[{"fulldoi":"https://doi.org/10.1145/3774904.3792868","language":[{"iso":"eng"}],"date_published":"2026-04-12T00:00:00Z","type":"conference","oa_version":"None","publication_identifier":{"isbn":["9798400723070"]},"conference":{"location":"Dubai","start_date":"2026-06-29","end_date":"2026-07-03","name":"WWW: Web Conference"},"date_created":"2026-05-24T22:01:32Z","citation":{"ieee":"M. Damie and E. A. L. Cyffers, “Fedivertex: A graph dataset based on decentralized Social Media,” in <i>2026 Proceedings of the ACM Web Conference</i>, Dubai, pp. 8393–8396.","chicago":"Damie, Marc, and Edwige Audrey Lucienne Cyffers. “Fedivertex: A Graph Dataset Based on Decentralized Social Media.” In <i>2026 Proceedings of the ACM Web Conference</i>, 8393–96. ACM, n.d. <a href=\"https://doi.org/10.1145/3774904.3792868\">https://doi.org/10.1145/3774904.3792868</a>.","short":"M. Damie, E.A.L. Cyffers, in:, 2026 Proceedings of the ACM Web Conference, ACM, n.d., pp. 8393–8396.","apa":"Damie, M., &#38; Cyffers, E. A. L. (n.d.). Fedivertex: A graph dataset based on decentralized Social Media. In <i>2026 Proceedings of the ACM Web Conference</i> (pp. 8393–8396). Dubai: ACM. <a href=\"https://doi.org/10.1145/3774904.3792868\">https://doi.org/10.1145/3774904.3792868</a>","mla":"Damie, Marc, and Edwige Audrey Lucienne Cyffers. “Fedivertex: A Graph Dataset Based on Decentralized Social Media.” <i>2026 Proceedings of the ACM Web Conference</i>, ACM, pp. 8393–96, doi:<a href=\"https://doi.org/10.1145/3774904.3792868\">10.1145/3774904.3792868</a>.","ista":"Damie M, Cyffers EAL. Fedivertex: A graph dataset based on decentralized Social Media. 2026 Proceedings of the ACM Web Conference. WWW: Web Conference, 8393–8396.","ama":"Damie M, Cyffers EAL. Fedivertex: A graph dataset based on decentralized Social Media. In: <i>2026 Proceedings of the ACM Web Conference</i>. ACM; :8393-8396. doi:<a href=\"https://doi.org/10.1145/3774904.3792868\">10.1145/3774904.3792868</a>"},"publication_status":"accepted","OA_type":"closed access","department":[{"_id":"ChLa"}],"article_processing_charge":"No","month":"04","day":"12","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","status":"public","scopus_import":"1","doi":"10.1145/3774904.3792868","date_updated":"2026-07-22T07:38:14Z","year":"2026","abstract":[{"text":"Social network graphs are central to graph learning research, serving as standard benchmarks for algorithm evaluation. However, existing datasets focus mainly on mainstream social media platforms whose structures are shaped notably by algorithmic recommendations. This raises an important question: would alternative, decentralized social networks exhibit different properties? We address this by studying the Fediverse; a collection of decentralized social networks (such as Mastodon and Lemmy). These platforms differ fundamentally from for-profit social media, notably in decentralization and absence of recommendation algorithms, which may yield distinct graph structures. We introduce Fedivertex, a dataset of over 400 graphs from seven decentralized networks, collected weekly over six months. The dataset, released with a companion Python package to facilitate its use, supports research on temporal and structural aspects of decentralized social networks. In particular, we benchmark applications to decentralized machine learning and community detection.","lang":"eng"}],"title":"Fedivertex: A graph dataset based on decentralized Social Media","_id":"21916","author":[{"full_name":"Damie, Marc","last_name":"Damie","first_name":"Marc"},{"first_name":"Edwige Audrey Lucienne","id":"20d4c299-977a-11ef-ae55-98b15ac64a57","last_name":"Cyffers","full_name":"Cyffers, Edwige Audrey Lucienne"}],"page":"8393-8396","publisher":"ACM","publication":"2026 Proceedings of the ACM Web Conference"},{"language":[{"iso":"eng"}],"acknowledgement":"We thank Rasmus Pagh, Christoph Lampert and Jalaj Upadhyay for valuable\r\ncomments on an early draft. We thank Ryan Mckenna for a fruitful discussion on the experiment\r\ndesign. We thank Antti Honkela for sharing insights on learning rate scheduling and DP.\r\nNikita P. Kalinin: Funded in part by the Austrian Science Fund (FWF) [10.55776/COE12].\r\nJoel Daniel Andersson: Funded by the European Union. Views and opinions expressed are however\r\nthose of the author(s) only and do not necessarily reflect those of the European Union or the European\r\nResearch Council Executive Agency. Neither the European Union nor the granting authority can be\r\nheld responsible for them. This project has received funding from the European Research Council\r\n(ERC) under the European Union’s Horizon 2020 research and innovation programme (MoDynStruct,\r\nNo. 101019564). Additional funding by Providentia, a Data Science Distinguished Investigator grant\r\nfrom Novo Nordisk Fonden, with additional support from VILLUM Investigator grant 54451.\r\n","oa_version":"Published Version","tmp":{"short":"CC BY (4.0)","legal_code_url":"https://creativecommons.org/licenses/by/4.0/legalcode","name":"Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)","image":"/images/cc_by.png"},"citation":{"ista":"Kalinin N, Andersson JD. 2026. Learning rate scheduling with matrix factorization for private training. 7th Symposium on Foundations of Responsible Computing. FORC: Symposium on Foundations of Responsible Computing, LIPIcs, vol. 368, 2:1-2:21.","mla":"Kalinin, Nikita, and Joel D. Andersson. “Learning Rate Scheduling with Matrix Factorization for Private Training.” <i>7th Symposium on Foundations of Responsible Computing</i>, vol. 368, 2:1-2:21, Schloss Dagstuhl - Leibniz-Zentrum für Informatik, 2026, doi:<a href=\"https://doi.org/10.4230/LIPIcs.FORC.2026.2\">10.4230/LIPIcs.FORC.2026.2</a>.","ama":"Kalinin N, Andersson JD. Learning rate scheduling with matrix factorization for private training. In: <i>7th Symposium on Foundations of Responsible Computing</i>. Vol 368. Schloss Dagstuhl - Leibniz-Zentrum für Informatik; 2026. doi:<a href=\"https://doi.org/10.4230/LIPIcs.FORC.2026.2\">10.4230/LIPIcs.FORC.2026.2</a>","chicago":"Kalinin, Nikita, and Joel D Andersson. “Learning Rate Scheduling with Matrix Factorization for Private Training.” In <i>7th Symposium on Foundations of Responsible Computing</i>, Vol. 368. Schloss Dagstuhl - Leibniz-Zentrum für Informatik, 2026. <a href=\"https://doi.org/10.4230/LIPIcs.FORC.2026.2\">https://doi.org/10.4230/LIPIcs.FORC.2026.2</a>.","ieee":"N. Kalinin and J. D. Andersson, “Learning rate scheduling with matrix factorization for private training,” in <i>7th Symposium on Foundations of Responsible Computing</i>, Cambridge, MA; United States, 2026, vol. 368.","apa":"Kalinin, N., &#38; Andersson, J. D. (2026). Learning rate scheduling with matrix factorization for private training. In <i>7th Symposium on Foundations of Responsible Computing</i> (Vol. 368). Cambridge, MA; United States: Schloss Dagstuhl - Leibniz-Zentrum für Informatik. <a href=\"https://doi.org/10.4230/LIPIcs.FORC.2026.2\">https://doi.org/10.4230/LIPIcs.FORC.2026.2</a>","short":"N. Kalinin, J.D. Andersson, in:, 7th Symposium on Foundations of Responsible Computing, Schloss Dagstuhl - Leibniz-Zentrum für Informatik, 2026."},"keyword":["differential privacy","machine learning","matrix factorization"],"publication_identifier":{"eissn":["1868-8969"],"isbn":["9783959774192"]},"file":[{"date_created":"2026-06-29T06:55:23Z","file_name":"2026_LIPIcsFORC_Kalinin.pdf","checksum":"c661f016d3861a1c1b590b87a744d087","content_type":"application/pdf","date_updated":"2026-06-29T06:55:23Z","access_level":"open_access","success":1,"creator":"dernst","file_size":1231914,"file_id":"22149","relation":"main_file"}],"file_date_updated":"2026-06-29T06:55:23Z","abstract":[{"lang":"eng","text":"We study differentially private model training with stochastic gradient descent under learning rate scheduling and correlated noise. Although correlated noise, in particular via matrix factorizations, has been shown to improve accuracy, prior theoretical work focused primarily on the prefix-sum workload. That workload assumes a constant learning rate, whereas in practice learning rate schedules are widely used to accelerate training and improve convergence. We close this gap by deriving general upper and lower bounds for a broad class of learning rate schedules in both single- and multi-epoch settings. Building on these results, we propose a learning-rate-aware factorization that achieves improvements over prefix-sum factorizations under both MaxSE and MeanSE error metrics. Our theoretical analysis yields memory-efficient constructions suitable for practical deployment, and experiments on CIFAR-10 and IMDB datasets confirm that schedule-aware factorizations improve accuracy in private training."}],"volume":368,"supplementarymaterial":"no","alternative_title":["LIPIcs"],"scopus_import":"1","day":"01","researchdata_availability":"no","project":[{"grant_number":"101019564","name":"The design and evaluation of modern fully dynamic data structures","call_identifier":"H2020","_id":"bd9ca328-d553-11ed-ba76-dc4f890cfe62"},{"_id":"d8f03aaa-b035-11f1-8588-d5147fa879e0","name":"Bilateral Artificial Intelligence (Lampert)","grant_number":"COE12"}],"publication":"7th Symposium on Foundations of Responsible Computing","ddc":["000"],"license":"https://creativecommons.org/licenses/by/4.0/","ec_funded":1,"das_tickbox":"0","fulldoi":"https://doi.org/10.4230/LIPIcs.FORC.2026.2","date_published":"2026-06-01T00:00:00Z","type":"conference","arxiv":1,"date_created":"2026-06-28T22:01:34Z","conference":{"name":"FORC: Symposium on Foundations of Responsible Computing","end_date":"2026-06-05","location":"Cambridge, MA; United States","start_date":"2026-06-03"},"external_id":{"arxiv":["2511.17994"]},"article_number":"2:1-2:21","article_processing_charge":"No","month":"06","OA_place":"publisher","department":[{"_id":"ChLa"},{"_id":"GradSch"},{"_id":"MoHe"}],"OA_type":"gold","publication_status":"published","year":"2026","date_updated":"2026-09-16T07:37:21Z","title":"Learning rate scheduling with matrix factorization for private training","doi":"10.4230/LIPIcs.FORC.2026.2","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","status":"public","has_accepted_license":"1","quality_controlled":"1","_id":"22146","intvolume":"       368","publisher":"Schloss Dagstuhl - Leibniz-Zentrum für Informatik","oa":1,"author":[{"last_name":"Kalinin","full_name":"Kalinin, Nikita","first_name":"Nikita","id":"4b14526e-14d2-11ed-ba64-c14c9553d137"},{"last_name":"Andersson","full_name":"Andersson, Joel D","first_name":"Joel D","id":"4a893819-d954-11f0-89b1-e360bad9ccc5"}],"corr_author":"1"},{"department":[{"_id":"GradSch"},{"_id":"ChLa"}],"month":"02","article_processing_charge":"No","OA_place":"publisher","publication_status":"published","supervisor":[{"orcid":"0000-0001-8622-7887","first_name":"Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","full_name":"Lampert, Christoph","last_name":"Lampert"}],"date_created":"2026-02-09T14:59:53Z","type":"dissertation","date_published":"2026-02-09T00:00:00Z","related_material":{"record":[{"relation":"part_of_dissertation","id":"21207","status":"public"},{"status":"public","id":"20819","relation":"part_of_dissertation"},{"id":"17411","relation":"part_of_dissertation","status":"public"},{"status":"public","relation":"part_of_dissertation","id":"18120"}]},"fulldoi":"https://doi.org/10.15479/AT-ISTA-21198","publisher":"Institute of Science and Technology Austria","oa":1,"author":[{"last_name":"Scott","full_name":"Scott, Jonathan A","id":"e499926b-f6e0-11ea-865d-9c63db0031e8","first_name":"Jonathan A"}],"corr_author":"1","_id":"21198","degree_awarded":"PhD","doi":"10.15479/AT-ISTA-21198","acknowledged_ssus":[{"_id":"ScienComp"}],"year":"2026","date_updated":"2026-10-01T09:29:03Z","title":"Data heterogeneity and personalization in federated learning","status":"public","has_accepted_license":"1","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","file":[{"content_type":"application/zip","date_created":"2026-02-17T11:46:22Z","checksum":"121c1d968bd86f3630aa7e81d5bbbcb0","file_name":"2026_Scott_Jonathan_Thesis_Source.zip","creator":"jscott","relation":"source_file","file_id":"21298","file_size":272379252,"date_updated":"2026-02-17T11:46:22Z","access_level":"closed"},{"content_type":"application/pdf","file_name":"2026_Jonathan_Scott_Thesis.pdf","checksum":"6e3e08ba474bbee8511cc8a839ab2077","date_created":"2026-02-27T10:25:41Z","file_size":15220298,"relation":"main_file","file_id":"21366","creator":"jscott","access_level":"open_access","success":1,"date_updated":"2026-02-27T10:25:41Z"}],"file_date_updated":"2026-02-27T10:25:41Z","citation":{"chicago":"Scott, Jonathan A. “Data Heterogeneity and Personalization in Federated Learning.” Institute of Science and Technology Austria, 2026. <a href=\"https://doi.org/10.15479/AT-ISTA-21198\">https://doi.org/10.15479/AT-ISTA-21198</a>.","ieee":"J. A. Scott, “Data heterogeneity and personalization in federated learning,” Institute of Science and Technology Austria, 2026.","apa":"Scott, J. A. (2026). <i>Data heterogeneity and personalization in federated learning</i>. Institute of Science and Technology Austria. <a href=\"https://doi.org/10.15479/AT-ISTA-21198\">https://doi.org/10.15479/AT-ISTA-21198</a>","short":"J.A. Scott, Data Heterogeneity and Personalization in Federated Learning, Institute of Science and Technology Austria, 2026.","ista":"Scott JA. 2026. Data heterogeneity and personalization in federated learning. Institute of Science and Technology Austria.","mla":"Scott, Jonathan A. <i>Data Heterogeneity and Personalization in Federated Learning</i>. Institute of Science and Technology Austria, 2026, doi:<a href=\"https://doi.org/10.15479/AT-ISTA-21198\">10.15479/AT-ISTA-21198</a>.","ama":"Scott JA. Data heterogeneity and personalization in federated learning. 