal Learning and Reasoning.' apa: 'Zhang, C., Janzing, D., van der Schaar, M., Locatello, F., Spirtes, P., Zhang, K., … Uhler, C. (n.d.). Causality in the time of LLMs: Round table discussion results of CLeaR 2023. In 2nd Conference on Causal Learning and Reasoning. Tübingen, Germany.' chicago: 'Zhang, Cheng, Dominik Janzing, Mihaela van der Schaar, Francesco Locatello, Peter Spirtes, Kun Zhang, Bernhard Schölkopf, and Caroline Uhler. “Causality in the Time of LLMs: Round Table Discussion Results of CLeaR 2023.” In 2nd Conference on Causal Learning and Reasoning, n.d.' ieee: 'C. Zhang et al., “Causality in the time of LLMs: Round table discussion results of CLeaR 2023,” in 2nd Conference on Causal Learning and Reasoning, Tübingen, Germany.' ista: 'Zhang C, Janzing D, van der Schaar M, Locatello F, Spirtes P, Zhang K, Schölkopf B, Uhler C. Causality in the time of LLMs: Round table discussion results of CLeaR 2023. 2nd Conference on Causal Learning and Reasoning. CLeaR: Conference on Causal Learning and Reasoning.' mla: 'Zhang, Cheng, et al. “Causality in the Time of LLMs: Round Table Discussion Results of CLeaR 2023.” 2nd Conference on Causal Learning and Reasoning.' short: C. Zhang, D. Janzing, M. van der Schaar, F. Locatello, P. Spirtes, K. Zhang, B. Schölkopf, C. Uhler, in:, 2nd Conference on Causal Learning and Reasoning, n.d. conference: end_date: 2023-04-14 location: Tübingen, Germany name: 'CLeaR: Conference on Causal Learning and Reasoning' start_date: 2023-04-11 date_created: 2024-02-08T16:03:18Z date_published: 2023-05-01T00:00:00Z date_updated: 2025-08-05T11:19:37Z day: '01' ddc: - '000' department: - _id: FrLo extern: '1' file: - access_level: open_access checksum: 105ff58e55de866ce76967f3a95e82f7 content_type: application/pdf creator: ptazenko date_created: 2024-02-08T16:03:08Z date_updated: 2024-02-08T16:03:08Z file_id: '14975' file_name: CLeaR23_roundtable_discussion.pdf file_size: 215629 relation: main_file file_date_updated: 2024-02-08T16:03:08Z has_accepted_license: '1' language: - iso: eng month: '05' oa: 1 oa_version: Submitted Version publication: 2nd Conference on Causal Learning and Reasoning publication_status: submitted quality_controlled: '1' status: public title: 'Causality in the time of LLMs: Round table discussion results of CLeaR 2023' type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2023' ... --- OA_place: publisher OA_type: gold _id: '14210' abstract: - lang: eng text: Research on recovering the latent factors of variation of high dimensional data has so far focused on simple synthetic settings. Mostly building on unsupervised and weakly-supervised objectives, prior work missed out on the positive implications for representation learning on real world data. In this work, we propose to leverage knowledge extracted from a diversified set of supervised tasks to learn a common disentangled representation. Assuming that each supervised task only depends on an unknown subset of the factors of variation, we disentangle the feature space of a supervised multi-task model, with features activating sparsely across different tasks and information being shared as appropriate. Importantly, we never directly observe the factors of variations, but establish that access to multiple tasks is sufficient for identifiability under sufficiency and minimality assumptions. We validate our approach on six real world distribution shift benchmarks, and different data modalities (images, text), demonstrating how disentangled representations can be transferred to real settings. acknowledgement: "Marco Fumero and Emanuele Rodolà were supported by the ERC grant no.802554 (SPECGEO),\r\nPRIN 2020 project no.2020TA3K9N (LEGO.AI), and PNRR MUR project PE0000013-FAIR. Marco\r\nFumero and Francesco Locatello were partially at Amazon while working at this project. We thank\r\nJulius von Kügelgen, Sebastian Lachapelle and the anonymous reviewers for their feedback and\r\nsuggestions." alternative_title: - Advances in Neural Information Processing Systems article_number: '1204' article_processing_charge: No arxiv: 1 author: - first_name: Marco full_name: Fumero, Marco last_name: Fumero - first_name: Florian full_name: Wenzel, Florian last_name: Wenzel - first_name: Luca full_name: Zancato, Luca last_name: Zancato - first_name: Alessandro full_name: Achille, Alessandro last_name: Achille - first_name: Emanuele full_name: Rodolà, Emanuele last_name: Rodolà - first_name: Stefano full_name: Soatto, Stefano last_name: Soatto - first_name: Bernhard full_name: Schölkopf, Bernhard last_name: Schölkopf - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 citation: ama: 'Fumero M, Wenzel F, Zancato L, et al. Leveraging sparse and shared feature activations for disentangled representation learning. In: 37th International Conference on Neural Information Processing Systems. Vol 36. Neural Information Processing Systems Foundation; 2023.' apa: 'Fumero, M., Wenzel, F., Zancato, L., Achille, A., Rodolà, E., Soatto, S., … Locatello, F. (2023). Leveraging sparse and shared feature activations for disentangled representation learning. In 37th International Conference on Neural Information Processing Systems (Vol. 36). New Orleans, LA, United States: Neural Information Processing Systems Foundation.' chicago: Fumero, Marco, Florian Wenzel, Luca Zancato, Alessandro Achille, Emanuele Rodolà, Stefano Soatto, Bernhard Schölkopf, and Francesco Locatello. “Leveraging Sparse and Shared Feature Activations for Disentangled Representation Learning.” In 37th International Conference on Neural Information Processing Systems, Vol. 36. Neural Information Processing Systems Foundation, 2023. ieee: M. Fumero et al., “Leveraging sparse and shared feature activations for disentangled representation learning,” in 37th International Conference on Neural Information Processing Systems, New Orleans, LA, United States, 2023, vol. 36. ista: 'Fumero M, Wenzel F, Zancato L, Achille A, Rodolà E, Soatto S, Schölkopf B, Locatello F. 2023. Leveraging sparse and shared feature activations for disentangled representation learning. 37th International Conference on Neural Information Processing Systems. NeurIPS: Neural Information Processing Systems, Advances in Neural Information Processing Systems, vol. 36, 1204.' mla: Fumero, Marco, et al. “Leveraging Sparse and Shared Feature Activations for Disentangled Representation Learning.” 