Task-agnostic Continual Learning with Hybrid Probabilistic Models

Abstract

Learning new tasks continuously without forgetting on a constantly changing data distribution is essential for real-world problems but extremely challenging for modern deep learning. In this work we propose HCL, a Hybrid generative-discriminative approach to Continual Learning for classification. We model the distribution of each task and each class with a normalizing flow. The flow is used to learn the data distribution, perform classification, identify task changes, and avoid forgetting, all leveraging the invertibility and exact likelihood which are uniquely enabled by the normalizing flow model. We use the generative capabilities of the flow to avoid catastrophic forgetting through generative replay and a novel functional regularization technique. For task identification, we use state-of-the-art anomaly detection techniques based on measuring the typicality of the model's statistics. We demonstrate the strong performance of HCL on a range of continual learning benchmarks such as split-MNIST, split-CIFAR, and SVHN-MNIST. * Work partially done as an intern at DeepMind.

Simulating Task-Free Continual Learning Streams From Existing Datasets

“…In other words, the stream consists of a number contiguous iid sub-streams (each one corresponding to a task), and the distribution only changes when there is a transition from one sub-stream to the next. In this setting, however, it is relatively easy to infer task labels during training [17,36].…”

Continual learning on 3D point clouds with random compressed rehearsal

“…This phenomena is called catastrophic forgetting. Finding an effective way to avoid it is a main object of research in the field of continual learning (Mai et al, 2021;Masana et al, 2020;Belouadah et al, 2020;Delange et al, 2021;Parisi et al, 2019), with most of the current research focusing on image (Rusu et al, 2016;Li & Hoiem, 2017;Kirkpatrick et al, 2017;Shin et al, 2017;Zenke et al, 2017;Kirichenko et al, 2021) and video data (Doshi & Yilmaz, 2020;2022). The further natural direction, being an application to the domain of 3D computer vision, is only just starting to be explored (Chowdhury et al, 2021;Dong et al, 2021).…”