Bridging Inter-task Gap of Continual Self-supervised Learning with External Data

01 May 2024 (modified: 06 Nov 2024)Submitted to NeurIPS 2024EveryoneRevisionsBibTeXCC BY-NC 4.0
Keywords: self-supervised learning, continual learning
Abstract: Recent research on Self-Supervised Learning (SSL) has demonstrated its ability to extract high-quality representations from unlabeled samples. However, in continual learning scenarios where training data arrives sequentially, SSL's performance tends to deteriorate. This study focuses on Continual Contrastive Self-Supervised Learning (CCSSL) and highlights that the absence of contrastive learning on inter-task data, due to the unavailability of historical samples, leads to a significant drop in performance. To tackle this issue, we introduce a simple and effective method called BGE, which Bridges the inter-task Gap of CCSSL using External data from publicly available datasets. BGE enables the contrastive learning of each task data with external data, allowing relationships between them to be passed along the tasks, thereby facilitating implicit inter-task data comparisons. To overcome the limitation of the external data selection and maintain its effectiveness, we further propose the One-Propose-One algorithm to collect more relevant and diverse high-quality samples from the chosen external data while filtering out distractions from the out-of-distribution data. Experiments show that BGE can generate better discriminative representation in CCSSL, especially for inter-task data, and improve classification results with various external data compositions. Additionally, the proposed method can be seamlessly integrated into existing continual learning methods yielding significant performance improvement.
Supplementary Material: zip
Primary Area: Online learning
Submission Number: 1299
Loading