Pseudo-Rehearsal for Continual Learning with Normalizing Flows

TL;DR

This paper introduces a novel continual learning method combining normalizing flows with regularization, effectively mitigating catastrophic forgetting while maintaining low computational and memory overheads.

Contribution

It proposes a unified approach using a single normalizing flow conditioned on tasks, which regularizes embeddings and replays past data efficiently.

Findings

Performs favorably compared to state-of-the-art methods

Maintains constant memory overhead with a single NF

Reduces catastrophic forgetting effectively

Abstract

Catastrophic forgetting (CF) happens whenever a neural network overwrites past knowledge while being trained on new tasks. Common techniques to handle CF include regularization of the weights (using, e.g., their importance on past tasks), and rehearsal strategies, where the network is constantly re-trained on past data. Generative models have also been applied for the latter, in order to have endless sources of data. In this paper, we propose a novel method that combines the strengths of regularization and generative-based rehearsal approaches. Our generative model consists of a normalizing flow (NF), a probabilistic and invertible neural network, trained on the internal embeddings of the network. By keeping a single NF conditioned on the task, we show that our memory overhead remains constant. In addition, exploiting the invertibility of the NF, we propose a simple approach to regularize the network's embeddings with respect to past tasks. We show that our method performs favorably with respect to state-of-the-art approaches in the literature, with bounded computational power and memory overheads.