Variational Prototype Replays for Continual Learning

Abstract

Continual learning refers to the ability to acquire and transfer knowledge without catastrophically forgetting what was previously learned. In this work, we consider \emph{few-shot} continual learning in classification tasks, and we propose a novel method, Variational Prototype Replays, that efficiently consolidates and recalls previous knowledge to avoid catastrophic forgetting. In each classification task, our method learns a set of variational prototypes with their means and variances, where embedding of the samples from the same class can be represented in a prototypical distribution and class-representative prototypes are separated apart. To alleviate catastrophic forgetting, our method replays one sample per class from previous tasks, and correspondingly matches newly predicted embeddings to their nearest class-representative prototypes stored from previous tasks. Compared with recent continual learning approaches, our method can readily adapt to new tasks with more classes without requiring the addition of new units. Furthermore, our method is more memory efficient since only class-representative prototypes with their means and variances, as well as only one sample per class from previous tasks need to be stored. Without tampering with the performance on initial tasks, our method learns novel concepts given a few training examples of each class in new tasks.

under submission

References in corpus (15)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Distilling the Knowledge in a Neural Network
• Prototypical Networks for Few-shot Learning
• Do Deep Nets Really Need to be Deep?
• Tutorial on Variational Autoencoders
• Understanding Black-box Predictions via Influence Functions
• Continual Learning Through Synaptic Intelligence
• FearNet: Brain-Inspired Model for Incremental Learning
• Variational Continual Learning
• Overcoming Catastrophic Forgetting by Incremental Moment Matching
• Measuring Catastrophic Forgetting in Neural Networks
• Generative replay with feedback connections as a general strategy for continual learning
• Dynamic Few-Shot Visual Learning without Forgetting
• Pseudo-Recursal: Solving the Catastrophic Forgetting Problem in Deep Neural Networks
• Few-Shot Self Reminder to Overcome Catastrophic Forgetting

Cited by in corpus (8)

• Understanding the Role of Training Regimes in Continual Learning
• Orthogonal Gradient Descent for Continual Learning
• Prototype Guided Federated Learning of Visual Feature Representations
• Linear Mode Connectivity in Multitask and Continual Learning
• ARCADe: A Rapid Continual Anomaly Detector
• Continual Learning via Bit-Level Information Preserving
• Task-agnostic Continual Learning with Hybrid Probabilistic Models
• Shared and Private VAEs with Generative Replay for Continual Learning