2026. doi:<a href=\"https://doi.org/10.15479/AT-ISTA-21198\">10.15479/AT-ISTA-21198</a>"},"publication_identifier":{"issn":["2663-337X"]},"oa_version":"Published Version","language":[{"iso":"eng"}],"acknowledgement":"This research was funded in part by the Austrian Science Fund (FWF)\r\n[10.55776/COE12]. Furthermore, the candidate acknowledges the support from the Scientific\r\nService Units (SSU) of ISTA through resources provided by Scientific Computing (SciComp).","doi_confirm":"1","ddc":["005"],"page":"158","project":[{"grant_number":"COE12","name":"Bilateral Artificial Intelligence (Lampert)","_id":"d8f03aaa-b035-11f1-8588-d5147fa879e0"}],"alternative_title":["ISTA Thesis"],"abstract":[{"text":"In recent years there has been a massive increase in the amount of data generated in a\r\ndecentralized manner. Ever more powerful edge devices, such as smartphones, have become\r\nubiquitous in most societies on earth. Through text typed, photos taken and apps used,\r\nthese devices, which we refer to as clients, generate enormous amounts of high quality and\r\ncomplex data. Moreover, the nature of these devices means the data they generate is often\r\nsensitive and privacy concerns prevent it being gathered and stored in a central location. This\r\npresents a challenge to the modern machine learning paradigm that requires central access\r\nto large amounts of data. Federated learning (FL) has emerged as one of the answers to\r\nthis problem. Rather than bringing the data to the model, FL sends the model to the data.\r\nModel training takes place on device, with periodically synchronized updates, allowing data to\r\nremain locally stored. While this approach offers significant privacy advantages it comes with\r\nits own set of unique challenges. These include: data heterogeneity, the notion that different\r\ndevices generate data in distinct ways which can negatively impact training dynamics; systems\r\nheterogeneity, meaning that different devices may have differing hardware specifications; high\r\ncommunication costs, which are induced by the repeated transferring of models over the\r\nnetwork and low device computational power, which limits the use of larger models on device.\r\nIn this thesis we present a range of methods for federated learning. We focus primarily on\r\nthe challenge of data heterogeneity, though the methods presented are designed to be well\r\nadapted to the other challenges of a federated setting, such as the constraints of limited\r\ncompute and communication overhead. We first present a method for explicitly modeling client\r\ndata heterogeneity. The approach formulates clients as samples from a certain probability\r\ndistribution and infers the parameters of this distribution from the available training clients.\r\nThis learned distribution then represents the heterogeneity present among the clients and can\r\nbe sampled from in order to create new simulated clients that are similar to the real clients we\r\nhave observed so far. Following this we present two methods for directly dealing with data\r\nheterogeneity through personalization. Highly heterogeneous client data distributions can mean\r\nthat learning a single global model becomes suboptimal, and some form of personalization of\r\nmodels to each individual client is required. Our approaches are based around hypernetworks,\r\nwhich we use to generate personalized model parameters without the need for additional\r\ntraining or finetuning. In the first approach we focus on generating full parameterizations of\r\nclient models using learned embeddings of client data and labels, with a hypernetwork located\r\non the central server. In the second approach we address the more challenging scenario where\r\nwe want to generate a personalized model for a client without any label information. The\r\nhypernetwork is trained to generate a low dimensional representation of a client’s personalized\r\nmodel parameters, allowing it to be transferred to and run on the client devices. In our final\r\npresented method, we change our focus and rather than aim to directly address the challenge\r\nof data heterogeneity, we instead ensure we are unaffected by it. This is done in the context\r\nof k-means clustering and we present a method for federated clustering with a focus on added\r\nprivacy guarantees.","lang":"eng"}],"day":"09"},{"status":"public","quality_controlled":"1","has_accepted_license":"1","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","date_updated":"2025-09-03T10:37:59Z","title":"Predictive monitoring of black-box dynamical systems","year":"2025","_id":"20256","corr_author":"1","oa":1,"author":[{"orcid":"0000-0002-2985-7724","first_name":"Thomas A","id":"40876CD8-F248-11E8-B48F-1D18A9856A87","full_name":"Henzinger, Thomas A","last_name":"Henzinger"},{"first_name":"Fabian","id":"faff3c84-23f6-11ef-9085-e5187b51c604","last_name":"Kresse","full_name":"Kresse, Fabian"},{"first_name":"Kaushik","id":"0834ff3c-6d72-11ec-94e0-b5b0a4fb8598","orcid":"0000-0001-9864-7475","last_name":"Mallik","full_name":"Mallik, Kaushik"},{"id":"20aa2ae8-f2f1-11ed-bbfa-8205053f1342","first_name":"Zhengqi","full_name":"Yu, Zhengqi","last_name":"Yu"},{"full_name":"Zikelic, Dorde","last_name":"Zikelic","orcid":"0000-0002-4681-1699","id":"294AA7A6-F248-11E8-B48F-1D18A9856A87","first_name":"Dorde"}],"publisher":"ML Research Press","intvolume":"       283","arxiv":1,"date_published":"2025-06-01T00:00:00Z","type":"conference","external_id":{"arxiv":["2412.16564"]},"conference":{"end_date":"2025-06-06","name":"L4DC: Learning for Dynamics & Control","start_date":"2025-06-04","location":"Ann Arbor, MI, United States"},"date_created":"2025-08-31T22:01:32Z","publication_status":"published","OA_place":"publisher","article_processing_charge":"No","month":"06","department":[{"_id":"ToHe"},{"_id":"ChLa"}],"OA_type":"gold","scopus_import":"1","day":"01","abstract":[{"lang":"eng","text":"We study the problem of predictive runtime monitoring of black-box dynamical systems with quantitative safety properties. The black-box setting stipulates that the exact semantics of the dynamical system and the controller are unknown, and that we are only able to observe the state of the controlled (aka, closed-loop) system at finitely many time points. We present a novel framework for predicting future states of the system based on the states observed in the past. The numbers of past states and of predicted future states are parameters provided by the user. Our method is based on a combination of Taylor’s expansion and the backward difference operator for numerical differentiation. We also derive an upper bound on the prediction error under the assumption that the system dynamics and the controller are smooth. The predicted states are then used to predict safety violations ahead in time. Our experiments demonstrate practical applicability of our method for complex black-box systems, showing that it is computationally lightweight and yet significantly more accurate than the state-of-the-art predictive safety monitoring techniques."}],"volume":283,"alternative_title":["PMLR"],"project":[{"grant_number":"101020093","name":"Vigilant Algorithmic Monitoring of Software","call_identifier":"H2020","_id":"62781420-2b32-11ec-9570-8d9b63373d4d"}],"page":"804-816","ddc":["000"],"publication":"7th Annual Learning for Dynamics & Control Conference","ec_funded":1,"acknowledgement":"This work was supported in part by the ERC project ERC-2020-AdG 101020093.\r\n","language":[{"iso":"eng"}],"oa_version":"Published Version","publication_identifier":{"eissn":["2640-3498"]},"citation":{"short":"T.A. Henzinger, F. Kresse, K. Mallik, E. Yu, D. Zikelic, in:, 7th Annual Learning for Dynamics &#38; Control Conference, ML Research Press, 2025, pp. 804–816.","apa":"Henzinger, T. A., Kresse, F., Mallik, K., Yu, E., &#38; Zikelic, D. (2025). Predictive monitoring of black-box dynamical systems. In <i>7th Annual Learning for Dynamics &#38; Control Conference</i> (Vol. 283, pp. 804–816). Ann Arbor, MI, United States: ML Research Press.","chicago":"Henzinger, Thomas A, Fabian Kresse, Kaushik Mallik, Emily Yu, and Dorde Zikelic. “Predictive Monitoring of Black-Box Dynamical Systems.” In <i>7th Annual Learning for Dynamics &#38; Control Conference</i>, 283:804–16. ML Research Press, 2025.","ieee":"T. A. Henzinger, F. Kresse, K. Mallik, E. Yu, and D. Zikelic, “Predictive monitoring of black-box dynamical systems,” in <i>7th Annual Learning for Dynamics &#38; Control Conference</i>, Ann Arbor, MI, United States, 2025, vol. 283, pp. 804–816.","ama":"Henzinger TA, Kresse F, Mallik K, Yu E, Zikelic D. Predictive monitoring of black-box dynamical systems. In: <i>7th Annual Learning for Dynamics &#38; Control Conference</i>. Vol 283. ML Research Press; 2025:804-816.","mla":"Henzinger, Thomas A., et al. “Predictive Monitoring of Black-Box Dynamical Systems.” <i>7th Annual Learning for Dynamics &#38; Control Conference</i>, vol. 283, ML Research Press, 2025, pp. 804–16.","ista":"Henzinger TA, Kresse F, Mallik K, Yu E, Zikelic D. 2025. Predictive monitoring of black-box dynamical systems. 7th Annual Learning for Dynamics &#38; Control Conference. L4DC: Learning for Dynamics &#38; Control, PMLR, vol. 283, 804–816."},"file_date_updated":"2025-09-03T10:32:12Z","file":[{"success":1,"access_level":"open_access","date_updated":"2025-09-03T10:32:12Z","relation":"main_file","file_id":"20283","file_size":489639,"creator":"dernst","checksum":"d5236e561560635f5ae1d17de4903033","file_name":"2025_L4DC_HenzingerT.pdf","date_created":"2025-09-03T10:32:12Z","content_type":"application/pdf"}]},{"article_processing_charge":"No","month":"05","OA_place":"publisher","OA_type":"diamond","department":[{"_id":"ChLa"}],"publication_status":"published","date_created":"2025-09-07T22:01:34Z","conference":{"location":"Mai Khao, Thailand","start_date":"2025-05-03","end_date":"2025-05-05","name":"AISTATS: Conference on Artificial Intelligence and Statistics"},"external_id":{"arxiv":["2402.04840"]},"type":"conference","date_published":"2025-05-01T00:00:00Z","arxiv":1,"publisher":"ML Research Press","intvolume":"       258","oa":1,"corr_author":"1","author":[{"first_name":"Nikita","id":"4b14526e-14d2-11ed-ba64-c14c9553d137","last_name":"Kalinin","full_name":"Kalinin, Nikita"},{"first_name":"Lukas","full_name":"Steinberger, Lukas","last_name":"Steinberger"}],"_id":"20298","year":"2025","title":"Efficient estimation of a Gaussian mean with local differential privacy","date_updated":"2025-09-09T08:28:41Z","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","status":"public","has_accepted_license":"1","quality_controlled":"1","file":[{"file_name":"2025_AISTATS_Kalinin.pdf","checksum":"3dcd59988ca974b98662ba09a516e616","date_created":"2025-09-09T08:26:44Z","content_type":"application/pdf","access_level":"open_access","success":1,"date_updated":"2025-09-09T08:26:44Z","file_size":395864,"relation":"main_file","file_id":"20316","creator":"dernst"}],"file_date_updated":"2025-09-09T08:26:44Z","citation":{"apa":"Kalinin, N., &#38; Steinberger, L. (2025). Efficient estimation of a Gaussian mean with local differential privacy. In <i>Proceedings of the 28th International Conference on Artificial Intelligence and Statistics</i> (Vol. 258, pp. 118–126). Mai Khao, Thailand: ML Research Press.","short":"N. Kalinin, L. Steinberger, in:, Proceedings of the 28th International Conference on Artificial Intelligence and Statistics, ML Research Press, 2025, pp. 118–126.","ieee":"N. Kalinin and L. Steinberger, “Efficient estimation of a Gaussian mean with local differential privacy,” in <i>Proceedings of the 28th International Conference on Artificial Intelligence and Statistics</i>, Mai Khao, Thailand, 2025, vol. 258, pp. 118–126.","chicago":"Kalinin, Nikita, and Lukas Steinberger. “Efficient Estimation of a Gaussian Mean with Local Differential Privacy.” In <i>Proceedings of the 28th International Conference on Artificial Intelligence and Statistics</i>, 258:118–26. ML Research Press, 2025.","ama":"Kalinin N, Steinberger L. Efficient estimation of a Gaussian mean with local differential privacy. In: <i>Proceedings of the 28th International Conference on Artificial Intelligence and Statistics</i>. Vol 258. ML Research Press; 2025:118-126.","ista":"Kalinin N, Steinberger L. 2025. Efficient estimation of a Gaussian mean with local differential privacy. Proceedings of the 28th International Conference on Artificial Intelligence and Statistics. AISTATS: Conference on Artificial Intelligence and Statistics, PMLR, vol. 258, 118–126.","mla":"Kalinin, Nikita, and Lukas Steinberger. “Efficient Estimation of a Gaussian Mean with Local Differential Privacy.” <i>Proceedings of the 28th International Conference on Artificial Intelligence and Statistics</i>, vol. 258, ML Research Press, 2025, pp. 118–26."