Not Just Selection, but Exploration: Online Class-Incremental Continual Learning via Dual View Consistency

Citations Over TimeTop 10% of 2022 papers

Abstract

Online class-incremental continual learning aims to learn new classes continually from a never-ending and single-pass data stream, while not forgetting the learned knowledge of old classes. Existing replay-based methods have shown promising performance by storing a subset of old class data. Unfortunately, these methods only focus on selecting samples from the memory bank for replay and ignore the adequate exploration of semantic information in the single-pass data stream, leading to poor classification accuracy. In this paper, we propose a novel yet effective framework for online class-incremental continual learning, which considers not only the selection of stored samples, but also the full exploration of the data stream. Specifically, we propose a gradient-based sample selection strategy, which selects the stored samples whose gradients generated in the network are most interfered by the new incoming samples. We believe such samples are beneficial for updating the neural network based on back gradient propagation. More importantly, we seek to explore the semantic information between two different views of training images by maximizing their mutual information, which is conducive to the improvement of classification accuracy. Extensive experimental results demonstrate that our method achieves state-of-the-art performance on a variety of benchmark datasets. Our code is available on https://github.com/YananGu/DVC.

Related Papers

• → Ways of Forgetting: Israel and the Obliterated Memory of the Palestinian Nakba(2009)71 cited
• → Measurements of everyday memory: Toward the prevention of forgetting(1984)134 cited
• → Measures of Forgetting(2018)2 cited
• → Selective Forgetting of Deep Networks at a Finer Level than Samples(2020)2 cited
• → The Work of Forgetting(2021)