37th International Conference on Neural Information Processing Systems, vol. 36, 1204, Neural Information Processing Systems Foundation, 2023. short: M. Fumero, F. Wenzel, L. Zancato, A. Achille, E. Rodolà, S. Soatto, B. Schölkopf, F. Locatello, in:, 37th International Conference on Neural Information Processing Systems, Neural Information Processing Systems Foundation, 2023. conference: end_date: 2023-12-16 location: New Orleans, LA, United States name: 'NeurIPS: Neural Information Processing Systems' start_date: 2023-12-10 corr_author: '1' das_tickbox: '0' date_created: 2023-08-22T14:19:03Z date_published: 2023-12-10T00:00:00Z date_updated: 2026-07-29T08:02:05Z day: '10' ddc: - '000' department: - _id: FrLo external_id: arxiv: - '2304.07939' file: - access_level: open_access checksum: c70e8a74e205528750c71996ba39c6c1 content_type: application/pdf creator: dernst date_created: 2026-07-29T07:58:19Z date_updated: 2026-07-29T07:58:19Z file_id: '22604' file_name: 2026_Neurips_Fumero.pdf file_size: 1466102 relation: main_file success: 1 file_date_updated: 2026-07-29T07:58:19Z has_accepted_license: '1' intvolume: ' 36' language: - iso: eng month: '12' oa: 1 oa_version: Published Version publication: 37th International Conference on Neural Information Processing Systems publication_identifier: issn: - 1049-5258 publication_status: published publisher: Neural Information Processing Systems Foundation quality_controlled: '1' researchdata_availability: no status: public supplementarymaterial: no title: Leveraging sparse and shared feature activations for disentangled representation learning type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 volume: 36 year: '2023' ... --- OA_place: publisher OA_type: gold _id: '14952' abstract: - lang: eng text: "While different neural models often exhibit latent spaces that are alike when exposed to semantically related data, this intrinsic similarity is not always immediately discernible. Towards a better understanding of this phenomenon, our work shows how representations learned from these neural modules can be translated between different pre-trained networks via simpler transformations than previously thought. An advantage of this approach is the ability to\r\nestimate these transformations using standard, well-understood algebraic procedures that have closed-form solutions. Our method directly estimates a transformation between two given latent spaces, thereby enabling effective stitching of encoders and decoders without additional training. We extensively validate the adaptability of this translation procedure in different\r\nexperimental settings: across various trainings, domains, architectures (e.g., ResNet, CNN, ViT), and in multiple downstream tasks (classification, reconstruction). Notably, we show how it is possible to zero-shot stitch text encoders and vision decoders, or vice-versa, yielding surprisingly good classification performance in this multimodal setting." acknowledgement: "This work is supported by the ERC grant no.802554 (SPECGEO), PRIN 2020 project no.2020TA3K9N (LEGO.AI), and PNRR MUR project PE0000013-FAIR. Francesco\r\nLocatello did not contribute to this work at Amazon." alternative_title: - Advances in Neural Information Processing Systems article_processing_charge: No arxiv: 1 author: - first_name: Valentino full_name: Maiorca, Valentino last_name: Maiorca - first_name: Luca full_name: Moschella, Luca last_name: Moschella - first_name: Antonio full_name: Norelli, Antonio last_name: Norelli - first_name: Marco full_name: Fumero, Marco last_name: Fumero - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Emanuele full_name: Rodolà, Emanuele last_name: Rodolà citation: ama: 'Maiorca V, Moschella L, Norelli A, Fumero M, Locatello F, Rodolà E. Latent space translation via semantic alignment. In: 37th Conference on Neural Information Processing Systems. Neural Information Processing Systems Foundation; 2023:55394-55414. doi:10.52202/075280-2418' apa: 'Maiorca, V., Moschella, L., Norelli, A., Fumero, M., Locatello, F., & Rodolà, E. (2023). Latent space translation via semantic alignment. In 37th Conference on Neural Information Processing Systems (pp. 55394–55414). New Orleans, LA, United States: Neural Information Processing Systems Foundation. https://doi.org/10.52202/075280-2418' chicago: Maiorca, Valentino, Luca Moschella, Antonio Norelli, Marco Fumero, Francesco Locatello, and Emanuele Rodolà. “Latent Space Translation via Semantic Alignment.” In 37th Conference on Neural Information Processing Systems, 55394–414. Neural Information Processing Systems Foundation, 2023. https://doi.org/10.52202/075280-2418. ieee: V. Maiorca, L. Moschella, A. Norelli, M. Fumero, F. Locatello, and E. Rodolà, “Latent space translation via semantic alignment,” in 37th Conference on Neural Information Processing Systems, New Orleans, LA, United States, 2023, pp. 55394–55414. ista: 'Maiorca V, Moschella L, Norelli A, Fumero M, Locatello F, Rodolà E. 2023. Latent space translation via semantic alignment. 37th Conference on Neural Information Processing Systems. NeurIPS: Neural Information Processing Systems, Advances in Neural Information Processing Systems, , 55394–55414.' mla: Maiorca, Valentino, et al. “Latent Space Translation via Semantic Alignment.” 37th Conference on Neural Information Processing Systems, Neural Information Processing Systems Foundation, 2023, pp. 55394–414, doi:10.52202/075280-2418. short: V. Maiorca, L. Moschella, A. Norelli, M. Fumero, F. Locatello, E. Rodolà, in:, 37th Conference on Neural Information Processing Systems, Neural Information Processing Systems Foundation, 2023, pp. 55394–55414. conference: end_date: 2026-12-16 location: New Orleans, LA, United States name: 'NeurIPS: Neural Information Processing Systems' start_date: 2023-12-10 das_tickbox: '0' date_created: 2024-02-07T15:08:55Z date_published: 2023-12-01T00:00:00Z date_updated: 2026-07-29T08:24:56Z day: '01' ddc: - '000' department: - _id: FrLo doi: 10.52202/075280-2418 external_id: arxiv: - '2311.00664' file: - access_level: open_access checksum: 44ec045f30f8e25ce1c8ee70feffc9b4 content_type: application/pdf creator: dernst date_created: 2026-07-29T08:23:16Z date_updated: 2026-07-29T08:23:16Z file_id: '22606' file_name: 2023_Neurips_Malorca.pdf file_size: 4858561 relation: main_file success: 1 file_date_updated: 2026-07-29T08:23:16Z fulldoi: https://doi.org/10.52202/075280-2418 has_accepted_license: '1' language: - iso: eng month: '12' oa: 1 oa_version: Published Version page: 55394-55414 publication: 37th Conference on Neural Information Processing Systems