},"publication_identifier":{"eissn":["2640-3498"]},"oa_version":"Published Version","tmp":{"short":"CC BY (4.0)","legal_code_url":"https://creativecommons.org/licenses/by/4.0/legalcode","name":"Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)","image":"/images/cc_by.png"},"language":[{"iso":"eng"}],"acknowledgement":"We would like to express our gratitude to Christoph Lampert for his valuable insights and fruitful discussions that significantly contributed to the development of this paper.\r\nWe also thank Salil Vadhan for his constructive feedback on an earlier version of this draft.\r\nThe second author gratefully acknowledges support by the Austrian Science Fund (FWF): I 5484-N, as part of the Research Unit 5381 of the German Research Foundation.","publication":"Proceedings of the 28th International Conference on Artificial Intelligence and Statistics","ddc":["000"],"page":"118-126","abstract":[{"lang":"eng","text":"In this paper, we study the problem of estimating the unknown mean θ of a unit variance Gaussian distribution in a locally differentially private (LDP) way. In the high-privacy regime (ϵ≤1\r\n), we identify an optimal privacy mechanism that minimizes the variance of the estimator asymptotically. Our main technical contribution is the maximization of the Fisher-Information of the sanitized data with respect to the local privacy mechanism Q. We find that the exact solution Qθ,ϵ of this maximization is the sign mechanism that applies randomized response to the sign of Xi−θ, where X1,…,Xn are the confidential iid original samples. However, since this optimal local mechanism depends on the unknown mean θ, we employ a two-stage LDP parameter estimation procedure which requires splitting agents into two groups. The first n1 observations are used to consistently but not necessarily efficiently estimate the parameter θ by θn1~\r\n. Then this estimate is updated by applying the sign mechanism with θ~n1 instead of θ\r\n to the remaining n−n1 observations, to obtain an LDP and efficient estimator of the unknown mean."}],"volume":258,"alternative_title":["PMLR"],"scopus_import":"1","day":"01"},{"department":[{"_id":"MaMo"},{"_id":"GradSch"},{"_id":"ChLa"}],"OA_type":"free access","month":"12","OA_place":"publisher","article_processing_charge":"No","main_file_link":[{"open_access":"1","url":"https://doi.org/10.52202/085713-1450"}],"publication_status":"published","date_created":"2026-09-06T22:01:59Z","external_id":{"arxiv":["2505.15239"]},"conference":{"name":"NeurIPS: Neural Information Processing Systems","end_date":"2025-12-07","location":"San Diego, CA, United States","start_date":"2025-12-02"},"type":"conference","date_published":"2025-12-02T00:00:00Z","arxiv":1,"das_tickbox":"0","fulldoi":"https://doi.org/10.52202/085713-1450","intvolume":"        38","publisher":"Neural Information Processing Systems Foundation","author":[{"first_name":"Peter","id":"d64d6a8d-eb8e-11eb-b029-96fd216dec3c","last_name":"Súkeník","full_name":"Súkeník, Peter"},{"last_name":"Lampert","full_name":"Lampert, Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","first_name":"Christoph","orcid":"0000-0001-8622-7887"},{"full_name":"Mondelli, Marco","last_name":"Mondelli","orcid":"0000-0002-3242-7020","id":"27EB676C-8706-11E9-9510-7717E6697425","first_name":"Marco"}],"oa":1,"corr_author":"1","_id":"22825","doi":"10.52202/085713-1450","acknowledged_ssus":[{"_id":"ScienComp"}],"date_updated":"2026-09-10T08:11:46Z","title":"Neural collapse is globally optimal in deep regularized ResNets and transformers","year":"2025","status":"public","quality_controlled":"1","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","citation":{"ieee":"P. Súkeník, C. Lampert, and M. Mondelli, “Neural collapse is globally optimal in deep regularized ResNets and transformers,” in <i>39th Conference on Neural Information Processing Systems</i>, San Diego, CA, United States, 2025, vol. 38, pp. 48646–48677.","chicago":"Súkeník, Peter, Christoph Lampert, and Marco Mondelli. “Neural Collapse Is Globally Optimal in Deep Regularized ResNets and Transformers.” In <i>39th Conference on Neural Information Processing Systems</i>, 38:48646–77. Neural Information Processing Systems Foundation, 2025. <a href=\"https://doi.org/10.52202/085713-1450\">https://doi.org/10.52202/085713-1450</a>.","short":"P. Súkeník, C. Lampert, M. Mondelli, in:, 39th Conference on Neural Information Processing Systems, Neural Information Processing Systems Foundation, 2025, pp. 48646–48677.","apa":"Súkeník, P., Lampert, C., &#38; Mondelli, M. (2025). Neural collapse is globally optimal in deep regularized ResNets and transformers. In <i>39th Conference on Neural Information Processing Systems</i> (Vol. 38, pp. 48646–48677). San Diego, CA, United States: Neural Information Processing Systems Foundation. <a href=\"https://doi.org/10.52202/085713-1450\">https://doi.org/10.52202/085713-1450</a>","mla":"Súkeník, Peter, et al. “Neural Collapse Is Globally Optimal in Deep Regularized ResNets and Transformers.” <i>39th Conference on Neural Information Processing Systems</i>, vol. 38, Neural Information Processing Systems Foundation, 2025, pp. 48646–77, doi:<a href=\"https://doi.org/10.52202/085713-1450\">10.52202/085713-1450</a>.","ista":"Súkeník P, Lampert C, Mondelli M. 2025. Neural collapse is globally optimal in deep regularized ResNets and transformers. 39th Conference on Neural Information Processing Systems. NeurIPS: Neural Information Processing Systems, Advances in Neural Information Processing Systems, vol. 38, 48646–48677.","ama":"Súkeník P, Lampert C, Mondelli M. Neural collapse is globally optimal in deep regularized ResNets and transformers. In: <i>39th Conference on Neural Information Processing Systems</i>. Vol 38. Neural Information Processing Systems Foundation; 2025:48646-48677. doi:<a href=\"https://doi.org/10.52202/085713-1450\">10.52202/085713-1450</a>"},"publication_identifier":{"isbn":["9798331338275"],"issn":["1049-5258"]},"oa_version":"Published Version","language":[{"iso":"eng"}],"acknowledgement":"M. M. and P. S. are funded by the European Union (ERC, INF2\r\n, project number 101161364). Views\r\nand opinions expressed are however those of the author(s) only and do not necessarily reflect those\r\nof the European Union or the European Research Council Executive Agency. Neither the European\r\nUnion nor the granting authority can be held responsible for them. This research was supported\r\nby the Scientific Service Units (SSU) of ISTA through resources provided by Scientific Computing (SciComp).","ddc":["000"],"publication":"39th Conference on Neural Information Processing Systems","page":"48646-48677","researchdata_availability":"no","project":[{"grant_number":"101161364","name":"Inference in High Dimensions: Light-speed Algorithms and Information Limits","_id":"911e6d1f-16d5-11f0-9cad-c5c68c6a1cdf"}],"volume":38,"supplementarymaterial":"no","alternative_title":["Advances in Neural Information Processing Systems"],"abstract":[{"text":"The empirical emergence of neural collapse—a surprising symmetry in the feature representations of the training data in the penultimate layer of deep neural\r\nnetworks—has spurred a line of theoretical research aimed at its understanding.\r\nHowever, existing work focuses on data-agnostic models or, when data structure is\r\ntaken into account, it remains limited to multi-layer perceptrons. Our paper fills\r\nboth these gaps by analyzing modern architectures in a data-aware regime: we\r\nprove that global optima of deep regularized transformers and residual networks\r\n(ResNets) with LayerNorm trained with cross entropy or mean squared error loss\r\nare approximately collapsed, and the approximation gets tighter as the depth grows.\r\nMore generally, we formally reduce any end-to-end large-depth ResNet or transformer training into an equivalent unconstrained features model, thus justifying its\r\nwide use in the literature even beyond data-agnostic settings. Our theoretical results\r\nare supported by experiments on computer vision and language datasets showing\r\nthat, as the depth grows, neural collapse indeed becomes more prominent.","lang":"eng"}],"day":"02","scopus_import":"1"},{"researchdata_availability":"no","abstract":[{"lang":"eng","text":"We present new fast-rate PAC-Bayesian generalization bounds for multi-task and\r\nmeta-learning in the unbalanced setting, i.e. when the tasks have training sets of\r\ndifferent sizes, as is typically the case in real-world scenarios. Previously, only\r\nstandard-rate bounds were known for this situation, while fast-rate bounds were\r\nlimited to the setting where all training sets are of equal size. Our new bounds\r\nare numerically computable as well as interpretable, and we demonstrate their\r\nflexibility in handling a number of cases where they give stronger guarantees\r\nthan previous bounds. Besides the bounds themselves, we also make conceptual\r\ncontributions: we demonstrate that the unbalanced multi-task setting has different\r\nstatistical properties than the balanced situation, specifically that proofs from\r\nthe balanced situation do not carry over to the unbalanced setting. Additionally,\r\nwe shed light on the fact that the unbalanced situation allows two meaningful\r\ndefinitions of multi-task risk, depending on whether all tasks should be considered\r\nequally important or if sample-rich tasks should receive more weight than samplepoor ones."}],"volume":38,"supplementarymaterial":"yes","alternative_title":["Advances in Neural Information Processing Systems"],"scopus_import":"1","day":"02","publication":"39th Conference on Neural Information Processing Systems","page":"9062-9093","oa_version":"Published Version","language":[{"iso":"eng"}],"acknowledgement":"This research was supported by the Scientific Service Units (SSU) of ISTA through resources provided by Scientific Computing (SciComp).","citation":{"short":"H. Zakerinia, C. Lampert, in:, 39th Conference on Neural Information Processing Systems, Neural Information Processing Systems Foundation, 2025, pp. 9062–9093.","apa":"Zakerinia, H., &#38; Lampert, C. (2025). Fast rate bounds for multi-task and meta-learning with different sample sizes. In <i>39th Conference on Neural Information Processing Systems</i> (Vol. 38, pp. 9062–9093). San Diego, CA, United States: Neural Information Processing Systems Foundation. <a href=\"https://doi.org/10.52202/085713-0278\">https://doi.org/10.52202/085713-0278</a>","chicago":"Zakerinia, Hossein, and Christoph Lampert. “Fast Rate Bounds for Multi-Task and Meta-Learning with Different Sample Sizes.” In <i>39th Conference on Neural Information Processing Systems</i>, 38:9062–93. Neural Information Processing Systems Foundation, 2025. <a href=\"https://doi.org/10.52202/085713-0278\">https://doi.org/10.52202/085713-0278</a>.","ieee":"H. Zakerinia and C. Lampert, “Fast rate bounds for multi-task and meta-learning with different sample sizes,” in <i>39th Conference on Neural Information Processing Systems</i>, San Diego, CA, United States, 2025, vol. 38, pp. 9062–9093.","ama":"Zakerinia H, Lampert C. Fast rate bounds for multi-task and meta-learning with different sample sizes. In: <i>39th Conference on Neural Information Processing Systems</i>. Vol 38. Neural Information Processing Systems Foundation; 2025:9062-9093. doi:<a href=\"https://doi.org/10.52202/085713-0278\">10.52202/085713-0278</a>","mla":"Zakerinia, Hossein, and Christoph Lampert. “Fast Rate Bounds for Multi-Task and Meta-Learning with Different Sample Sizes.” <i>39th Conference on Neural Information Processing Systems</i>, vol. 38, Neural Information Processing Systems Foundation, 2025, pp. 9062–93, doi:<a href=\"https://doi.org/10.52202/085713-0278\">10.52202/085713-0278</a>.","ista":"Zakerinia H, Lampert C. 2025. Fast rate bounds for multi-task and meta-learning with different sample sizes. 39th Conference on Neural Information Processing Systems. NeurIPS: Neural Information Processing Systems, Advances in Neural Information Processing Systems, vol. 38, 9062–9093."