Continual Learning on Noisy Data Streams via Self-Purified Replay

Continually learning in the real world must overcome many challenges, among which noisy labels are a common and inevitable issue. In this work, we present a replaybased continual learning framework that simultaneously addresses both catastrophic forgetting and noisy labels for the first time. Our solution is based on two observations; (i) forgetting can be mitigated even with noisy labels via selfsupervised learning, and (ii) the purity of the replay buffer is crucial. Building on this regard, we propose two key components of our method: (i) a self-supervised replay technique named Self-Replay which can circumvent erroneous training signals arising from noisy labeled data, and (ii) the SelfCentered filter that maintains a purified replay buffer via centrality-based stochastic graph ensembles. The empirical results on MNIST, CIFAR-10, CIFAR-100, and WebVision with real-world noise demonstrate that our framework can maintain a highly pure replay buffer amidst noisy streamed data while greatly outperforming the combinations of the state-of-the-art continual learning and noisy label learning methods.