publication_status: published publisher: Neural Information Processing Systems Foundation quality_controlled: '1' researchdata_availability: no status: public supplementarymaterial: yes title: Latent space translation via semantic alignment type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2023' ... --- OA_place: publisher OA_type: gold _id: '14207' abstract: - lang: eng text: The binding problem in human cognition, concerning how the brain represents and connects objects within a fixed network of neural connections, remains a subject of intense debate. Most machine learning efforts addressing this issue in an unsupervised setting have focused on slot-based methods, which may be limiting due to their discrete nature and difficulty to express uncertainty. Recently, the Complex AutoEncoder was proposed as an alternative that learns continuous and distributed object-centric representations. However, it is only applicable to simple toy data. In this paper, we present Rotating Features, a generalization of complex-valued features to higher dimensions, and a new evaluation procedure for extracting objects from distributed representations. Additionally, we show the applicability of our approach to pre-trained features. Together, these advancements enable us to scale distributed object-centric representations from simple toy to real-world data. We believe this work advances a new paradigm for addressing the binding problem in machine learning and has the potential to inspire further innovation in the field. acknowledgement: "We thank Jascha Sohl-Dickstein and Sjoerd van Steenkiste for insightful discussions. Francesco\r\nLocatello worked on this paper outside of Amazon. Sindy Lowe was supported by the Google PhD ¨Fellowship.\r\n" alternative_title: - Advances in Neural Information Processing Systems article_processing_charge: No arxiv: 1 author: - first_name: Sindy full_name: Löwe, Sindy last_name: Löwe - first_name: Phillip full_name: Lippe, Phillip last_name: Lippe - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Max full_name: Welling, Max last_name: Welling citation: ama: 'Löwe S, Lippe P, Locatello F, Welling M. Rotating features for object discovery. In: 37th Conference on Neural Information Processing Systems. Vol 36. Neural Information Processing Systems Foundation; 2023.' apa: 'Löwe, S., Lippe, P., Locatello, F., & Welling, M. (2023). Rotating features for object discovery. In 37th Conference on Neural Information Processing Systems (Vol. 36). New Orleans, LA, United States: Neural Information Processing Systems Foundation.' chicago: Löwe, Sindy, Phillip Lippe, Francesco Locatello, and Max Welling. “Rotating Features for Object Discovery.” In 37th Conference on Neural Information Processing Systems, Vol. 36. Neural Information Processing Systems Foundation, 2023. ieee: S. Löwe, P. Lippe, F. Locatello, and M. Welling, “Rotating features for object discovery,” in 37th Conference on Neural Information Processing Systems, New Orleans, LA, United States, 2023, vol. 36. ista: 'Löwe S, Lippe P, Locatello F, Welling M. 2023. Rotating features for object discovery. 37th Conference on Neural Information Processing Systems. NeurIPS: Neural Information Processing Systems, Advances in Neural Information Processing Systems, vol. 36.' mla: Löwe, Sindy, et al. “Rotating Features for Object Discovery.” 37th Conference on Neural Information Processing Systems, vol. 36, Neural Information Processing Systems Foundation, 2023. short: S. Löwe, P. Lippe, F. Locatello, M. Welling, in:, 37th Conference on Neural Information Processing Systems, Neural Information Processing Systems Foundation, 2023. conference: end_date: 2023-12-16 location: New Orleans, LA, United States name: 'NeurIPS: Neural Information Processing Systems' start_date: 2023-12-10 das_tickbox: '0' date_created: 2023-08-22T14:18:00Z date_published: 2023-09-21T00:00:00Z date_updated: 2026-07-29T07:36:23Z day: '21' ddc: - '000' department: - _id: FrLo external_id: arxiv: - '2306.00600' file: - access_level: open_access checksum: 67860dc8ce60f60cfa366baa623f6a55 content_type: application/pdf creator: dernst date_created: 2026-07-29T07:35:33Z date_updated: 2026-07-29T07:35:33Z file_id: '22603' file_name: 2026_Neurips_Loewe.pdf file_size: 10788023 relation: main_file success: 1 file_date_updated: 2026-07-29T07:35:33Z has_accepted_license: '1' intvolume: ' 36' language: - iso: eng month: '09' oa: 1 oa_version: Published Version publication: 37th Conference on Neural Information Processing Systems publication_identifier: issn: - 1049-5258 publication_status: published publisher: Neural Information Processing Systems Foundation quality_controlled: '1' researchdata_availability: no scopus_import: '1' status: public supplementarymaterial: yes title: Rotating features for object discovery type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 volume: 36 year: '2023' ... --- OA_place: publisher OA_type: gold _id: '14948' abstract: - lang: eng text: "The extraction of modular object-centric representations for downstream tasks\r\nis an emerging area of research. Learning grounded representations of objects\r\nthat are guaranteed to be stable and invariant promises robust performance\r\nacross different tasks and environments. Slot Attention (SA) learns\r\nobject-centric representations by assigning objects to \\textit{slots}, but\r\npresupposes a \\textit{single} distribution from which all slots are randomly\r\ninitialised. This results in an inability to learn \\textit{specialized} slots\r\nwhich bind to specific object types and remain invariant to identity-preserving\r\nchanges in object appearance. To address this, we present\r\n\\emph{\\textsc{Co}nditional \\textsc{S}lot \\textsc{A}ttention} (\\textsc{CoSA})\r\nusing a novel concept of \\emph{Grounded Slot Dictionary} (GSD) inspired by\r\nvector quantization. Our proposed GSD comprises (i) canonical object-level\r\nproperty vectors and (ii) parametric Gaussian distributions, which define a\r\nprior over the slots. We demonstrate the benefits of our method in multiple\r\ndownstream tasks such as scene generation, composition, and task adaptation,\r\nwhilst remaining competitive with SA in popular object discovery benchmarks." acknowledgement: "A. Kori is supported by UKRI (grant agreement no. EP/S023356/1), as part of the UKRI Centre\r\nfor Doctoral Training in Safe and Trusted AI. F.D.S. Ribeiro and B. Glocker received funding from\r\nthe European Research Council (ERC) under the European Union’s