Measuring Catastrophic Forgetting in Neural Networks

Abstract

Deep neural networks are used in many state-of-the-art systems for machine perception. Once a network is trained to do a specific task, e.g., bird classification, it cannot easily be trained to do new tasks, e.g., incrementally learning to recognize additional bird species or learning an entirely different task such as flower recognition. When new tasks are added, typical deep neural networks are prone to catastrophically forgetting previous tasks. Networks that are capable of assimilating new information incrementally, much like how humans form new memories over time, will be more efficient than re-training the model from scratch each time a new task needs to be learned. There have been multiple attempts to develop schemes that mitigate catastrophic forgetting, but these methods have not been directly compared, the tests used to evaluate them vary considerably, and these methods have only been evaluated on small-scale problems (e.g., MNIST). In this paper, we introduce new metrics and benchmarks for directly comparing five different mechanisms designed to mitigate catastrophic forgetting in neural networks: regularization, ensembling, rehearsal, dual-memory, and sparse-coding. Our experiments on real-world images and sounds show that the mechanism(s) that are critical for optimal performance vary based on the incremental training paradigm and type of data being used, but they all demonstrate that the catastrophic forgetting problem has yet to be solved.

To appear in AAAI 2018

Cited by in corpus (16)

• Orthogonal Gradient Descent for Continual Learning
• Defining Benchmarks for Continual Few-Shot Learning
• Anatomy of Catastrophic Forgetting: Hidden Representations and Task Semantics
• The Next Big Thing(s) in Unsupervised Machine Learning: Five Lessons from Infant Learning
• Exploring Benefits of Transfer Learning in Neural Machine Translation
• Never Forget: Balancing Exploration and Exploitation via Learning Optical Flow
• Autoencoder-Based Incremental Class Learning without Retraining on Old Data
• Learning Sparse Representations Incrementally in Deep Reinforcement Learning
• Dissecting Catastrophic Forgetting in Continual Learning by Deep Visualization
• Neural Network Retraining for Model Serving
• Answer Them All! Toward Universal Visual Question Answering Models
• Continual Learning for Natural Language Generation in Task-oriented Dialog Systems
• Towards a practical measure of interference for reinforcement learning
• Seq2Biseq: Bidirectional Output-wise Recurrent Neural Networks for Sequence Modelling
• Plug-and-Play Anomaly Detection with Expectation Maximization Filtering
• Continuous Learning for Large-scale Personalized Domain Classification