KRNet: Towards Efficient Knowledge Replay

TL;DR

KRNet introduces an efficient knowledge replay method that significantly reduces storage costs by directly mapping sample identities to data, enhancing continual learning without needing an encoder.

Contribution

The paper presents KRNet, a novel model that directly maps sample identities to data, reducing storage and training complexity compared to autoencoders.

Findings

KRNet requires 400x less storage than autoencoders.

KRNet effectively supports continual learning tasks.

Extensive experiments validate KRNet's efficiency.

Abstract

The knowledge replay technique has been widely used in many tasks such as continual learning and continuous domain adaptation. The key lies in how to effectively encode the knowledge extracted from previous data and replay them during current training procedure. A simple yet effective model to achieve knowledge replay is autoencoder. However, the number of stored latent codes in autoencoder increases linearly with the scale of data and the trained encoder is redundant for the replaying stage. In this paper, we propose a novel and efficient knowledge recording network (KRNet) which directly maps an arbitrary sample identity number to the corresponding datum. Compared with autoencoder, our KRNet requires significantly ($400\times$) less storage cost for the latent codes and can be trained without the encoder sub-network. Extensive experiments validate the efficiency of KRNet, and as a showcase, it is successfully applied in the task of continual learning.