Horizon 2020 research and\r\ninnovation programme (grant agreement No 757173, project MIRA, ERC-2017-STG)." article_processing_charge: No arxiv: 1 author: - first_name: Avinash full_name: Kori, Avinash last_name: Kori - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Fabio De Sousa full_name: Ribeiro, Fabio De Sousa last_name: Ribeiro - first_name: Francesca full_name: Toni, Francesca last_name: Toni - first_name: Ben full_name: Glocker, Ben last_name: Glocker citation: ama: 'Kori A, Locatello F, Ribeiro FDS, Toni F, Glocker B. Grounded object centric learning. In: 12th International Conference on Learning Representations. ICLR; 2023.' apa: 'Kori, A., Locatello, F., Ribeiro, F. D. S., Toni, F., & Glocker, B. (2023). Grounded object centric learning. In 12th International Conference on Learning Representations. Vienna, Austria: ICLR.' chicago: Kori, Avinash, Francesco Locatello, Fabio De Sousa Ribeiro, Francesca Toni, and Ben Glocker. “Grounded Object Centric Learning.” In 12th International Conference on Learning Representations. ICLR, 2023. ieee: A. Kori, F. Locatello, F. D. S. Ribeiro, F. Toni, and B. Glocker, “Grounded object centric learning,” in 12th International Conference on Learning Representations, Vienna, Austria, 2023. ista: 'Kori A, Locatello F, Ribeiro FDS, Toni F, Glocker B. 2023. Grounded object centric learning. 12th International Conference on Learning Representations. ICLR: International Conference on Learning Representations.' mla: Kori, Avinash, et al. “Grounded Object Centric Learning.” 12th International Conference on Learning Representations, ICLR, 2023. short: A. Kori, F. Locatello, F.D.S. Ribeiro, F. Toni, B. Glocker, in:, 12th International Conference on Learning Representations, ICLR, 2023. conference: end_date: 2024-05-11 location: Vienna, Austria name: 'ICLR: International Conference on Learning Representations' start_date: 2024-05-07 das_tickbox: '0' date_created: 2024-02-07T14:47:04Z date_published: 2023-07-18T00:00:00Z date_updated: 2026-07-29T08:17:51Z day: '18' ddc: - '000' department: - _id: FrLo external_id: arxiv: - '2307.09437' file: - access_level: open_access checksum: b9687ba09b6ecfdfb319fe167ed07816 content_type: application/pdf creator: dernst date_created: 2026-07-29T08:13:30Z date_updated: 2026-07-29T08:13:30Z file_id: '22605' file_name: 2024_ICLR_Kori.pdf file_size: 19986955 relation: main_file success: 1 file_date_updated: 2026-07-29T08:13:30Z has_accepted_license: '1' language: - iso: eng month: '07' oa: 1 oa_version: Published Version publication: 12th International Conference on Learning Representations publication_status: published publisher: ICLR quality_controlled: '1' researchdata_availability: no status: public supplementarymaterial: yes title: Grounded object centric learning type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2023' ... --- OA_place: publisher OA_type: gold _id: '14954' abstract: - lang: eng text: "When domain knowledge is limited and experimentation is restricted by ethical,\r\nfinancial, or time constraints, practitioners turn to observational causal discovery\r\nmethods to recover the causal structure, exploiting the statistical properties of their\r\ndata. Because causal discovery without further assumptions is an ill-posed problem,\r\neach algorithm comes with its own set of usually untestable assumptions, some\r\nof which are hard to meet in real datasets. Motivated by these considerations, this\r\npaper extensively benchmarks the empirical performance of recent causal discovery\r\nmethods on observational iid data generated under different background conditions,\r\nallowing for violations of the critical assumptions required by each selected approach. Our experimental findings show that score matching-based methods demonstrate surprising performance in the false positive and false negative rate of the\r\ninferred graph in these challenging scenarios, and we provide theoretical insights\r\ninto their performance. This work is also the first effort to benchmark the stability of\r\ncausal discovery algorithms with respect to the values of their hyperparameters. Finally, we hope this paper will set a new standard for the evaluation of causal discovery methods and can serve as an accessible entry point for practitioners interested\r\nin the field, highlighting the empirical implications of different algorithm choices." acknowledgement: "We thank Kun Zhang and Carl-Johann Simon-Gabriel for the insightful discussions. This work\r\nhas been supported by AFOSR, grant n. FA8655-20-1-7035. FM is supported by Programma\r\nOperativo Nazionale ricerca e innovazione 2014-2020. FM partially contributed to this work during an internship at Amazon Web Services with FL. FL partially contributed while at AWS." alternative_title: - Advances in Neural Information Processing Systems article_processing_charge: No arxiv: 1 author: - first_name: Francesco full_name: Montagna, Francesco last_name: Montagna - first_name: Atalanti A. full_name: Mastakouri, Atalanti A. last_name: Mastakouri - first_name: Elias full_name: Eulig, Elias last_name: Eulig - first_name: Nicoletta full_name: Noceti, Nicoletta last_name: Noceti - first_name: Lorenzo full_name: Rosasco, Lorenzo last_name: Rosasco - first_name: Dominik full_name: Janzing, Dominik last_name: Janzing - first_name: Bryon full_name: Aragam, Bryon last_name: Aragam - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 citation: ama: 'Montagna F, Mastakouri AA, Eulig E, et al. Assumption violations in causal discovery and the robustness of score matching. In: 37th Conference on Neural Information Processing Systems. Vol 36. Neural Information Processing Systems Foundation; 2023. doi:10.52202/075280-2050' apa: 'Montagna, F., Mastakouri, A. A., Eulig, E., Noceti, N., Rosasco, L., Janzing, D., … Locatello, F. (2023). Assumption violations in causal discovery and the robustness of score matching. In 37th Conference on Neural Information Processing Systems (Vol. 36). New Orleans, LA, United States: Neural Information Processing Systems Foundation. https://doi.org/10.52202/075280-2050' chicago: Montagna, Francesco, Atalanti A. Mastakouri, Elias Eulig, Nicoletta Noceti, Lorenzo Rosasco, Dominik Janzing, Bryon Aragam, and Francesco Locatello. “Assumption Violations in Causal Discovery and the Robustness of Score Matching.” In 37th Conference on Neural Information Processing Systems, Vol. 36. Neural Information Processing Systems Foundation, 2023. https://doi.org/10.52202/075280-2050. ieee: F. Montagna et al., “Assumption violations in causal discovery and the robustness of score matching,” in 37th Conference on Neural Information Processing Systems, New Orleans, LA, United States, 2023, vol. 36. ista: 'Montagna F, Mastakouri AA, Eulig E, Noceti N, Rosasco L, Janzing D, Aragam B, Locatello F. 2023. Assumption violations in causal discovery and the robustness of score matching. 