AggSS: An Aggregated Self-Supervised Approach for Class-Incremental Learning

TL;DR

AggSS introduces a self-supervised aggregation method that improves class-incremental learning by enhancing feature robustness and can be integrated into existing frameworks, demonstrated on CIFAR-100 and ImageNet-Subset.

Contribution

The paper proposes AggSS, a plug-and-play self-supervised approach that boosts class-incremental learning performance by aggregating rotation-based predictions for better feature learning.

Findings

AggSS improves accuracy on CIFAR-100 and ImageNet-Subset.

It shifts neural attention towards intrinsic object features.

AggSS enhances various class-incremental learning methods.

Abstract

This paper investigates the impact of self-supervised learning, specifically image rotations, on various class-incremental learning paradigms. Here, each image with a predefined rotation is considered as a new class for training. At inference, all image rotation predictions are aggregated for the final prediction, a strategy we term Aggregated Self-Supervision (AggSS). We observe a shift in the deep neural network's attention towards intrinsic object features as it learns through AggSS strategy. This learning approach significantly enhances class-incremental learning by promoting robust feature learning. AggSS serves as a plug-and-play module that can be seamlessly incorporated into any class-incremental learning framework, leveraging its powerful feature learning capabilities to enhance performance across various class-incremental learning approaches. Extensive experiments conducted on standard incremental learning datasets CIFAR-100 and ImageNet-Subset demonstrate the significant role of AggSS in improving performance within these paradigms.