},"publication_identifier":{"issn":["1049-5258"],"isbn":["9798331338275"]},"_id":"22824","year":"2025","title":"Fast rate bounds for multi-task and meta-learning with different sample sizes","acknowledged_ssus":[{"_id":"ScienComp"}],"date_updated":"2026-09-10T08:37:33Z","doi":"10.52202/085713-0278","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","status":"public","quality_controlled":"1","intvolume":"        38","publisher":"Neural Information Processing Systems Foundation","oa":1,"author":[{"full_name":"Zakerinia, Hossein","last_name":"Zakerinia","orcid":"0009-0007-3977-6462","first_name":"Hossein","id":"653bd8b6-f394-11eb-9cf6-c0bbf6cd78d4"},{"full_name":"Lampert, Christoph","last_name":"Lampert","orcid":"0000-0001-8622-7887","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","first_name":"Christoph"}],"corr_author":"1","type":"conference","date_published":"2025-12-02T00:00:00Z","das_tickbox":"0","fulldoi":"https://doi.org/10.52202/085713-0278","article_processing_charge":"No","OA_place":"publisher","month":"12","department":[{"_id":"GradSch"},{"_id":"ChLa"}],"OA_type":"free access","publication_status":"published","main_file_link":[{"open_access":"1","url":"https://doi.org/10.52202/085713-0278"}],"date_created":"2026-09-06T22:01:59Z","conference":{"start_date":"2025-12-02","location":"San Diego, CA, United States","end_date":"2025-12-07","name":"NeurIPS: Neural Information Processing Systems"}},{"project":[{"call_identifier":"H2020","name":"Vigilant Algorithmic Monitoring of Software","grant_number":"101020093","_id":"62781420-2b32-11ec-9570-8d9b63373d4d"},{"_id":"d8f03aaa-b035-11f1-8588-d5147fa879e0","grant_number":"COE12","name":"Bilateral Artificial Intelligence (Lampert)"}],"scopus_import":"1","day":"01","abstract":[{"lang":"eng","text":"Learning-based systems are increasingly deployed across various domains, yet the complexity of traditional neural networks poses significant challenges for formal verification. Unlike conventional neural networks, learned Logic Gate Networks (LGNs) replace multiplications with Boolean logic gates, yielding a sparse, netlist-like architecture that is inherently more amenable to symbolic verification, while still delivering promising performance. In this paper, we introduce a SAT encoding for verifying global robustness and fairness in LGNs. We evaluate our method on five benchmark datasets, including a newly constructed 5-class variant, and find that LGNs are both verification-friendly and maintain strong predictive performance."}],"alternative_title":["PMLR"],"volume":288,"ec_funded":1,"publication":"2nd International Conferenceon Neuro-Symbolic Systems","ddc":["000"],"oa_version":"Published Version","acknowledgement":"This work is supported in part by the ERC grant under Grant No. ERC-2020-AdG 101020093 and\r\nthe Austrian Science Fund (FWF) [10.55776/COE12]. This research was supported by the Scientific\r\nService Units (SSU) of ISTA through resources provided by Scientific Computing (SciComp).","language":[{"iso":"eng"}],"file_date_updated":"2025-09-09T08:10:13Z","file":[{"date_updated":"2025-09-09T08:10:13Z","success":1,"access_level":"open_access","creator":"dernst","file_id":"20314","relation":"main_file","file_size":295466,"date_created":"2025-09-09T08:10:13Z","checksum":"90a32defed34787e771a5c1623b6b0d2","file_name":"2025_NeuS_Kresse.pdf","content_type":"application/pdf"}],"publication_identifier":{"eissn":["2640-3498"]},"citation":{"ieee":"F. Kresse, E. Yu, C. Lampert, and T. A. Henzinger, “Logic gate neural networks are good for verification,” in <i>2nd International Conferenceon Neuro-Symbolic Systems</i>, Philadephia, PA, United States, 2025, vol. 288.","chicago":"Kresse, Fabian, Emily Yu, Christoph Lampert, and Thomas A Henzinger. “Logic Gate Neural Networks Are Good for Verification.” In <i>2nd International Conferenceon Neuro-Symbolic Systems</i>, Vol. 288. ML Research Press, 2025.","apa":"Kresse, F., Yu, E., Lampert, C., &#38; Henzinger, T. A. (2025). Logic gate neural networks are good for verification. In <i>2nd International Conferenceon Neuro-Symbolic Systems</i> (Vol. 288). Philadephia, PA, United States: ML Research Press.","short":"F. Kresse, E. Yu, C. Lampert, T.A. Henzinger, in:, 2nd International Conferenceon Neuro-Symbolic Systems, ML Research Press, 2025.","ista":"Kresse F, Yu E, Lampert C, Henzinger TA. 2025. Logic gate neural networks are good for verification. 2nd International Conferenceon Neuro-Symbolic Systems. NeuS: International Conferenceon Neuro-Symbolic Systems, PMLR, vol. 288, 26.","mla":"Kresse, Fabian, et al. “Logic Gate Neural Networks Are Good for Verification.” <i>2nd International Conferenceon Neuro-Symbolic Systems</i>, vol. 288, 26, ML Research Press, 2025.","ama":"Kresse F, Yu E, Lampert C, Henzinger TA. Logic gate neural networks are good for verification. In: <i>2nd International Conferenceon Neuro-Symbolic Systems</i>. Vol 288. ML Research Press; 2025."},"_id":"20296","status":"public","quality_controlled":"1","has_accepted_license":"1","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","year":"2025","acknowledged_ssus":[{"_id":"ScienComp"}],"date_updated":"2026-09-16T07:02:18Z","title":"Logic gate neural networks are good for verification","oa":1,"corr_author":"1","author":[{"first_name":"Fabian","id":"faff3c84-23f6-11ef-9085-e5187b51c604","full_name":"Kresse, Fabian","last_name":"Kresse"},{"first_name":"Zhengqi","id":"20aa2ae8-f2f1-11ed-bbfa-8205053f1342","orcid":"0000-0002-4993-773X","last_name":"Yu","full_name":"Yu, Zhengqi"},{"full_name":"Lampert, Christoph","last_name":"Lampert","orcid":"0000-0001-8622-7887","first_name":"Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87"},{"last_name":"Henzinger","full_name":"Henzinger, Thomas A","first_name":"Thomas A","id":"40876CD8-F248-11E8-B48F-1D18A9856A87","orcid":"0000-0002-2985-7724"}],"publisher":"ML Research Press","intvolume":"       288","arxiv":1,"type":"conference","date_published":"2025-06-01T00:00:00Z","publication_status":"published","month":"06","OA_place":"publisher","article_processing_charge":"No","OA_type":"diamond","department":[{"_id":"ChLa"},{"_id":"ToHe"}],"conference":{"end_date":"2025-05-30","name":"NeuS: International Conferenceon Neuro-Symbolic Systems","location":"Philadephia, PA, United States","start_date":"2025-05-28"},"external_id":{"arxiv":["2505.19932"]},"article_number":"26","date_created":"2025-09-07T22:01:34Z"},{"file":[{"access_level":"open_access","success":1,"date_updated":"2025-12-16T12:38:29Z","file_size":746612,"relation":"main_file","file_id":"20829","creator":"dernst","file_name":"2025_ICML_Scott.pdf","checksum":"815b32b463023ca21e569c2158745c15","date_created":"2025-12-16T12:38:29Z","content_type":"application/pdf"}],"file_date_updated":"2025-12-16T12:38:29Z","citation":{"chicago":"Scott, Jonathan A, Christoph Lampert, and David Saulpic. “Differentially Private Federated K-Means Clustering with Server-Side Data.” In <i>42nd International Conference on Machine Learning</i>, 267:53757–90. ML Research Press, 2025.","ieee":"J. A. Scott, C. Lampert, and D. Saulpic, “Differentially private federated k-means clustering with server-side data,” in <i>42nd International Conference on Machine Learning</i>, Vancouver, Canada, 2025, vol. 267, pp. 53757–53790.","short":"J.A. Scott, C. Lampert, D. Saulpic, in:, 42nd International Conference on Machine Learning, ML Research Press, 2025, pp. 53757–53790.","apa":"Scott, J. A., Lampert, C., &#38; Saulpic, D. (2025). Differentially private federated k-means clustering with server-side data. In <i>42nd International Conference on Machine Learning</i> (Vol. 267, pp. 53757–53790). Vancouver, Canada: ML Research Press.","mla":"Scott, Jonathan A., et al. “Differentially Private Federated K-Means Clustering with Server-Side Data.” <i>42nd International Conference on Machine Learning</i>, vol. 267, ML Research Press, 2025, pp. 53757–90.","ista":"Scott JA, Lampert C, Saulpic D. 2025. Differentially private federated k-means clustering with server-side data. 42nd International Conference on Machine Learning. ICML: International Conference on Machine Learning, PMLR, vol. 267, 53757–53790.","ama":"Scott JA, Lampert C, Saulpic D. Differentially private federated k-means clustering with server-side data. In: <i>42nd International Conference on Machine Learning</i>. Vol 267. ML Research Press; 2025:53757-53790."},"publication_identifier":{"eissn":["2640-3498"]},"tmp":{"short":"CC BY (4.0)","legal_code_url":"https://creativecommons.org/licenses/by/4.0/legalcode","name":"Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)","image":"/images/cc_by.png"},"oa_version":"Published Version","language":[{"iso":"eng"}],"acknowledgement":"This research was funded in part by the Austrian Science Fund (FWF) [10.55776/COE12] and supported by the Scientific Service Units (SSU) of ISTA through resources provided by Scientific Computing (SciComp).\r\n","publication":"42nd International Conference on Machine Learning","ddc":["000"],"page":"53757-53790","project":[{"_id":"d8f03aaa-b035-11f1-8588-d5147fa879e0","name":"Bilateral Artificial Intelligence (Lampert)","grant_number":"COE12"}],"alternative_title":["PMLR"],"volume":267,"abstract":[{"text":"Clustering is a cornerstone of data analysis that is particularly suited to identifying coherent subgroups or substructures in unlabeled data, as are generated continuously in large amounts these days. However, in many cases traditional clustering methods are not applicable, because data are increasingly being produced and stored in a distributed way, e.g. on edge devices, and privacy concerns prevent it from being transferred to a central server. To address this challenge, we present FedDP-KMeans, a new algorithm for \r\n-means clustering that is fully-federated as well as differentially private. Our approach leverages (potentially small and out-of-distribution) server-side data to overcome the primary challenge of differentially private clustering methods: the need for a good initialization. Combining our initialization with a simple federated DP-Lloyds algorithm we obtain an algorithm that achieves excellent results on synthetic and real-world benchmark tasks. We also provide a theoretical analysis of our method that provides bounds on the convergence speed and cluster identification success.","lang":"eng"}],"day":"01","scopus_import":"1","department":[{"_id":"ChLa"},{"_id":"MoHe"}],"OA_type":"gold","month":"05","OA_place":"publisher","article_processing_charge":"No","publication_status":"published","date_created":"2025-12-14T23:02:05Z","conference":{"name":"ICML: International Conference on Machine Learning","end_date":"2025-07-19","start_date":"2025-07-13","location":"Vancouver, Canada"},"external_id":{"arxiv":["2506.05408"]},"date_published":"2025-05-01T00:00:00Z","type":"conference","related_material":{"record":[{"id":"21198","relation":"dissertation_contains","status":"public"}]},"arxiv":1,"publisher":"ML Research Press","intvolume":"       267","author":[{"id":"e499926b-f6e0-11ea-865d-9c63db0031e8","first_name":"Jonathan A","last_name":"Scott","full_name":"Scott, Jonathan A"},{"first_name":"Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","orcid":"0000-0001-8622-7887","last_name":"Lampert","full_name":"Lampert, Christoph"},{"id":"f8e48cf0-b0ff-11ed-b0e9-b4c35598f964","first_name":"David","full_name":"Saulpic, David","last_name":"Saulpic"}],"oa":1,"corr_author":"1","_id":"20819","title":"Differentially private federated k-means clustering with server-side data","year":"2025","acknowledged_ssus":[{"_id":"ScienComp"}],"date_updated":"2026-09-16T07:26:12Z","has_accepted_license":"1","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","status":"public","quality_controlled":"1"},{"corr_author":"1","oa":1,"author":[{"last_name":"Zakerinia","full_name":"Zakerinia, Hossein","first_name":"Hossein","id":"653bd8b6-f394-11eb-9cf6-c0bbf6cd78d4","orcid":"0009-0007-3977-6462"},{"full_name":"Scott, Jonathan A","last_name":"Scott","first_name":"Jonathan A","id":"e499926b-f6e0-11ea-865d-9c63db0031e8"},{"first_name":"Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","orcid":"0000-0001-8622-7887","last_name":"Lampert","full_name":"Lampert, Christoph"}],"_id":"21207","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","status":"public","doi":"10.48550/ARXIV.2505.15579","year":"2025","date_updated":"2026-09-16T07:26:12Z","title":"Federated learning with unlabeled clients: Personalization can happen in low dimensions","publication_status":"draft","main_file_link":[{"url":"https://doi.org/10.48550/arXiv.2505.15579","open_access":"1"}],"OA_type":"green","department":[{"_id":"ChLa"}],"OA_place":"repository","article_processing_charge":"No","month":"05","article_number":"2505.15579","date_created":"2026-02-10T08:20:59Z","related_material":{"record":[{"id":"21198","relation":"dissertation_contains","status":"public"}]},"type":"preprint","date_published":"2025-05-21T00:00:00Z","fulldoi":"https://doi.org/10.48550/ARXIV.2505.15579","das_tickbox":"1","publication":"arXiv","day":"21","abstract":[{"lang":"eng","text":"Personalized federated learning has emerged as a popular approach to training on devices holding statistically heterogeneous data, known as clients. However, most existing approaches require a client to have labeled data for training or finetuning in order to obtain their own personalized model. In this paper we address this by proposing FLowDUP, a novel method that is able to generate a personalized model using only a forward pass with unlabeled data. The generated model parameters reside in a low-dimensional subspace, enabling efficient communication and computation. FLowDUP's learning objective is theoretically motivated by our new transductive multi-task PAC-Bayesian generalization bound, that provides performance guarantees for unlabeled clients. The objective is structured in such a way that it allows both clients with labeled data and clients with only unlabeled data to contribute to the training process. To supplement our theoretical results we carry out a thorough experimental evaluation of FLowDUP, demonstrating strong empirical performance on a range of datasets with differing sorts of statistically heterogeneous clients. Through numerous ablation studies, we test the efficacy of the individual components of the method."