37th Conference on Neural Information Processing Systems. NeurIPS: Neural Information Processing Systems, Advances in Neural Information Processing Systems, vol. 36.' mla: Montagna, Francesco, et al. “Assumption Violations in Causal Discovery and the Robustness of Score Matching.” 37th Conference on Neural Information Processing Systems, vol. 36, Neural Information Processing Systems Foundation, 2023, doi:10.52202/075280-2050. short: F. Montagna, A.A. Mastakouri, E. Eulig, N. Noceti, L. Rosasco, D. Janzing, B. Aragam, F. Locatello, in:, 37th Conference on Neural Information Processing Systems, Neural Information Processing Systems Foundation, 2023. conference: end_date: 2023-12-16 location: New Orleans, LA, United States name: 'NeurIPS: Neural Information Processing Systems' start_date: 2023-12-12 das_tickbox: '0' date_created: 2024-02-07T15:11:56Z date_published: 2023-12-20T00:00:00Z date_updated: 2026-09-17T12:59:55Z day: '20' ddc: - '000' department: - _id: FrLo doi: 10.52202/075280-2050 external_id: arxiv: - '2310.13387' file: - access_level: open_access checksum: 71d38402b4edef4c084b39f3e2f04526 content_type: application/pdf creator: dernst date_created: 2026-08-13T07:49:38Z date_updated: 2026-08-13T07:49:38Z file_id: '22701' file_name: 2023_Neurips_Montagna.pdf file_size: 7640984 relation: main_file success: 1 file_date_updated: 2026-08-13T07:49:38Z fulldoi: https://doi.org/10.52202/075280-2050 has_accepted_license: '1' intvolume: ' 36' language: - iso: eng month: '12' oa: 1 oa_version: Published Version publication: 37th Conference on Neural Information Processing Systems publication_identifier: eissn: - 1049-5258 publication_status: published publisher: Neural Information Processing Systems Foundation quality_controlled: '1' researchdata_availability: no status: public supplementarymaterial: yes title: Assumption violations in causal discovery and the robustness of score matching type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 volume: 36 year: '2023' ... --- OA_place: publisher OA_type: gold _id: '14953' abstract: - lang: eng text: "This paper provides statistical sample complexity bounds for score-matching and\r\nits applications in causal discovery. We demonstrate that accurate estimation of the\r\nscore function is achievable by training a standard deep ReLU neural network using\r\nstochastic gradient descent. We establish bounds on the error rate of recovering\r\ncausal relationships using the score-matching-based causal discovery method of\r\nRolland et al. [2022], assuming a sufficiently good estimation of the score function.\r\nFinally, we analyze the upper bound of score-matching estimation within the scorebased generative modeling, which has been applied for causal discovery but is also\r\nof independent interest within the domain of generative models.y" acknowledgement: "We are thankful to the reviewers for providing constructive feedback and Kun Zhang and Dominik\r\nJanzing for helpful discussion on the special case of deterministic children. This work was supported\r\nby Hasler Foundation Program: Hasler Responsible AI (project number 21043). This work was\r\nsupported by the Swiss National Science Foundation (SNSF) under grant number 200021_205011.\r\nFrancesco Locatello did not contribute to this work at Amazon. " alternative_title: - Advances in Neural Information Processing Systems article_processing_charge: No arxiv: 1 author: - first_name: Zhenyu full_name: Zhu, Zhenyu last_name: Zhu - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Volkan full_name: Cevher, Volkan last_name: Cevher citation: ama: 'Zhu Z, Locatello F, Cevher V. Sample complexity bounds for score-matching: Causal discovery and generative modeling. In: 37th Conference on Neural Information Processing Systems. Vol 36. Neural Information Processing Systems Foundation; 2023:3325-3337. doi:10.52202/075280-0147' apa: 'Zhu, Z., Locatello, F., & Cevher, V. (2023). Sample complexity bounds for score-matching: Causal discovery and generative modeling. In 37th Conference on Neural Information Processing Systems (Vol. 36, pp. 3325–3337). New Orleans, LA, United States: Neural Information Processing Systems Foundation. https://doi.org/10.52202/075280-0147' chicago: 'Zhu, Zhenyu, Francesco Locatello, and Volkan Cevher. “Sample Complexity Bounds for Score-Matching: Causal Discovery and Generative Modeling.” In 37th Conference on Neural Information Processing Systems, 36:3325–37. Neural Information Processing Systems Foundation, 2023. https://doi.org/10.52202/075280-0147.' ieee: 'Z. Zhu, F. Locatello, and V. Cevher, “Sample complexity bounds for score-matching: Causal discovery and generative modeling,” in 37th Conference on Neural Information Processing Systems, New Orleans, LA, United States, 2023, vol. 36, pp. 3325–3337.' ista: 'Zhu Z, Locatello F, Cevher V. 2023. Sample complexity bounds for score-matching: Causal discovery and generative modeling. 37th Conference on Neural Information Processing Systems. NeurIPS: Neural Information Processing Systems, Advances in Neural Information Processing Systems, vol. 36, 3325–3337.' mla: 'Zhu, Zhenyu, et al. “Sample Complexity Bounds for Score-Matching: Causal Discovery and Generative Modeling.” 