NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:4164
Title:Continual Unsupervised Representation Learning

The task of continual unsupervised learning without task labels and boundaries is interesting, and would be important for learning in the real-world system. The novelty of the proposed method is high. The paper is well written. Experiments with larger-scale data would strengthen the paper.