}],"citation":{"ista":"Zakerinia H, Scott JA, Lampert C. Federated learning with unlabeled clients: Personalization can happen in low dimensions. arXiv, 2505.15579.","mla":"Zakerinia, Hossein, et al. “Federated Learning with Unlabeled Clients: Personalization Can Happen in Low Dimensions.” <i>ArXiv</i>, 2505.15579, doi:<a href=\"https://doi.org/10.48550/ARXIV.2505.15579\">10.48550/ARXIV.2505.15579</a>.","ama":"Zakerinia H, Scott JA, Lampert C. Federated learning with unlabeled clients: Personalization can happen in low dimensions. <i>arXiv</i>. doi:<a href=\"https://doi.org/10.48550/ARXIV.2505.15579\">10.48550/ARXIV.2505.15579</a>","chicago":"Zakerinia, Hossein, Jonathan A Scott, and Christoph Lampert. “Federated Learning with Unlabeled Clients: Personalization Can Happen in Low Dimensions.” <i>ArXiv</i>, n.d. <a href=\"https://doi.org/10.48550/ARXIV.2505.15579\">https://doi.org/10.48550/ARXIV.2505.15579</a>.","ieee":"H. Zakerinia, J. A. Scott, and C. Lampert, “Federated learning with unlabeled clients: Personalization can happen in low dimensions,” <i>arXiv</i>. .","apa":"Zakerinia, H., Scott, J. A., &#38; Lampert, C. (n.d.). Federated learning with unlabeled clients: Personalization can happen in low dimensions. <i>arXiv</i>. <a href=\"https://doi.org/10.48550/ARXIV.2505.15579\">https://doi.org/10.48550/ARXIV.2505.15579</a>","short":"H. Zakerinia, J.A. Scott, C. Lampert, ArXiv (n.d.)."},"tmp":{"short":"CC BY (4.0)","legal_code_url":"https://creativecommons.org/licenses/by/4.0/legalcode","name":"Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)","image":"/images/cc_by.png"},"oa_version":"Preprint","language":[{"iso":"eng"}]},{"conference":{"name":"NeurIPS: Neural Information Processing Systems","location":"San Diego, CA, United States"},"date_created":"2026-09-06T22:02:00Z","publication_status":"published","main_file_link":[{"url":"https://doi.org/10.52202/085713-1924","open_access":"1"}],"OA_type":"free access","department":[{"_id":"GradSch"},{"_id":"ChLa"}],"month":"12","article_processing_charge":"No","OA_place":"publisher","fulldoi":"https://doi.org/10.52202/085713-1924","das_tickbox":"0","date_published":"2025-12-02T00:00:00Z","type":"conference","author":[{"first_name":"Nikita","id":"4b14526e-14d2-11ed-ba64-c14c9553d137","full_name":"Kalinin, Nikita","last_name":"Kalinin"},{"first_name":"Jalaj","last_name":"Upadhyay","full_name":"Upadhyay, Jalaj"},{"orcid":"0000-0001-8622-7887","first_name":"Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","full_name":"Lampert, Christoph","last_name":"Lampert"}],"oa":1,"corr_author":"1","publisher":"Neural Information Processing Systems Foundation","intvolume":"        38","status":"public","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","quality_controlled":"1","doi":"10.52202/085713-1924","year":"2025","title":"Continual release moment estimation with differential privacy","date_updated":"2026-09-16T07:40:56Z","_id":"22828","publication_identifier":{"issn":["1049-5258"],"isbn":["9798331338275"]},"citation":{"apa":"Kalinin, N., Upadhyay, J., &#38; Lampert, C. (2025). Continual release moment estimation with differential privacy. In <i>Advances in Neural Information Processing Systems</i> (Vol. 38, pp. 64002–64047). San Diego, CA, United States: Neural Information Processing Systems Foundation. <a href=\"https://doi.org/10.52202/085713-1924\">https://doi.org/10.52202/085713-1924</a>","short":"N. Kalinin, J. Upadhyay, C. Lampert, in:, Advances in Neural Information Processing Systems, Neural Information Processing Systems Foundation, 2025, pp. 64002–64047.","ieee":"N. Kalinin, J. Upadhyay, and C. Lampert, “Continual release moment estimation with differential privacy,” in <i>Advances in Neural Information Processing Systems</i>, San Diego, CA, United States, 2025, vol. 38, pp. 64002–64047.","chicago":"Kalinin, Nikita, Jalaj Upadhyay, and Christoph Lampert. “Continual Release Moment Estimation with Differential Privacy.” In <i>Advances in Neural Information Processing Systems</i>, 38:64002–47. Neural Information Processing Systems Foundation, 2025. <a href=\"https://doi.org/10.52202/085713-1924\">https://doi.org/10.52202/085713-1924</a>.","ama":"Kalinin N, Upadhyay J, Lampert C. Continual release moment estimation with differential privacy. In: <i>Advances in Neural Information Processing Systems</i>. Vol 38. Neural Information Processing Systems Foundation; 2025:64002-64047. doi:<a href=\"https://doi.org/10.52202/085713-1924\">10.52202/085713-1924</a>","ista":"Kalinin N, Upadhyay J, Lampert C. 2025. Continual release moment estimation with differential privacy. Advances in Neural Information Processing Systems. NeurIPS: Neural Information Processing Systems, Advances in Neural Information Processing Systems, vol. 38, 64002–64047.","mla":"Kalinin, Nikita, et al. “Continual Release Moment Estimation with Differential Privacy.” <i>Advances in Neural Information Processing Systems</i>, vol. 38, Neural Information Processing Systems Foundation, 2025, pp. 64002–47, doi:<a href=\"https://doi.org/10.52202/085713-1924\">10.52202/085713-1924</a>."},"acknowledgement":"We thank Monika Henzinger for her valuable feedback and insightful discussions on earlier versions\r\nof this draft. We are also grateful to Mher Safaryan for his contributions to discussions on DP-Adam.\r\nAdditionally, we thank Ryan McKenna for suggesting Joint Clipping as a baseline.\r\nJalaj Upadhyay’s research was funded by the Rutgers Decanal Grant no. 302918, NSF CNS 2433628,\r\nGoogle Research Scholar Award, and Google Seed Fund Grant. A part of this work was done while\r\nvisiting the Institute of Science and Technology Austria (ISTA).\r\nNikita Kalinin’s research was funded in part by the Austrian Science Fund (FWF) [10.55776/COE12]","language":[{"iso":"eng"}],"oa_version":"Published Version","page":"64002-64047","publication":"Advances in Neural Information Processing Systems","day":"02","scopus_import":"1","alternative_title":["Advances in Neural Information Processing Systems"],"supplementarymaterial":"no","volume":38,"abstract":[{"text":"We propose Joint Moment Estimation (JME), a method for continually and privately estimating both the first and second moments of a data stream with reduced noise compared to naive approaches. JME supports the matrix mechanism and exploits a joint sensitivity analysis to identify a privacy regime in which the second-moment estimation incurs no additional privacy cost, thereby improving accuracy while maintaining privacy. We demonstrate JME’s effectiveness in two applications: estimating the running mean and covariance matrix for Gaussian density estimation and model training with DP-Adam.","lang":"eng"}],"project":[{"_id":"d8f03aaa-b035-11f1-8588-d5147fa879e0","grant_number":"COE12","name":"Bilateral Artificial Intelligence (Lampert)"}],"researchdata_availability":"no"},{"fulldoi":"https://doi.org/10.1007/s00521-024-10616-1","das_tickbox":"1","arxiv":1,"date_published":"2025-10-01T00:00:00Z","type":"journal_article","dataavailabilitystatement":"Our manuscript does not include or refer to any data sources.","external_id":{"arxiv":["2208.13499"]},"date_created":"2023-02-20T08:23:06Z","publication_status":"published","OA_type":"hybrid","department":[{"_id":"ChLa"}],"OA_place":"publisher","article_processing_charge":"Yes (via OA deal)","month":"10","quality_controlled":"1","has_accepted_license":"1","status":"public","user_id":"317138e5-6ab7-11ef-aa6d-ffef3953e345","doi":"10.1007/s00521-024-10616-1","date_updated":"2026-10-01T09:29:43Z","year":"2025","title":"Generalization in multi-objective machine learning","_id":"12662","PlanS_conform":"1","corr_author":"1","oa":1,"author":[{"first_name":"Peter","id":"d64d6a8d-eb8e-11eb-b029-96fd216dec3c","full_name":"Súkeník, Peter","last_name":"Súkeník"},{"first_name":"Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","orcid":"0000-0001-8622-7887","last_name":"Lampert","full_name":"Lampert, Christoph"}],"publisher":"Springer Nature","intvolume":"        37","acknowledgement":"Open access funding provided by Institute of Science and Technology (IST Austria).","language":[{"iso":"eng"}],"tmp":{"short":"CC BY (4.0)","legal_code_url":"https://creativecommons.org/licenses/by/4.0/legalcode","name":"Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)","image":"/images/cc_by.png"},"oa_version":"Published Version","publication_identifier":{"eissn":["1433-3058"],"issn":["0941-0643"]},"citation":{"ieee":"P. Súkeník and C. Lampert, “Generalization in multi-objective machine learning,” <i>Neural Computing and Applications</i>, vol. 37. Springer Nature, pp. 24669–24683, 2025.","chicago":"Súkeník, Peter, and Christoph Lampert. “Generalization in Multi-Objective Machine Learning.” <i>Neural Computing and Applications</i>. Springer Nature, 2025. <a href=\"https://doi.org/10.1007/s00521-024-10616-1\">https://doi.org/10.1007/s00521-024-10616-1</a>.","short":"P. Súkeník, C. Lampert, Neural Computing and Applications 37 (2025) 24669–24683.","apa":"Súkeník, P., &#38; Lampert, C. (2025). Generalization in multi-objective machine learning. <i>Neural Computing and Applications</i>. Springer Nature. <a href=\"https://doi.org/10.1007/s00521-024-10616-1\">https://doi.org/10.1007/s00521-024-10616-1</a>","mla":"Súkeník, Peter, and Christoph Lampert. “Generalization in Multi-Objective Machine Learning.” <i>Neural Computing and Applications</i>, vol. 37, Springer Nature, 2025, pp. 24669–24683, doi:<a href=\"https://doi.org/10.1007/s00521-024-10616-1\">10.1007/s00521-024-10616-1</a>.","ista":"Súkeník P, Lampert C. 2025. Generalization in multi-objective machine learning. Neural Computing and Applications. 37, 24669–24683.","ama":"Súkeník P, Lampert C. Generalization in multi-objective machine learning. <i>Neural Computing and Applications</i>. 2025;37:24669–24683. doi:<a href=\"https://doi.org/10.1007/s00521-024-10616-1\">10.1007/s00521-024-10616-1</a>"},"file_date_updated":"2025-12-30T06:39:11Z","file":[{"date_created":"2025-12-30T06:39:11Z","checksum":"61ad4591aee16b1e02daf6c164321a42","file_name":"2025_NeuralCompApplic_Sukenik.pdf","content_type":"application/pdf","date_updated":"2025-12-30T06:39:11Z","success":1,"access_level":"open_access","creator":"dernst","relation":"main_file","file_id":"20877","file_size":500213}],"article_type":"original","day":"01","scopus_import":"1","supplementarymaterial":"no","volume":37,"abstract":[{"lang":"eng","text":"Modern machine learning tasks often require considering not just one but multiple objectives. For example, besides the prediction quality, this could be the efficiency, robustness or fairness of the learned models, or any of their combinations. Multi-objective learning offers a natural framework for handling such problems without having to commit to early trade-offs. Surprisingly, statistical learning theory so far offers almost no insight into the generalization properties of multi-objective learning. In this work, we make first steps to fill this gap: We establish foundational generalization bounds for the multi-objective setting as well as generalization and excess bounds for learning with scalarizations. We also provide the first theoretical analysis of the relation between the Pareto-optimal sets of the true objectives and the Pareto-optimal sets of their empirical approximations from training data. In particular, we show a surprising asymmetry: All Pareto-optimal solutions can be approximated by empirically Pareto-optimal ones, but not vice versa."}],"researchdata_availability":"not applicable","page":"24669–24683","publication":"Neural Computing and Applications","ddc":["004"]},{"publisher":"Institute of Science and Technology Austria","corr_author":"1","oa":1,"author":[{"full_name":"Prach, Bernd","last_name":"Prach","id":"2D561D42-C427-11E9-89B4-9C1AE6697425","first_name":"Bernd"}],"degree_awarded":"PhD","_id":"19759","doi":"10.15479/10.15479/at-ista-19759","title":"Robust image classification with 1-Lipschitz networks","year":"2025","date_updated":"2026-10-02T11:16:04Z","user_id":"8b945eb4-e2f2-11eb-945a-df72226e66a9","status":"public","has_accepted_license":"1","department":[{"_id":"GradSch"},{"_id":"ChLa"}],"article_processing_charge":"No","month":"05","OA_place":"publisher","publication_status":"published","date_created":"2025-05-28T16:20:48Z","supervisor":[{"last_name":"Lampert","full_name":"Lampert, Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","first_name":"Christoph","orcid":"0000-0001-8622-7887"}],"type":"dissertation","date_published":"2025-05-30T00:00:00Z","related_material":{"record":[{"status":"public","relation":"part_of_dissertation","id":"15039"},{"id":"11839","relation":"part_of_dissertation","status":"public"},{"id":"18874","relation":"part_of_dissertation","status":"public"},{"status":"public","relation":"part_of_dissertation","id":"17426"}]},"fulldoi":"https://doi.org/10.15479/10.15479/at-ista-19759","doi_confirm":"1","ddc":["000"],"page":"84","publisher_comment":"In reference to IEEE copyrighted material which is used with permission in this thesis, the IEEE does not\r\nendorse any of ISTA's products or services. Internal or personal use of this\r\nmaterial is permitted. If interested in reprinting/republishing IEEE copyrighted material for advertising or promotional\r\npurposes or for creating new collective works for resale or redistribution, please go to\r\nhttp://www.ieee.org/publications_standards/publications/rights/rights_link.html to learn how to obtain a License from\r\nRightsLink.  