37th Conference on Neural Information Processing Systems, vol. 36, Neural Information Processing Systems Foundation, 2023, pp. 3325–37, doi:10.52202/075280-0147.' short: Z. Zhu, F. Locatello, V. Cevher, in:, 37th Conference on Neural Information Processing Systems, Neural Information Processing Systems Foundation, 2023, pp. 3325–3337. conference: end_date: 2023-12-16 location: New Orleans, LA, United States name: 'NeurIPS: Neural Information Processing Systems' start_date: 2023-12-12 corr_author: '1' das_tickbox: '0' date_created: 2024-02-07T15:11:11Z date_published: 2023-10-27T00:00:00Z date_updated: 2026-09-17T12:59:35Z day: '27' ddc: - '000' department: - _id: FrLo doi: 10.52202/075280-0147 external_id: arxiv: - '2310.18123' file: - access_level: open_access checksum: 6f10adadefceb3bb50ad0b2637493381 content_type: application/pdf creator: dernst date_created: 2026-08-13T07:02:51Z date_updated: 2026-08-13T07:02:51Z file_id: '22700' file_name: 2023_Neurips_Zhu.pdf file_size: 305362 relation: main_file success: 1 file_date_updated: 2026-08-13T07:02:51Z fulldoi: https://doi.org/10.52202/075280-0147 has_accepted_license: '1' intvolume: ' 36' language: - iso: eng month: '10' oa: 1 oa_version: Published Version page: 3325-3337 publication: 37th Conference on Neural Information Processing Systems publication_identifier: eissn: - 1049-5258 isbn: - '9781713899921' publication_status: published publisher: Neural Information Processing Systems Foundation quality_controlled: '1' researchdata_availability: no scopus_import: '1' status: public supplementarymaterial: yes title: 'Sample complexity bounds for score-matching: Causal discovery and generative modeling' type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 volume: 36 year: '2023' ... --- _id: '14093' abstract: - lang: eng text: ' We propose a stochastic conditional gradient method (CGM) for minimizing convex finite-sum objectives formed as a sum of smooth and non-smooth terms. Existing CGM variants for this template either suffer from slow convergence rates, or require carefully increasing the batch size over the course of the algorithm’s execution, which leads to computing full gradients. In contrast, the proposed method, equipped with a stochastic average gradient (SAG) estimator, requires only one sample per iteration. Nevertheless, it guarantees fast convergence rates on par with more sophisticated variance reduction techniques. In applications we put special emphasis on problems with a large number of separable constraints. Such problems are prevalent among semidefinite programming (SDP) formulations arising in machine learning and theoretical computer science. We provide numerical experiments on matrix completion, unsupervised clustering, and sparsest-cut SDPs. ' alternative_title: - PMLR article_processing_charge: No arxiv: 1 author: - first_name: Gideon full_name: Dresdner, Gideon last_name: Dresdner - first_name: Maria-Luiza full_name: Vladarean, Maria-Luiza last_name: Vladarean - first_name: Gunnar full_name: Rätsch, Gunnar last_name: Rätsch - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Volkan full_name: Cevher, Volkan last_name: Cevher - first_name: Alp full_name: Yurtsever, Alp last_name: Yurtsever citation: ama: 'Dresdner G, Vladarean M-L, Rätsch G, Locatello F, Cevher V, Yurtsever A. Faster one-sample stochastic conditional gradient method for composite convex minimization. In: Proceedings of the 25th International Conference on Artificial Intelligence and Statistics. Vol 151. ML Research Press; 2022:8439-8457.' apa: 'Dresdner, G., Vladarean, M.-L., Rätsch, G., Locatello, F., Cevher, V., & Yurtsever, A. (2022). Faster one-sample stochastic conditional gradient method for composite convex minimization. In Proceedings of the 25th International Conference on Artificial Intelligence and Statistics (Vol. 151, pp. 8439–8457). Virtual: ML Research Press.' chicago: Dresdner, Gideon, Maria-Luiza Vladarean, Gunnar Rätsch, Francesco Locatello, Volkan Cevher, and Alp Yurtsever. “ Faster One-Sample Stochastic Conditional Gradient Method for Composite Convex Minimization.” In Proceedings of the 25th International Conference on Artificial Intelligence and Statistics, 151:8439–57. ML Research Press, 2022. ieee: G. Dresdner, M.-L. Vladarean, G. Rätsch, F. Locatello, V. Cevher, and A. Yurtsever, “ Faster one-sample stochastic conditional gradient method for composite convex minimization,” in Proceedings of the 25th International Conference on Artificial Intelligence and Statistics, Virtual, 2022, vol. 151, pp. 8439–8457. ista: 'Dresdner G, Vladarean M-L, Rätsch G, Locatello F, Cevher V, Yurtsever A. 2022. Faster one-sample stochastic conditional gradient method for composite convex minimization. Proceedings of the 25th International Conference on Artificial Intelligence and Statistics. AISTATS: Conference on Artificial Intelligence and Statistics, PMLR, vol. 151, 8439–8457.' mla: Dresdner, Gideon, et al. “ Faster One-Sample Stochastic Conditional Gradient Method for Composite Convex Minimization.” Proceedings of the 25th International Conference on Artificial Intelligence and Statistics, vol. 151, ML Research Press, 2022, pp. 8439–57. short: G. Dresdner, M.-L. Vladarean, G. Rätsch, F. Locatello, V. Cevher, A. Yurtsever, in:, Proceedings of the 25th International Conference on Artificial Intelligence and Statistics, ML Research Press, 2022, pp. 8439–8457. conference: end_date: 2022-03-30 location: Virtual name: 'AISTATS: Conference on Artificial Intelligence and Statistics' start_date: 2022-03-28 date_created: 2023-08-21T09:27:43Z date_published: 2022-04-01T00:00:00Z date_updated: 2023-09-06T10:28:17Z day: '01' department: - _id: FrLo extern: '1' external_id: arxiv: - '2202.13212' intvolume: ' 151' language: - iso: eng main_file_link: - open_access: '1' url: https://arxiv.org/abs/2202.13212 month: '04' oa: 1 oa_version: Preprint page: 8439-8457 publication: Proceedings of the 25th International Conference on Artificial Intelligence and Statistics publication_identifier: issn: - 2640-3498 publication_status: published publisher: ML Research Press quality_controlled: '1' scopus_import: '1' status: public title: ' Faster one-sample stochastic conditional gradient method for composite convex minimization' type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 volume: 151 year: '2022' ... --- _id: '14106' abstract: - lang: eng text: "We show that deep networks trained to satisfy demographic parity often do so\r\nthrough a form of race or gender awareness, and that the more we force a network\r\nto be fair, the more accurately we can recover race or gender from the internal state\r\nof the network. Based on this observation, we investigate an alternative fairness\r\napproach: we add a second classification head to the network to explicitly predict\r\nthe protected attribute (such as race or gender) alongside the original task. After\r\ntraining the two-headed network, we enforce demographic parity by merging the\r\ntwo heads, creating a network with the same architecture as the original network.