M2KD: Multi-model and Multi-level Knowledge Distillation for Incremental Learning

TL;DR

This paper introduces a multi-model, multi-level knowledge distillation approach for incremental learning, leveraging all previous models and intermediate features to better retain old knowledge and improve performance.

Contribution

It proposes a novel strategy that distills knowledge from all previous models and intermediate features, enhancing old class retention in incremental learning.

Findings

Improves performance on old classes compared to standard methods.

Uses mask-based pruning for memory-efficient model reconstruction.

Achieves better overall accuracy on benchmark datasets.

Abstract

Incremental learning targets at achieving good performance on new categories without forgetting old ones. Knowledge distillation has been shown critical in preserving the performance on old classes. Conventional methods, however, sequentially distill knowledge only from the last model, leading to performance degradation on the old classes in later incremental learning steps. In this paper, we propose a multi-model and multi-level knowledge distillation strategy. Instead of sequentially distilling knowledge only from the last model, we directly leverage all previous model snapshots. In addition, we incorporate an auxiliary distillation to further preserve knowledge encoded at the intermediate feature levels. To make the model more memory efficient, we adapt mask based pruning to reconstruct all previous models with a small memory footprint. Experiments on standard incremental learning benchmarks show that our method preserves the knowledge on old classes better and improves the overall performance over standard distillation techniques.