If applicable, University Microfilms and/or ProQuest Library, or the Archives of Canada may supply single copies of the dissertation.","alternative_title":["ISTA Thesis"],"abstract":[{"lang":"eng","text":"Despite generating remarkable results in various computer vision tasks, deep learning comes\r\nwith some surprising shortcomings. For example, tiny perturbations, often imperceptible to\r\nthe human eye, can completely change the predictions of image classifiers. Despite a decade\r\nof research, the field has made limited progress in developing image classifiers that are both\r\naccurate and robust. This thesis aims to address this gap.\r\nAs our first contribution, we aim to simplify the process of training certifiably robust image\r\nclassifiers. We do this by designing a convolutional layer that does not require executing an\r\niterative procedure in every forward pass, but relies on an explicit bound instead. We also\r\npropose a loss function that allows optimizing for a particular margin more precisely.\r\nNext, we provide an overview and comparison of various methods that create robust image\r\nclassifiers by constraining the Lipschitz constant. This is important since generally longer\r\ntraining times and more parameters improve the performance of robust classifiers, making it\r\nchallenging to determine the most practical and effective methods from existing literature.\r\nIn 1-Lipschitz classification, the performance of current methods is still much worse than what\r\nwe expect on the simple tasks we consider. Therefore, we next investigate potential causes of\r\nthis shortcoming. We first consider the role of the activation function. We prove a theoretical\r\nshortcoming of the commonly used activation function, and provide an alternative without it.\r\nHowever this theoretical improvement does barely translate to the empirical performance of\r\nrobust classifiers, suggesting a different bottleneck.\r\nTherefore, in the final chapter, we study how the performance depends on the amount of\r\ntraining data. We prove that in the worst case, we might require far more data to train a\r\nrobust classifier compared to a normal one. We furthermore find that the amount of training\r\ndata is a key determinant of the performance current methods achieve on popular datasets.\r\nAdditionally, we show that linear subspaces exist with tiny data variance, and yet we can\r\nstill train very accurate classifiers after projecting into those subspaces. This shows that on\r\nthe datasets considered, enforcing robustness in classification makes the task strictly more\r\nchallenging.\r\n\r\n"}],"day":"30","file":[{"content_type":"application/pdf","date_created":"2025-06-10T18:11:05Z","checksum":"e5108e759014e2a9020c973c778fafc9","file_name":"ThesisFinal.pdf","creator":"bprach","relation":"main_file","file_id":"19829","file_size":3578077,"date_updated":"2025-06-10T18:11:05Z","access_level":"open_access"},{"access_level":"closed","date_updated":"2025-06-10T18:14:03Z","relation":"source_file","file_id":"19830","file_size":74894357,"creator":"bprach","checksum":"51bf6c11fb6d8a9f8010b458c600a83f","file_name":"ThesisFinal.zip","date_created":"2025-06-10T18:14:03Z","content_type":"application/x-zip-compressed"}],"file_date_updated":"2025-06-10T18:14:03Z","citation":{"short":"B. Prach, Robust Image Classification with 1-Lipschitz Networks, Institute of Science and Technology Austria, 2025.","apa":"Prach, B. (2025). <i>Robust image classification with 1-Lipschitz networks</i>. Institute of Science and Technology Austria. <a href=\"https://doi.org/10.15479/10.15479/at-ista-19759\">https://doi.org/10.15479/10.15479/at-ista-19759</a>","chicago":"Prach, Bernd. “Robust Image Classification with 1-Lipschitz Networks.” Institute of Science and Technology Austria, 2025. <a href=\"https://doi.org/10.15479/10.15479/at-ista-19759\">https://doi.org/10.15479/10.15479/at-ista-19759</a>.","ieee":"B. Prach, “Robust image classification with 1-Lipschitz networks,” Institute of Science and Technology Austria, 2025.","ama":"Prach B. Robust image classification with 1-Lipschitz networks. 2025. doi:<a href=\"https://doi.org/10.15479/10.15479/at-ista-19759\">10.15479/10.15479/at-ista-19759</a>","mla":"Prach, Bernd. <i>Robust Image Classification with 1-Lipschitz Networks</i>. Institute of Science and Technology Austria, 2025, doi:<a href=\"https://doi.org/10.15479/10.15479/at-ista-19759\">10.15479/10.15479/at-ista-19759</a>.","ista":"Prach B. 2025. Robust image classification with 1-Lipschitz networks. Institute of Science and Technology Austria."},"publication_identifier":{"issn":["2663-337X"]},"oa_version":"Published Version","language":[{"iso":"eng"}]},{"main_file_link":[{"url":"https://doi.org/10.48550/arXiv.2412.04245","open_access":"1"}],"publication_status":"published","department":[{"_id":"ChLa"}],"OA_type":"green","OA_place":"repository","article_processing_charge":"No","month":"06","external_id":{"arxiv":["2412.04245"]},"conference":{"start_date":"2025-06-11","location":"Nashville, TN, United States","end_date":"2025-06-12","name":"CVPR: Conference on Computer Vision and Pattern Recognition"},"date_created":"2025-10-12T22:01:26Z","related_material":{"record":[{"status":"public","id":"18874","relation":"earlier_version"}]},"arxiv":1,"type":"conference","date_published":"2025-06-15T00:00:00Z","fulldoi":"https://doi.org/10.1109/CVPRW67362.2025.00071","author":[{"id":"2D561D42-C427-11E9-89B4-9C1AE6697425","first_name":"Bernd","last_name":"Prach","full_name":"Prach, Bernd"},{"full_name":"Lampert, Christoph","last_name":"Lampert","orcid":"0000-0001-8622-7887","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","first_name":"Christoph"}],"corr_author":"1","oa":1,"publisher":"IEEE","_id":"20455","status":"public","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","quality_controlled":"1","doi":"10.1109/CVPRW67362.2025.00071","title":"Intriguing properties of robust classification","date_updated":"2026-10-02T11:15:42Z","year":"2025","publication_identifier":{"issn":["2160-7508"],"isbn":["9798331599942"],"eissn":["2160-7516"]},"citation":{"ama":"Prach B, Lampert C. Intriguing properties of robust classification. In: <i>2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops</i>. IEEE; 2025:660-669. doi:<a href=\"https://doi.org/10.1109/CVPRW67362.2025.00071\">10.1109/CVPRW67362.2025.00071</a>","mla":"Prach, Bernd, and Christoph Lampert. “Intriguing Properties of Robust Classification.” <i>2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops</i>, IEEE, 2025, pp. 660–69, doi:<a href=\"https://doi.org/10.1109/CVPRW67362.2025.00071\">10.1109/CVPRW67362.2025.00071</a>.","ista":"Prach B, Lampert C. 2025. Intriguing properties of robust classification. 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. CVPR: Conference on Computer Vision and Pattern Recognition, 660–669.","short":"B. Prach, C. Lampert, in:, 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, IEEE, 2025, pp. 660–669.","apa":"Prach, B., &#38; Lampert, C. (2025). Intriguing properties of robust classification. In <i>2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops</i> (pp. 660–669). Nashville, TN, United States: IEEE. <a href=\"https://doi.org/10.1109/CVPRW67362.2025.00071\">https://doi.org/10.1109/CVPRW67362.2025.00071</a>","chicago":"Prach, Bernd, and Christoph Lampert. “Intriguing Properties of Robust Classification.” In <i>2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops</i>, 660–69. IEEE, 2025. <a href=\"https://doi.org/10.1109/CVPRW67362.2025.00071\">https://doi.org/10.1109/CVPRW67362.2025.00071</a>.","ieee":"B. Prach and C. Lampert, “Intriguing properties of robust classification,” in <i>2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops</i>, Nashville, TN, United States, 2025, pp. 660–669."},"oa_version":"Preprint","language":[{"iso":"eng"}],"page":"660-669","publication":"2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops","day":"15","scopus_import":"1","abstract":[{"text":"Despite extensive research since the community learned about adversarial examples 10 years ago, we still do not know how to train high-accuracy classifiers that are guaranteed to be robust to small perturbations of their inputs. Previous works often argued that this might be because no classifier exists that is robust and accurate at the same time. However, in computer vision this assumption does not match reality where humans are usually accurate and robust on most tasks of interest. We offer an alternative explanation and show that in certain settings robust generalization is only possible with unrealistically large amounts of data. Specifically, we find a setting where a robust classifier exists, it is easy to learn an accurate classifier, yet it requires an exponential amount of data to learn a robust classifier. Based on this theoretical result, we evaluate the influence of the amount of training data on datasets such as CIFAR10. Our findings indicate that the the amount of training data is the main factor determining the robust performance. Furthermore we show that that there are low magnitude directions in the data which are useful for non-robust generalization but are not available for robust classifiers. This implies that robust classification is a strictly harder tasks than normal classification, thereby providing an explanation why robust classification requires more data.","lang":"eng"}]},{"oa_version":"Published Version","language":[{"iso":"eng"}],"acknowledgement":"This research was funded in whole by the Austrian Science Fund (FWF) 10.55776/COE12. I thank\r\nAlberto Naibo for the discussion and his invitation to \"Logique, droit, IA : penser les algorithmes\"\r\ncolloquium. I thank Simone Bombari for the discussions. I thank EurIPS for the active poster session.","citation":{"mla":"Cyffers, Edwige Audrey Lucienne. “Setting ε Is Not the Issue in Differential Privacy.” <i>39th Conference on Neural Information Processing Systems</i>, vol. 38, Neural Information Processing Systems Foundation, 2025, pp. 159016–28, doi:<a href=\"https://doi.org/10.52202/085713-4798\">10.52202/085713-4798</a>.","ista":"Cyffers EAL. 2025. Setting ε is not the issue in differential privacy. 39th Conference on Neural Information Processing Systems. NeurIPS: Neural Information Processing Systems, Advances in Neural Information Processing Systems, vol. 38, 159016–159028.","ama":"Cyffers EAL. Setting ε is not the issue in differential privacy. In: <i>39th Conference on Neural Information Processing Systems</i>. Vol 38. Neural Information Processing Systems Foundation; 2025:159016-159028. doi:<a href=\"https://doi.org/10.52202/085713-4798\">10.52202/085713-4798</a>","chicago":"Cyffers, Edwige Audrey Lucienne. “Setting ε Is Not the Issue in Differential Privacy.” In <i>39th Conference on Neural Information Processing Systems</i>, 38:159016–28. Neural Information Processing Systems Foundation, 2025. <a href=\"https://doi.org/10.52202/085713-4798\">https://doi.org/10.52202/085713-4798</a>.","ieee":"E. A. L. Cyffers, “Setting ε is not the issue in differential privacy,” in <i>39th Conference on Neural Information Processing Systems</i>, San Diego, CA, United States, 2025, vol. 38, pp. 159016–159028.","short":"E.A.L. Cyffers, in:, 39th Conference on Neural Information Processing Systems, Neural Information Processing Systems Foundation, 2025, pp. 159016–159028.","apa":"Cyffers, E. A. L. (2025). Setting ε is not the issue in differential privacy. In <i>39th Conference on Neural Information Processing Systems</i> (Vol. 38, pp. 159016–159028). San Diego, CA, United States: Neural Information Processing Systems Foundation. <a href=\"https://doi.org/10.52202/085713-4798\">https://doi.org/10.52202/085713-4798</a>"},"publication_identifier":{"eissn":["1049-5258"],"isbn":["9798331338275"]},"researchdata_availability":"no","project":[{"_id":"d8f03aaa-b035-11f1-8588-d5147fa879e0","grant_number":"COE12","name":"Bilateral Artificial Intelligence (Lampert)"}],"volume":38,"alternative_title":["Advances in Neural Information Processing Systems"],"supplementarymaterial":"yes","abstract":[{"lang":"eng","text":"This position paper argues that setting the privacy budget in differential privacy should not be viewed as an important limitation of differential privacy compared to alternative methods for privacy-preserving machine learning. The so-called problem of interpreting the privacy budget is often presented as a major hindrance to the wider adoption of differential privacy in real-world deployments and is sometimes used to promote alternative mitigation techniques for data protection. We believe this misleads decision-makers into choosing unsafe methods. We argue that the difficulty in interpreting privacy budgets does not stem from the definition of differential privacy itself, but from the intrinsic difficulty of estimating privacy risks in context, a challenge that any rigorous method for privacy risk assessment face. Moreover, we claim that any sound method for estimating privacy risks should, given the current state of research, be expressible within the differential privacy framework or justify why it cannot."