Association: Remind Your GAN not to Forget

TL;DR

This paper introduces a brain-inspired associative learning approach to mitigate catastrophic forgetting in neural networks, enabling continual learning without access to original data, inspired by human memory mechanisms.

Contribution

It proposes a novel associative memory-inspired framework with potentiation and depression mechanisms for continual learning, avoiding the need for original data access.

Findings

Effective in reducing catastrophic forgetting on image-to-image translation tasks

No requirement for original data during training

Inspired by human cognitive processes

Abstract

Neural networks are susceptible to catastrophic forgetting. They fail to preserve previously acquired knowledge when adapting to new tasks. Inspired by human associative memory system, we propose a brain-like approach that imitates the associative learning process to achieve continual learning. We design a heuristics mechanism to potentiatively stimulate the model, which guides the model to recall the historical episodes based on the current circumstance and obtained association experience. Besides, a distillation measure is added to depressively alter the efficacy of synaptic transmission, which dampens the feature reconstruction learning for new task. The framework is mediated by potentiation and depression stimulation that play opposing roles in directing synaptic and behavioral plasticity. It requires no access to the original data and is more similar to human cognitive process. Experiments demonstrate the effectiveness of our method in alleviating catastrophic forgetting on image-to-image translation tasks.