AutoTransfer: Subject Transfer Learning with Censored Representations on Biosignals Data
This addresses the challenge of generalizing machine learning models across different subjects in biosignal applications, though it appears incremental as it builds on existing transfer learning concepts with new regularization techniques.
The authors tackled the problem of subject transfer learning in biosignals data by developing a regularization framework that minimizes classification loss while penalizing dependence between latent representations and subject labels, showing that their AutoTransfer method improves performance on EEG, EMG, and ECoG datasets.
We provide a regularization framework for subject transfer learning in which we seek to train an encoder and classifier to minimize classification loss, subject to a penalty measuring independence between the latent representation and the subject label. We introduce three notions of independence and corresponding penalty terms using mutual information or divergence as a proxy for independence. For each penalty term, we provide several concrete estimation algorithms, using analytic methods as well as neural critic functions. We provide a hands-off strategy for applying this diverse family of regularization algorithms to a new dataset, which we call "AutoTransfer". We evaluate the performance of these individual regularization strategies and our AutoTransfer method on EEG, EMG, and ECoG datasets, showing that these approaches can improve subject transfer learning for challenging real-world datasets.