Remembering for the Right Reasons: Explanations Reduce Catastrophic Forgetting

TL;DR

This paper introduces a novel continual learning method called Remembering for the Right Reasons (RRR) that uses explanations to reduce forgetting and improve model interpretability across tasks.

Contribution

The paper proposes a simple training paradigm that incorporates explanation consistency into continual learning, reducing forgetting and enhancing explainability.

Findings

RRR reduces catastrophic forgetting across various CL approaches.

Incorporating explanations improves model interpretability.

The method shows consistent gains in both standard and few-shot learning settings.

Abstract

The goal of continual learning (CL) is to learn a sequence of tasks without suffering from the phenomenon of catastrophic forgetting. Previous work has shown that leveraging memory in the form of a replay buffer can reduce performance degradation on prior tasks. We hypothesize that forgetting can be further reduced when the model is encouraged to remember the \textit{evidence} for previously made decisions. As a first step towards exploring this hypothesis, we propose a simple novel training paradigm, called Remembering for the Right Reasons (RRR), that additionally stores visual model explanations for each example in the buffer and ensures the model has "the right reasons" for its predictions by encouraging its explanations to remain consistent with those used to make decisions at training time. Without this constraint, there is a drift in explanations and increase in forgetting as conventional continual learning algorithms learn new tasks. We demonstrate how RRR can be easily added to any memory or regularization-based approach and results in reduced forgetting, and more importantly, improved model explanations. We have evaluated our approach in the standard and few-shot settings and observed a consistent improvement across various CL approaches using different architectures and techniques to generate model explanations and demonstrated our approach showing a promising connection between explainability and continual learning. Our code is available at \url{https://github.com/SaynaEbrahimi/Remembering-for-the-Right-Reasons}.