Continual learning with Bayesian model based on a fixed pre-trained feature extractor

Authors:

Abstract
Deep learning has shown its human-level performance in various applications. However, current deep learning models are characterized by catastrophic forgetting of old knowledge when learning new classes. This poses a challenge such as in intelligent diagnosis systems where initially only training data of a limited number of diseases are available. In this case, updating the intelligent system with data of new diseases would inevitably downgrade its performance on previously learned diseases. Inspired by the process of learning new knowledge in human brains, we propose a Bayesian generative model for continual learning built on a fixed pre-trained feature extractor. In this model, knowledge of each old class can be compactly represented by a collection of statistical distributions, e.g., with Gaussian mixture models, and naturally kept from forgetting in continual learning over time. Unlike existing class-incremental learning methods, the proposed approach is not sensitive to the continual learning process and can be additionally well applied to the data-incremental learning scenario. Experiments on multiple medical and natural image classification tasks reveal that the proposed approach outperforms state-of-the-art approaches that even keep some images of old classes during continual learning of new classes.
Funding Information
  • Innovative Research Group Project of the National Natural Science Foundation of China (62071502, U1811461)
  • Special Project for Research and Development in Key areas of Guangdong Province (2020B1111190001)

This publication has 15 references indexed in Scilit: