Extracting robust and accurate features via a robust information bottleneck
IEEE J. Sel. Areas Inf. Theory, pp. 131-144, 2020.
We propose a novel strategy for extracting features in supervised learning that can be used to construct a classifier which is more robust to small perturbations in the input space. Our method builds upon the idea of the information bottleneck, by introducing an additional penalty term that encourages the Fisher information of the extract...More
PPT (Upload PPT)