References & Citations
Computer Science > Machine Learning
Title: Representational Continuity for Unsupervised Continual Learning
(Submitted on 13 Oct 2021 (v1), last revised 5 Apr 2022 (this version, v3))
Abstract: Continual learning (CL) aims to learn a sequence of tasks without forgetting the previously acquired knowledge. However, recent CL advances are restricted to supervised continual learning (SCL) scenarios. Consequently, they are not scalable to real-world applications where the data distribution is often biased and unannotated. In this work, we focus on unsupervised continual learning (UCL), where we learn the feature representations on an unlabelled sequence of tasks and show that reliance on annotated data is not necessary for continual learning. We conduct a systematic study analyzing the learned feature representations and show that unsupervised visual representations are surprisingly more robust to catastrophic forgetting, consistently achieve better performance, and generalize better to out-of-distribution tasks than SCL. Furthermore, we find that UCL achieves a smoother loss landscape through qualitative analysis of the learned representations and learns meaningful feature representations. Additionally, we propose Lifelong Unsupervised Mixup (LUMP), a simple yet effective technique that interpolates between the current task and previous tasks' instances to alleviate catastrophic forgetting for unsupervised representations.
Submission history
From: Divyam Madaan [view email][v1] Wed, 13 Oct 2021 18:38:06 GMT (17833kb,D)
[v2] Fri, 15 Oct 2021 20:19:13 GMT (17833kb,D)
[v3] Tue, 5 Apr 2022 02:33:03 GMT (18059kb,D)
Link back to: arXiv, form interface, contact.