Adaptive Robust Learning using Latent Bernoulli Variables
2024 (English)In: Proceedings of the 41st International Conference on Machine Learning / [ed] Ruslan Salakhutdinov; Zico Kolter; Katherine Heller, Adrian Weller; Nuria Oliver; Jonathan Scarlett; Felix Berkenkamp, PLMR -Proceedings of Machine Learning Research , 2024, p. 23105-23122Conference paper, Published paper (Refereed)
Abstract [en]
We present an adaptive approach for robust learning from corrupted training sets. We identify corrupted and non-corrupted samples with latent Bernoulli variables and thus formulate the learning problem as maximization of the likelihood where latent variables are marginalized. The resulting problem is solved via variational inference, using an efficient Expectation-Maximization based method. The proposed approach improves over the state-of-the-art by automatically inferring the corruption level, while adding minimal computational overhead. We demonstrate our robust learning method and its parameter-free nature on a wide variety of machine learning tasks including online learning and deep learning where it adapts to different levels of noise and maintains high prediction accuracy.
Place, publisher, year, edition, pages
PLMR -Proceedings of Machine Learning Research , 2024. p. 23105-23122
Series
Proceedings of Machine Learning Research, ISSN 2640-3498 ; 235
Keywords [en]
robustness, statistical learning, adaptive method, probabilistic method, latent variables, variational inference
National Category
Probability Theory and Statistics
Research subject
Machine learning
Identifiers
URN: urn:nbn:se:uu:diva-535364OAI: oai:DiVA.org:uu-535364DiVA, id: diva2:1885926
Conference
The 41st International Conference on Machine Learning, Vienna, Austria, 21-27 July, 2024
Projects
eSSENCE - An eScience Collaboration
Funder
Swedish Research Council, 2018-05040Swedish Research Council, 2023-055932024-07-272024-07-272025-01-07Bibliographically approved