ExCon: Explanation-driven Supervised Contrastive Learning for Image Classification

Abstract

Contrastive learning has led to substantial improvements in the quality of learned embedding representations for tasks such as image classification. However, a key drawback of existing contrastive augmentation methods is that they may lead to the modification of the image content which can yield undesired alterations of its semantics. This can affect the performance of the model on downstream tasks. Hence, in this paper, we ask whether we can augment image data in contrastive learning such that the task-relevant semantic content of an image is preserved. For this purpose, we propose to leverage saliency-based explanation methods to create content-preserving masked augmentations for contrastive learning. Our novel explanation-driven supervised contrastive learning (ExCon) methodology critically serves the dual goals of encouraging nearby image embeddings to have similar content and explanation. To quantify the impact of ExCon, we conduct experiments on the CIFAR-100 and the Tiny ImageNet datasets. We demonstrate that ExCon outperforms vanilla supervised contrastive learning in terms of classification, explanation quality, adversarial robustness as well as probabilistic calibration in the context of distributional shift.

References in corpus (16)

• Adam: A Method for Stochastic Optimization
• Grad-CAM++: Improved Visual Explanations for Deep Convolutional Networks
• Axiomatic Attribution for Deep Networks
• On Calibration of Modern Neural Networks
• Learning deep representations by mutual information estimation and maximization
• SGDR: Stochastic Gradient Descent with Warm Restarts
• Towards Deep Learning Models Resistant to Adversarial Attacks
• Can You Trust Your Model's Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shift
• Data-Efficient Image Recognition with Contrastive Predictive Coding
• SmoothGrad: removing noise by adding noise
• AutoAugment: Learning Augmentation Policies from Data
• Big Self-Supervised Models are Strong Semi-Supervised Learners
• A Downsampled Variant of ImageNet as an Alternative to the CIFAR datasets
• Supervised Contrastive Learning
• Revisiting the Calibration of Modern Neural Networks
• Robust Representation Learning via Perceptual Similarity Metrics