Selfless Sequential Learning

Abstract

Sequential learning, also called lifelong learning, studies the problem of learning tasks in a sequence with access restricted to only the data of the current task. In this paper we look at a scenario with fixed model capacity, and postulate that the learning process should not be selfish, i.e. it should account for future tasks to be added and thus leave enough capacity for them. To achieve Selfless Sequential Learning we study different regularization strategies and activation functions. We find that imposing sparsity at the level of the representation (i.e.~neuron activations) is more beneficial for sequential learning than encouraging parameter sparsity. In particular, we propose a novel regularizer, that encourages representation sparsity by means of neural inhibition. It results in few active neurons which in turn leaves more free neurons to be utilized by upcoming tasks. As neural inhibition over an entire layer can be too drastic, especially for complex tasks requiring strong representations, our regularizer only inhibits other neurons in a local neighbourhood, inspired by lateral inhibition processes in the brain. We combine our novel regularizer, with state-of-the-art lifelong learning methods that penalize changes to important previously learned parts of the network. We show that our new regularizer leads to increased sparsity which translates in consistent performance improvement %over alternative regularizers we studied on diverse datasets.

arXiv commentPublished as a conference paper at ICLR 2019

References in corpus (10)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Overcoming catastrophic forgetting in neural networks
• Riemannian Walk for Incremental Learning: Understanding Forgetting and Intransigence
• An Empirical Investigation of Catastrophic Forgetting in Gradient-Based Neural Networks
• PathNet: Evolution Channels Gradient Descent in Super Neural Networks
• Gradient Episodic Memory for Continual Learning
• Bayesian Compression for Deep Learning
• Overcoming catastrophic forgetting with hard attention to the task
• Regularizing CNNs with Locally Constrained Decorrelations
• Rotate your Networks: Better Weight Consolidation and Less Catastrophic Forgetting

Cited by in corpus (20)

• A continual learning survey: Defying forgetting in classification tasks
• Continual Learning for Recurrent Neural Networks: an Empirical Evaluation
• Gradient based sample selection for online continual learning
• La-MAML: Look-ahead Meta Learning for Continual Learning
• An Investigation of Replay-based Approaches for Continual Learning
• Continual Learning with Adaptive Weights (CLAW)
• Meta-Learning Representations for Continual Learning
• Continuous learning of spiking networks trained with local rules
• Beneficial Perturbation Network for designing general adaptive artificial intelligence systems
• Understanding Continual Learning Settings with Data Distribution Drift Analysis
• Continual Representation Learning for Biometric Identification
• A Study of Continual Learning Methods for Q-Learning
• Neural Network Retraining for Model Serving
• Overcoming Catastrophic Forgetting by Neuron-level Plasticity Control
• Frosting Weights for Better Continual Training
• Optimizing Reusable Knowledge for Continual Learning via Metalearning
• Learning Continually from Low-shot Data Stream
• Learning by Active Forgetting for Neural Networks
• Zero-shot task adaptation by homoiconic meta-mapping
• Class Incremental Online Streaming Learning