\r\nWe establish a close relationship between existing approaches and our approach\r\nby showing (1) that the decisions of a fair classifier are well-approximated by our\r\napproach, and (2) that an unfair and optimally accurate classifier can be recovered\r\nfrom a fair classifier and our second head predicting the protected attribute. We use\r\nour explicit formulation to argue that the existing fairness approaches, just as ours,\r\ndemonstrate disparate treatment and that they are likely to be unlawful in a wide\r\nrange of scenarios under US law." alternative_title: - Advances in Neural Information Processing Systems article_processing_charge: No arxiv: 1 author: - first_name: Michael full_name: Lohaus, Michael last_name: Lohaus - first_name: Matthäus full_name: Kleindessner, Matthäus last_name: Kleindessner - first_name: Krishnaram full_name: Kenthapadi, Krishnaram last_name: Kenthapadi - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Chris full_name: Russell, Chris last_name: Russell citation: ama: 'Lohaus M, Kleindessner M, Kenthapadi K, Locatello F, Russell C. Are two heads the same as one? Identifying disparate treatment in fair neural networks. In: 36th Conference on Neural Information Processing Systems. Vol 35. Neural Information Processing Systems Foundation; 2022:16548-16562.' apa: 'Lohaus, M., Kleindessner, M., Kenthapadi, K., Locatello, F., & Russell, C. (2022). Are two heads the same as one? Identifying disparate treatment in fair neural networks. In 36th Conference on Neural Information Processing Systems (Vol. 35, pp. 16548–16562). New Orleans, LA, United States: Neural Information Processing Systems Foundation.' chicago: Lohaus, Michael, Matthäus Kleindessner, Krishnaram Kenthapadi, Francesco Locatello, and Chris Russell. “Are Two Heads the Same as One? Identifying Disparate Treatment in Fair Neural Networks.” In 36th Conference on Neural Information Processing Systems, 35:16548–62. Neural Information Processing Systems Foundation, 2022. ieee: M. Lohaus, M. Kleindessner, K. Kenthapadi, F. Locatello, and C. Russell, “Are two heads the same as one? Identifying disparate treatment in fair neural networks,” in 36th Conference on Neural Information Processing Systems, New Orleans, LA, United States, 2022, vol. 35, pp. 16548–16562. ista: 'Lohaus M, Kleindessner M, Kenthapadi K, Locatello F, Russell C. 2022. Are two heads the same as one? Identifying disparate treatment in fair neural networks. 36th Conference on Neural Information Processing Systems. NeurIPS: Neural Information Processing Systems, Advances in Neural Information Processing Systems, vol. 35, 16548–16562.' mla: Lohaus, Michael, et al. “Are Two Heads the Same as One? Identifying Disparate Treatment in Fair Neural Networks.” 36th Conference on Neural Information Processing Systems, vol. 35, Neural Information Processing Systems Foundation, 2022, pp. 16548–62. short: M. Lohaus, M. Kleindessner, K. Kenthapadi, F. Locatello, C. Russell, in:, 36th Conference on Neural Information Processing Systems, Neural Information Processing Systems Foundation, 2022, pp. 16548–16562. conference: end_date: 2022-12-09 location: New Orleans, LA, United States name: 'NeurIPS: Neural Information Processing Systems' start_date: 2022-11-28 date_created: 2023-08-21T12:12:42Z date_published: 2022-12-15T00:00:00Z date_updated: 2024-10-14T12:27:01Z day: '15' department: - _id: FrLo extern: '1' external_id: arxiv: - '2204.04440' intvolume: ' 35' language: - iso: eng main_file_link: - open_access: '1' url: https://arxiv.org/abs/2204.04440 month: '12' oa: 1 oa_version: Preprint page: 16548-16562 publication: 36th Conference on Neural Information Processing Systems publication_identifier: isbn: - '9781713871088' publication_status: published publisher: Neural Information Processing Systems Foundation quality_controlled: '1' scopus_import: '1' status: public title: Are two heads the same as one? Identifying disparate treatment in fair neural networks type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 volume: 35 year: '2022' ... --- _id: '14107' abstract: - lang: eng text: "Amodal perception requires inferring the full shape of an object that is partially occluded. This task is particularly challenging on two levels: (1) it requires more information than what is contained in the instant retina or imaging sensor, (2) it is difficult to obtain enough well-annotated amodal labels for supervision. To this end, this paper develops a new framework of\r\nSelf-supervised amodal Video object segmentation (SaVos). Our method efficiently leverages the visual information of video temporal sequences to infer the amodal mask of objects. The key intuition is that the occluded part of an object can be explained away if that part is visible in other frames, possibly deformed as long as the deformation can be reasonably learned.\r\nAccordingly, we derive a novel self-supervised learning paradigm that efficiently utilizes the visible object parts as the supervision to guide the training on videos. In addition to learning type prior to complete masks for known types, SaVos also learns the spatiotemporal prior, which is also useful for the amodal task and could generalize to unseen types. The proposed\r\nframework achieves the state-of-the-art performance on the synthetic amodal segmentation benchmark FISHBOWL and the real world benchmark KINS-Video-Car. Further, it lends itself well to being transferred to novel distributions using test-time adaptation, outperforming existing models even after the transfer to a new distribution." article_processing_charge: No arxiv: 1 author: - first_name: Jian full_name: Yao, Jian last_name: Yao - first_name: Yuxin full_name: Hong, Yuxin last_name: Hong - first_name: Chiyu full_name: Wang, Chiyu last_name: Wang - first_name: Tianjun full_name: Xiao, Tianjun last_name: Xiao - first_name: Tong full_name: He, Tong last_name: He - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: David full_name: Wipf, David last_name: Wipf - first_name: Yanwei full_name: Fu, Yanwei last_name: Fu - first_name: Zheng full_name: Zhang, Zheng last_name: Zhang citation: ama: 'Yao J, Hong Y, Wang C, et al. Self-supervised amodal video object segmentation. In: 36th Conference on Neural Information Processing Systems. ; 2022. doi:10.48550/arXiv.2210.12733' apa: Yao, J., Hong, Y., Wang, C., Xiao, T., He, T., Locatello, F., … Zhang, Z. (2022). Self-supervised amodal video object segmentation. In 36th Conference on Neural Information Processing Systems. New Orleans, LA, United States. https://doi.org/10.48550/arXiv.2210.12733 chicago: Yao, Jian, Yuxin Hong, Chiyu Wang, Tianjun Xiao, Tong He, Francesco Locatello, David Wipf, Yanwei Fu, and Zheng Zhang. “Self-Supervised Amodal Video Object Segmentation.” In 36th Conference on Neural Information Processing Systems, 2022. https://doi.org/10.48550/arXiv.2210.12733. ieee: J. Yao et al., “Self-supervised amodal video object segmentation,” in 36th Conference on Neural Information Processing Systems, New Orleans, LA, United States, 2022. ista: 'Yao J, Hong Y, Wang C, Xiao T, He T, Locatello F, Wipf D, Fu Y, Zhang Z. 2022. Self-supervised amodal video object segmentation. 