Learning Task-Relevant Features via Contrastive Input Morphing
A fundamental challenge in artificial intelligence is learning useful representations of data that yield good performance on a downstream classification task, without overfitting to spurious input features. Extracting task-relevant predictive information becomes particularly challenging for high-dimensional, noisy, real-world data. We propose Contrastive Input Morphing (CIM), a representation learning framework that learns input-space transformations of the data to mitigate the effect of irrelevant input features on downstream performance via a triplet loss. Empirically, we demonstrate the efficacy of our approach on various tasks which typically suffer from the presence of spurious correlations, and show that CIM improves the performance of other representation learning methods such as variational information bottleneck (VIB) when used in conjunction.
PDF Abstract
ImageNet
CelebA