}],"day":"02","scopus_import":"1","publication":"39th Conference on Neural Information Processing Systems","page":"159016-159028","type":"conference","date_published":"2025-12-02T00:00:00Z","das_tickbox":"0","fulldoi":"https://doi.org/10.52202/085713-4798","OA_type":"free access","department":[{"_id":"ChLa"}],"month":"12","OA_place":"publisher","article_processing_charge":"No","publication_status":"published","main_file_link":[{"open_access":"1","url":"https://doi.org/10.52202/085713-4798"}],"date_created":"2026-09-06T22:01:59Z","conference":{"end_date":"2025-12-07","name":"NeurIPS: Neural Information Processing Systems","start_date":"2025-12-02","location":"San Diego, CA, United States"},"_id":"22823","doi":"10.52202/085713-4798","title":"Setting ε is not the issue in differential privacy","year":"2025","date_updated":"2026-10-15T09:34:50Z","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","quality_controlled":"1","status":"public","intvolume":"        38","publisher":"Neural Information Processing Systems Foundation","oa":1,"corr_author":"1","author":[{"full_name":"Cyffers, Edwige Audrey Lucienne","last_name":"Cyffers","id":"20d4c299-977a-11ef-ae55-98b15ac64a57","first_name":"Edwige Audrey Lucienne"}]},{"_id":"19408","year":"2024","date_updated":"2025-03-20T09:21:02Z","title":"Continual learning: Applications and the road forward","quality_controlled":"1","status":"public","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","has_accepted_license":"1","intvolume":"      2024","publisher":"Transactions on Machine Learning Research","oa":1,"author":[{"first_name":"Eli","last_name":"Verwimp","full_name":"Verwimp, Eli"},{"last_name":"Aljundi","full_name":"Aljundi, Rahaf","first_name":"Rahaf"},{"full_name":"Ben-David, Shai","last_name":"Ben-David","first_name":"Shai"},{"last_name":"Bethge","full_name":"Bethge, Matthias","first_name":"Matthias"},{"full_name":"Cossu, Andrea","last_name":"Cossu","first_name":"Andrea"},{"first_name":"Alexander","full_name":"Gepperth, Alexander","last_name":"Gepperth"},{"first_name":"Tyler L.","last_name":"Hayes","full_name":"Hayes, Tyler L."},{"last_name":"Hüllermeier","full_name":"Hüllermeier, Eyke","first_name":"Eyke"},{"last_name":"Kanan","full_name":"Kanan, Christopher","first_name":"Christopher"},{"first_name":"Dhireesha","full_name":"Kudithipudi, Dhireesha","last_name":"Kudithipudi"},{"last_name":"Lampert","full_name":"Lampert, Christoph","first_name":"Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","orcid":"0000-0001-8622-7887"},{"first_name":"Martin","last_name":"Mundt","full_name":"Mundt, Martin"},{"full_name":"Pascanu, Razvan","last_name":"Pascanu","first_name":"Razvan"},{"first_name":"Adrian","full_name":"Popescu, Adrian","last_name":"Popescu"},{"first_name":"Andreas S.","full_name":"Tolias, Andreas S.","last_name":"Tolias"},{"last_name":"Van De Weijer","full_name":"Van De Weijer, Joost","first_name":"Joost"},{"last_name":"Liu","full_name":"Liu, Bing","first_name":"Bing"},{"first_name":"Vincenzo","full_name":"Lomonaco, Vincenzo","last_name":"Lomonaco"},{"full_name":"Tuytelaars, Tinne","last_name":"Tuytelaars","first_name":"Tinne"},{"first_name":"Gido M.","last_name":"Van De Ven","full_name":"Van De Ven, Gido M."}],"date_published":"2024-04-12T00:00:00Z","type":"journal_article","arxiv":1,"department":[{"_id":"ChLa"}],"OA_type":"diamond","OA_place":"publisher","article_processing_charge":"No","month":"04","publication_status":"published","date_created":"2025-03-16T23:01:25Z","external_id":{"arxiv":["2311.11908"]},"volume":2024,"alternative_title":["TMLR"],"abstract":[{"lang":"eng","text":"Continual learning is a subfield of machine learning, which aims to allow machine learning models to continuously learn on new data, by accumulating knowledge without forgetting what was learned in the past. In this work, we take a step back, and ask: \"Why should one care about continual learning in the first place?\". We set the stage by examining recent continual learning papers published at four major machine learning conferences, and show that memory-constrained settings dominate the field. Then, we discuss five open problems in machine learning, and even though they might seem unrelated to continual learning at first sight, we show that continual learning will inevitably be part of their solution. These problems are model editing, personalization and specialization, on-device learning, faster (re-)training and reinforcement learning. Finally, by comparing the desiderata from these unsolved problems and the current assumptions in continual learning, we highlight and discuss four future directions for continual learning research. We hope that this work offers an interesting perspective on the future of continual learning, while displaying its potential value and the paths we have to pursue in order to make it successful. This work is the result of the many discussions the authors had at the Dagstuhl seminar on Deep Continual Learning, in March 2023."}],"article_type":"original","day":"12","scopus_import":"1","publication":"Transactions on Machine Learning Research","ddc":["000"],"tmp":{"short":"CC BY (4.0)","legal_code_url":"https://creativecommons.org/licenses/by/4.0/legalcode","name":"Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)","image":"/images/cc_by.png"},"oa_version":"Published Version","language":[{"iso":"eng"}],"file":[{"creator":"dernst","relation":"main_file","file_id":"19426","file_size":1367966,"date_updated":"2025-03-20T09:02:18Z","success":1,"access_level":"open_access","content_type":"application/pdf","date_created":"2025-03-20T09:02:18Z","checksum":"0714e12f7423cd098976ed9974561155","file_name":"2024_TMLR_Verwimp.pdf"}],"file_date_updated":"2025-03-20T09:02:18Z","citation":{"mla":"Verwimp, Eli, et al. “Continual Learning: Applications and the Road Forward.” <i>Transactions on Machine Learning Research</i>, vol. 2024, Transactions on Machine Learning Research, 2024.","ista":"Verwimp E, Aljundi R, Ben-David S, Bethge M, Cossu A, Gepperth A, Hayes TL, Hüllermeier E, Kanan C, Kudithipudi D, Lampert C, Mundt M, Pascanu R, Popescu A, Tolias AS, Van De Weijer J, Liu B, Lomonaco V, Tuytelaars T, Van De Ven GM. 2024. Continual learning: Applications and the road forward. Transactions on Machine Learning Research. 2024.","ama":"Verwimp E, Aljundi R, Ben-David S, et al. Continual learning: Applications and the road forward. <i>Transactions on Machine Learning Research</i>. 2024;2024.","ieee":"E. Verwimp <i>et al.</i>, “Continual learning: Applications and the road forward,” <i>Transactions on Machine Learning Research</i>, vol. 2024. Transactions on Machine Learning Research, 2024.","chicago":"Verwimp, Eli, Rahaf Aljundi, Shai Ben-David, Matthias Bethge, Andrea Cossu, Alexander Gepperth, Tyler L. Hayes, et al. “Continual Learning: Applications and the Road Forward.” <i>Transactions on Machine Learning Research</i>. Transactions on Machine Learning Research, 2024.","short":"E. Verwimp, R. Aljundi, S. Ben-David, M. Bethge, A. Cossu, A. Gepperth, T.L. Hayes, E. Hüllermeier, C. Kanan, D. Kudithipudi, C. Lampert, M. Mundt, R. Pascanu, A. Popescu, A.S. Tolias, J. Van De Weijer, B. Liu, V. Lomonaco, T. Tuytelaars, G.M. Van De Ven, Transactions on Machine Learning Research 2024 (2024).","apa":"Verwimp, E., Aljundi, R., Ben-David, S., Bethge, M., Cossu, A., Gepperth, A., … Van De Ven, G. M. (2024). Continual learning: Applications and the road forward. <i>Transactions on Machine Learning Research</i>. Transactions on Machine Learning Research."},"publication_identifier":{"eissn":["2835-8856"]}},{"author":[{"first_name":"Egor","id":"05162b19-1340-11ed-8f02-fa94e0e8c3bc","last_name":"Zverev","full_name":"Zverev, Egor"},{"full_name":"Abdelnabi, Sahar","last_name":"Abdelnabi","first_name":"Sahar"},{"last_name":"Tabesh","full_name":"Tabesh, Soroush","first_name":"Soroush","id":"06000900-6068-11ef-8d61-c2472ef2e752","orcid":"0009-0003-4119-6281"},{"first_name":"Mario","full_name":"Fritz, Mario","last_name":"Fritz"},{"orcid":"0000-0001-8622-7887","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","first_name":"Christoph","full_name":"Lampert, Christoph","last_name":"Lampert"}],"corr_author":"1","oa":1,"doi":"10.48550/arXiv.2403.06833","title":"Can LLMs separate instructions from data? And what do we even mean by that?","date_updated":"2026-08-13T07:26:54Z","acknowledged_ssus":[{"_id":"ScienComp"}],"year":"2024","status":"public","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","has_accepted_license":"1","_id":"19063","date_created":"2025-02-20T10:13:42Z","article_number":"2403.06833","external_id":{"arxiv":["2403.06833"]},"department":[{"_id":"GradSch"},{"_id":"ChLa"}],"OA_type":"green","OA_place":"repository","month":"03","article_processing_charge":"No","publication_status":"submitted","fulldoi":"https://doi.org/10.48550/arXiv.2403.06833","date_published":"2024-03-01T00:00:00Z","type":"preprint","related_material":{"link":[{"url":" https://github.com/egozverev/Shold-It-Be-Executed-Or-Processed","relation":"software"}]},"arxiv":1,"ddc":["000"],"publication":"arXiv","abstract":[{"text":"Instruction-tuned Large Language Models (LLMs) show impressive results in numerous practical applications, but they lack essential safety features that are common in other areas of computer science, particularly an explicit separation of instructions and data. This makes them vulnerable to manipulations such as indirect prompt injections and generally unsuitable for safety-critical tasks. Surprisingly, there is currently no established definition or benchmark to quantify this phenomenon. In this work, we close this gap by introducing a formal measure for instruction-data separation and an empirical variant that is calculable from a model's outputs. We also present a new dataset, SEP, that allows estimating the measure for real-world models. Our results on various LLMs show that the problem of instruction-data separation is real: all models fail to achieve high separation, and canonical mitigation techniques, such as prompt engineering and fine-tuning, either fail to substantially improve separation or reduce model utility. The source code and SEP dataset are openly accessible at https://github.com/egozverev/Shold-It-Be-Executed-Or-Processed.\r\n","lang":"eng"}],"day":"01","citation":{"ama":"Zverev E, Abdelnabi S, Tabesh S, Fritz M, Lampert C. Can LLMs separate instructions from data? And what do we even mean by that? <i>arXiv</i>. doi:<a href=\"https://doi.org/10.48550/arXiv.2403.06833\">10.48550/arXiv.2403.06833</a>","mla":"Zverev, Egor, et al. “Can LLMs Separate Instructions from Data? And What Do We Even Mean by That?” <i>ArXiv</i>, 2403.06833, doi:<a href=\"https://doi.org/10.48550/arXiv.2403.06833\">10.48550/arXiv.2403.06833</a>.","ista":"Zverev E, Abdelnabi S, Tabesh S, Fritz M, Lampert C. Can LLMs separate instructions from data? And what do we even mean by that? arXiv, 2403.06833.","short":"E. Zverev, S. Abdelnabi, S. Tabesh, M. Fritz, C. Lampert, ArXiv (n.d.).","apa":"Zverev, E., Abdelnabi, S., Tabesh, S., Fritz, M., &#38; Lampert, C. (n.d.). Can LLMs separate instructions from data? And what do we even mean by that? <i>arXiv</i>. <a href=\"https://doi.org/10.48550/arXiv.2403.06833\">https://doi.org/10.48550/arXiv.2403.06833</a>","ieee":"E. Zverev, S. Abdelnabi, S. Tabesh, M. Fritz, and C. Lampert, “Can LLMs separate instructions from data? And what do we even mean by that?,” <i>arXiv</i>. .","chicago":"Zverev, Egor, Sahar Abdelnabi, Soroush Tabesh, Mario Fritz, and Christoph Lampert. “Can LLMs Separate Instructions from Data? And What Do We Even Mean by That?” <i>ArXiv</i>, n.d. <a href=\"https://doi.org/10.48550/arXiv.2403.06833\">https://doi.org/10.48550/arXiv.2403.06833</a>."},"file":[{"access_level":"open_access","success":1,"date_updated":"2025-02-20T10:11:45Z","file_size":530972,"file_id":"19064","relation":"main_file","creator":"ezverev","file_name":"2403.06833v3.pdf","checksum":"35eb43968684b87be59144603ef10af0","date_created":"2025-02-20T10:11:45Z","content_type":"application/pdf"}],"file_date_updated":"2025-02-20T10:11:45Z","language":[{"iso":"eng"}],"acknowledgement":"The authors would like to sincerely thank Juan Rocamonde for valuable feedback to our manuscript. We acknowledge the support from the Scientific Service Units (SSU) of ISTA through resources provided by Scientific Computing (SciComp). We thank Dan Alistarh for providing us with computational resources. This work was partially funded by the German Federal Ministry of Education and Research (BMBF) under the grant AIgenCY (16KIS2012) and ELSA – European Lighthouse on Secure and Safe AI funded by the European Union under grant agreement No. 101070617. Views and opinions expressed are however those of the authors only and do not necessarily reflect those of the European Union or European Commission. Neither the European Union nor the European Commission can be held responsible for them.","tmp":{"short":"CC BY (4.0)","legal_code_url":"https://creativecommons.org/licenses/by/4.0/legalcode","name":"Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)","image":"/images/cc_by.png"},"oa_version":"Preprint"},{"publisher":"OpenReview","oa":1,"corr_author":"1","author":[{"full_name":"Scott, Jonathan A","last_name":"Scott","id":"e499926b-f6e0-11ea-865d-9c63db0031e8","first_name":"Jonathan A"},{"full_name":"Zakerinia, Hossein","last_name":"Zakerinia","orcid":"0009-0007-3977-6462","first_name":"Hossein","id":"653bd8b6-f394-11eb-9cf6-c0bbf6cd78d4"},{"last_name":"Lampert","full_name":"Lampert, Christoph","first_name":"Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","orcid":"0000-0001-8622-7887"}],"_id":"17411","acknowledged_ssus":[{"_id":"ScienComp"}],"year":"2024","title":"PEFLL: Personalized federated learning by learning to learn","date_updated":"2026-10-01T09:28:03Z","has_accepted_license":"1","status":"public","quality_controlled":"1","user_id":"317138e5-6ab7-11ef-aa6d-ffef3953e345","department":[{"_id":"ChLa"}],"month":"03","article_processing_charge":"No","publication_status":"published","date_created":"2024-08-11T22:01:12Z","external_id":{"arxiv":["2306.05515"]},"conference":{"end_date":"2024-03-07","name":"ICLR: International Conference on Learning Representations","location":"Vienna, Austria","start_date":"2024-03-07"},"type":"conference","date_published":"2024-03-07T00:00:00Z","related_material":{"record":[{"id":"21198","relation":"dissertation_contains","status":"public"}]},"arxiv":1,"das_tickbox":"0","publication":"12th International Conference on Learning Representations","ddc":["000"],"researchdata_availability":"no","supplementarymaterial":"no","abstract":[{"text":"We present PeFLL, a new personalized federated learning algorithm that improves\r\nover the state-of-the-art in three aspects: 1) it produces more accurate models,\r\nespecially in the low-data regime, and not only for clients present during its\r\ntraining phase, but also for any that may emerge in the future; 2) it reduces the\r\namount of on-client computation and client-server communication by providing\r\nfuture clients with ready-to-use personalized models that require no additional\r\nfinetuning or optimization; 3) it comes with theoretical guarantees that establish\r\ngeneralization from the observed clients to future ones.