36th Conference on Neural Information Processing Systems. NeurIPS: Neural Information Processing Systems.' mla: Yao, Jian, et al. “Self-Supervised Amodal Video Object Segmentation.” 36th Conference on Neural Information Processing Systems, 2022, doi:10.48550/arXiv.2210.12733. short: J. Yao, Y. Hong, C. Wang, T. Xiao, T. He, F. Locatello, D. Wipf, Y. Fu, Z. Zhang, in:, 36th Conference on Neural Information Processing Systems, 2022. conference: end_date: 2022-12-01 location: New Orleans, LA, United States name: 'NeurIPS: Neural Information Processing Systems' start_date: 2022-11-28 date_created: 2023-08-21T12:13:25Z date_published: 2022-10-23T00:00:00Z date_updated: 2023-09-11T09:34:17Z day: '23' department: - _id: FrLo doi: 10.48550/arXiv.2210.12733 extern: '1' external_id: arxiv: - '2210.12733' fulldoi: https://doi.org/10.48550/arXiv.2210.12733 language: - iso: eng main_file_link: - open_access: '1' url: https://doi.org/10.48550/arXiv.2210.12733 month: '10' oa: 1 oa_version: Preprint publication: 36th Conference on Neural Information Processing Systems publication_status: published status: public title: Self-supervised amodal video object segmentation type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2022' ... --- _id: '14114' abstract: - lang: eng text: Algorithmic fairness is frequently motivated in terms of a trade-off in which overall performance is decreased so as to improve performance on disadvantaged groups where the algorithm would otherwise be less accurate. Contrary to this, we find that applying existing fairness approaches to computer vision improve fairness by degrading the performance of classifiers across all groups (with increased degradation on the best performing groups). Extending the bias-variance decomposition for classification to fairness, we theoretically explain why the majority of fairness methods designed for low capacity models should not be used in settings involving high-capacity models, a scenario common to computer vision. We corroborate this analysis with extensive experimental support that shows that many of the fairness heuristics used in computer vision also degrade performance on the most disadvantaged groups. Building on these insights, we propose an adaptive augmentation strategy that, uniquely, of all methods tested, improves performance for the disadvantaged groups. article_processing_charge: No arxiv: 1 author: - first_name: Dominik full_name: Zietlow, Dominik last_name: Zietlow - first_name: Michael full_name: Lohaus, Michael last_name: Lohaus - first_name: Guha full_name: Balakrishnan, Guha last_name: Balakrishnan - first_name: Matthaus full_name: Kleindessner, Matthaus last_name: Kleindessner - first_name: Francesco full_name: Locatello, Francesco id: 26cfd52f-2483-11ee-8040-88983bcc06d4 last_name: Locatello orcid: 0000-0002-4850-0683 - first_name: Bernhard full_name: Scholkopf, Bernhard last_name: Scholkopf - first_name: Chris full_name: Russell, Chris last_name: Russell citation: ama: 'Zietlow D, Lohaus M, Balakrishnan G, et al. Leveling down in computer vision: Pareto inefficiencies in fair deep classifiers. In: 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Institute of Electrical and Electronics Engineers; 2022:10400-10411. doi:10.1109/cvpr52688.2022.01016' apa: 'Zietlow, D., Lohaus, M., Balakrishnan, G., Kleindessner, M., Locatello, F., Scholkopf, B., & Russell, C. (2022). Leveling down in computer vision: Pareto inefficiencies in fair deep classifiers. In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 10400–10411). New Orleans, LA, United States: Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/cvpr52688.2022.01016' chicago: 'Zietlow, Dominik, Michael Lohaus, Guha Balakrishnan, Matthaus Kleindessner, Francesco Locatello, Bernhard Scholkopf, and Chris Russell. “Leveling down in Computer Vision: Pareto Inefficiencies in Fair Deep Classifiers.” In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 10400–411. Institute of Electrical and Electronics Engineers, 2022. https://doi.org/10.1109/cvpr52688.2022.01016.' ieee: 'D. Zietlow et al., “Leveling down in computer vision: Pareto inefficiencies in fair deep classifiers,” in 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, New Orleans, LA, United States, 2022, pp. 10400–10411.' ista: 'Zietlow D, Lohaus M, Balakrishnan G, Kleindessner M, Locatello F, Scholkopf B, Russell C. 2022. Leveling down in computer vision: Pareto inefficiencies in fair deep classifiers. 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. CVPR: Conference on Computer Vision and Pattern Recognition, 10400–10411.' mla: 'Zietlow, Dominik, et al. “Leveling down in Computer Vision: Pareto Inefficiencies in Fair Deep Classifiers.” 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Institute of Electrical and Electronics Engineers, 2022, pp. 10400–11, doi:10.1109/cvpr52688.2022.01016.' short: D. Zietlow, M. Lohaus, G. Balakrishnan, M. Kleindessner, F. Locatello, B. Scholkopf, C. Russell, in:, 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Institute of Electrical and Electronics Engineers, 2022, pp. 10400–10411. conference: end_date: 2022-06-24 location: New Orleans, LA, United States name: 'CVPR: Conference on Computer Vision and Pattern Recognition' start_date: 2022-06-18 date_created: 2023-08-21T12:18:00Z date_published: 2022-07-01T00:00:00Z date_updated: 2023-09-11T09:19:14Z day: '01' department: - _id: FrLo doi: 10.1109/cvpr52688.2022.01016 extern: '1' external_id: arxiv: - '2203.04913' fulldoi: https://doi.org/10.1109/cvpr52688.2022.01016 language: - iso: eng main_file_link: - open_access: '1' url: https://arxiv.org/abs/2203.04913 month: '07' oa: 1 oa_version: Preprint page: 10400-10411 publication: 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition publication_identifier: eissn: - 2575-7075 isbn: - '9781665469470' issn: - 1063-6919 publication_status: published publisher: Institute of Electrical and Electronics Engineers quality_controlled: '1' scopus_import: '1' status: public title: 'Leveling down in computer vision: Pareto inefficiencies in fair deep classifiers' type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2022' ...