We present SOSELETO (SOurce SELEction for Target Optimization), a new method for exploiting a source dataset to solve a classification problem on a target dataset. SOSELETO is based on the following simple intuition: some source examples are more informative than others for the target problem. To capture this intuition, source samples are each given weights; these weights are solved for jointly with the source and target classification problems via a bilevel optimization scheme. The target therefore gets to choose the source samples which are most informative for its own classification task. Furthermore, the bilevel nature of the optimization acts as a kind of regularization on the target, mitigating overfitting. SOSELETO may be applied to both classic transfer learning, as well as the problem of training on datasets with noisy labels; we show state of the art results on both of these problems.
@article{arxiv.1805.09622,
title = {SOSELETO: A Unified Approach to Transfer Learning and Training with Noisy Labels},
author = {Or Litany and Daniel Freedman},
journal= {arXiv preprint arXiv:1805.09622},
year = {2019}
}
Comments
ICLR workshop on Learning from Limited Labeled Data (LLD) -- Best Paper Award