\r\nAt the core of PeFLL lies a learning-to-learn approach that jointly trains an\r\nembedding network and a hypernetwork. The embedding network is used to\r\nrepresent clients in a latent descriptor space in a way that reflects their similarity\r\nto each other. The hypernetwork takes as input such descriptors and outputs the\r\nparameters of fully personalized client models. In combination, both networks\r\nconstitute a learning algorithm that achieves state-of-the-art performance in several\r\npersonalized federated learning benchmarks","lang":"eng"}],"day":"07","scopus_import":"1","file":[{"date_updated":"2024-08-12T07:38:06Z","success":1,"access_level":"open_access","creator":"dernst","file_id":"17415","relation":"main_file","file_size":1029219,"date_created":"2024-08-12T07:38:06Z","checksum":"81b7ea2e667adaf9c7a7b6b376b1f251","file_name":"2024_ICLR_Scott.pdf","content_type":"application/pdf"}],"file_date_updated":"2024-08-12T07:38:06Z","citation":{"ieee":"J. A. Scott, H. Zakerinia, and C. Lampert, “PEFLL: Personalized federated learning by learning to learn,” in <i>12th International Conference on Learning Representations</i>, Vienna, Austria, 2024.","chicago":"Scott, Jonathan A, Hossein Zakerinia, and Christoph Lampert. “PEFLL: Personalized Federated Learning by Learning to Learn.” In <i>12th International Conference on Learning Representations</i>. OpenReview, 2024.","short":"J.A. Scott, H. Zakerinia, C. Lampert, in:, 12th International Conference on Learning Representations, OpenReview, 2024.","apa":"Scott, J. A., Zakerinia, H., &#38; Lampert, C. (2024). PEFLL: Personalized federated learning by learning to learn. In <i>12th International Conference on Learning Representations</i>. Vienna, Austria: OpenReview.","mla":"Scott, Jonathan A., et al. “PEFLL: Personalized Federated Learning by Learning to Learn.” <i>12th International Conference on Learning Representations</i>, OpenReview, 2024.","ista":"Scott JA, Zakerinia H, Lampert C. 2024. PEFLL: Personalized federated learning by learning to learn. 12th International Conference on Learning Representations. ICLR: International Conference on Learning Representations.","ama":"Scott JA, Zakerinia H, Lampert C. PEFLL: Personalized federated learning by learning to learn. In: <i>12th International Conference on Learning Representations</i>. OpenReview; 2024."},"oa_version":"Published Version","language":[{"iso":"eng"}],"acknowledgement":"This research was supported by the Scientific Service Units (SSU) of ISTA through resources provided by Scientific Computing (SciComp).\r\n"},{"user_id":"317138e5-6ab7-11ef-aa6d-ffef3953e345","quality_controlled":"1","status":"public","title":"Improved modelling of federated datasets using mixtures-of-Dirichlet-multinomials","year":"2024","date_updated":"2026-10-01T09:29:04Z","_id":"18120","corr_author":"1","oa":1,"author":[{"last_name":"Scott","full_name":"Scott, Jonathan A","id":"e499926b-f6e0-11ea-865d-9c63db0031e8","first_name":"Jonathan A"},{"first_name":"Áine","last_name":"Cahill","full_name":"Cahill, Áine"}],"publisher":"ML Research Press","intvolume":"       235","das_tickbox":"0","related_material":{"record":[{"status":"public","id":"21198","relation":"dissertation_contains"}]},"arxiv":1,"date_published":"2024-09-01T00:00:00Z","type":"conference","external_id":{"arxiv":["2406.02416"]},"conference":{"location":"Vienna, Austria","start_date":"2024-07-21","end_date":"2024-07-27","name":"ICML: International Conference on Machine Learning"},"date_created":"2024-09-22T22:01:45Z","main_file_link":[{"open_access":"1","url":"https://doi.org/10.48550/arXiv.2406.02416"}],"publication_status":"published","department":[{"_id":"ChLa"}],"article_processing_charge":"No","month":"09","day":"01","scopus_import":"1","volume":235,"alternative_title":["PMLR"],"supplementarymaterial":"no","abstract":[{"lang":"eng","text":"In practice, training using federated learning can be orders of magnitude slower than standard centralized training. This severely limits the amount of experimentation and tuning that can be done, making it challenging to obtain good performance on a given task. Server-side proxy data can be used to run training simulations, for instance for hyperparameter tuning. This can greatly speed up the training pipeline by reducing the number of tuning runs to be performed overall on the true clients. However, it is challenging to ensure that these simulations accurately reflect the dynamics of the real federated training. In particular, the proxy data used for simulations often comes as a single centralized dataset without a partition into distinct clients, and partitioning this data in a naive way can lead to simulations that poorly reflect real federated training. In this paper we address the challenge of how to partition centralized data in a way that reflects the statistical heterogeneity of the true federated clients. We propose a fully federated, theoretically justified, algorithm that efficiently learns the distribution of the true clients and observe improved server-side simulations when using the inferred distribution to create simulated clients from the centralized data."}],"researchdata_availability":"no","page":"44012-44037","publication":"Proceedings of the 41st International Conference on Machine Learning","acknowledgement":"We would like to thank: Mona Chitnis and everyone in the Private Federated Learning team at Apple for their help and support throughout the entire project; Audra McMillan, Martin Pelikan, Anosh Raj and Barry Theobold for feedback on the initial versions of the paper; and Christoph Lampert for valuable feedback on the paper structure and suggestions for additional experiments.","language":[{"iso":"eng"}],"oa_version":"Preprint","publication_identifier":{"eissn":["2640-3498"]},"citation":{"apa":"Scott, J. A., &#38; Cahill, Á. (2024). Improved modelling of federated datasets using mixtures-of-Dirichlet-multinomials. In <i>Proceedings of the 41st International Conference on Machine Learning</i> (Vol. 235, pp. 44012–44037). Vienna, Austria: ML Research Press.","short":"J.A. Scott, Á. Cahill, in:, Proceedings of the 41st International Conference on Machine Learning, ML Research Press, 2024, pp. 44012–44037.","ieee":"J. A. Scott and Á. Cahill, “Improved modelling of federated datasets using mixtures-of-Dirichlet-multinomials,” in <i>Proceedings of the 41st International Conference on Machine Learning</i>, Vienna, Austria, 2024, vol. 235, pp. 44012–44037.","chicago":"Scott, Jonathan A, and Áine Cahill. “Improved Modelling of Federated Datasets Using Mixtures-of-Dirichlet-Multinomials.” In <i>Proceedings of the 41st International Conference on Machine Learning</i>, 235:44012–37. ML Research Press, 2024.","ama":"Scott JA, Cahill Á. Improved modelling of federated datasets using mixtures-of-Dirichlet-multinomials. In: <i>Proceedings of the 41st International Conference on Machine Learning</i>. Vol 235. ML Research Press; 2024:44012-44037.","ista":"Scott JA, Cahill Á. 2024. Improved modelling of federated datasets using mixtures-of-Dirichlet-multinomials. Proceedings of the 41st International Conference on Machine Learning. ICML: International Conference on Machine Learning, PMLR, vol. 235, 44012–44037.","mla":"Scott, Jonathan A., and Áine Cahill. “Improved Modelling of Federated Datasets Using Mixtures-of-Dirichlet-Multinomials.” <i>Proceedings of the 41st International Conference on Machine Learning</i>, vol. 235, ML Research Press, 2024, pp. 44012–37."}},{"researchdata_availability":"no","abstract":[{"text":"We introduce a new framework for studying meta-learning methods using PAC-Bayesian theory. Its main advantage over previous work is that it allows for more flexibility in how the transfer of knowledge between tasks is realized. For previous approaches, this could only happen indirectly, by means of learning prior distributions over models. In contrast, the new generalization bounds that we prove express the process of meta-learning much more directly as learning the learning algorithm that should be used for future tasks. The flexibility of our framework makes it suitable to analyze a wide range of meta-learning mechanisms and even design new mechanisms. Other than our theoretical contributions we also show empirically that our framework improves the prediction quality in practical meta-learning mechanisms.","lang":"eng"}],"alternative_title":["PMLR"],"supplementarymaterial":"no","volume":235,"scopus_import":"1","day":"01","publication":"Proceedings of the 41st International Conference on Machine Learning","ddc":["000"],"page":"58122-58139","oa_version":"Published Version","language":[{"iso":"eng"}],"citation":{"mla":"Zakerinia, Hossein, et al. “More Flexible PAC-Bayesian Meta-Learning by Learning Learning Algorithms.” <i>Proceedings of the 41st International Conference on Machine Learning</i>, vol. 235, ML Research Press, 2024, pp. 58122–39.","ista":"Zakerinia H, Behjati A, Lampert C. 2024. More flexible PAC-Bayesian meta-learning by learning learning algorithms. Proceedings of the 41st International Conference on Machine Learning. ICML: International Conference on Machine Learning, PMLR, vol. 235, 58122–58139.","ama":"Zakerinia H, Behjati A, Lampert C. More flexible PAC-Bayesian meta-learning by learning learning algorithms. In: <i>Proceedings of the 41st International Conference on Machine Learning</i>. Vol 235. ML Research Press; 2024:58122-58139.","ieee":"H. Zakerinia, A. Behjati, and C. Lampert, “More flexible PAC-Bayesian meta-learning by learning learning algorithms,” in <i>Proceedings of the 41st International Conference on Machine Learning</i>, Vienna, Austria, 2024, vol. 235, pp. 58122–58139.","chicago":"Zakerinia, Hossein, Amin Behjati, and Christoph Lampert. “More Flexible PAC-Bayesian Meta-Learning by Learning Learning Algorithms.” In <i>Proceedings of the 41st International Conference on Machine Learning</i>, 235:58122–39. ML Research Press, 2024.","short":"H. Zakerinia, A. Behjati, C. Lampert, in:, Proceedings of the 41st International Conference on Machine Learning, ML Research Press, 2024, pp. 58122–58139.","apa":"Zakerinia, H., Behjati, A., &#38; Lampert, C. (2024). More flexible PAC-Bayesian meta-learning by learning learning algorithms. In <i>Proceedings of the 41st International Conference on Machine Learning</i> (Vol. 235, pp. 58122–58139). Vienna, Austria: ML Research Press."},"publication_identifier":{"eissn":["2640-3498"]},"_id":"18118","year":"2024","title":"More flexible PAC-Bayesian meta-learning by learning learning algorithms","date_updated":"2026-10-01T09:28:21Z","status":"public","quality_controlled":"1","user_id":"317138e5-6ab7-11ef-aa6d-ffef3953e345","publisher":"ML Research Press","intvolume":"       235","oa":1,"corr_author":"1","author":[{"last_name":"Zakerinia","full_name":"Zakerinia, Hossein","id":"653bd8b6-f394-11eb-9cf6-c0bbf6cd78d4","first_name":"Hossein","orcid":"0009-0007-3977-6462"},{"full_name":"Behjati, Amin","last_name":"Behjati","first_name":"Amin"},{"last_name":"Lampert","full_name":"Lampert, Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","first_name":"Christoph","orcid":"0000-0001-8622-7887"}],"date_published":"2024-09-01T00:00:00Z","type":"conference","arxiv":1,"das_tickbox":"0","month":"09","article_processing_charge":"No","department":[{"_id":"ChLa"}],"main_file_link":[{"url":" https://doi.org/10.48550/arXiv.2402.04054","open_access":"1"}],"publication_status":"published","date_created":"2024-09-22T22:01:45Z","external_id":{"arxiv":["2402.04054"]},"conference":{"location":"Vienna, Austria","start_date":"2024-07-21","end_date":"2024-07-27","name":"ICML: International Conference on Machine Learning"}}]