Learn to Bind and Grow Neural Structures
This work provides an incremental improvement for researchers working on continual learning by offering a new way to manage neural network expansion.
This paper addresses task-incremental learning by proposing a framework that learns neural architectures for new tasks by either binding with similar task layers or expanding conflicting layers. The method achieves comparable performance to existing expansion-based approaches on continual learning benchmarks and can find multiple optimal solutions with performance-size trade-offs.
Task-incremental learning involves the challenging problem of learning new tasks continually, without forgetting past knowledge. Many approaches address the problem by expanding the structure of a shared neural network as tasks arrive, but struggle to grow optimally, without losing past knowledge. We present a new framework, Learn to Bind and Grow, which learns a neural architecture for a new task incrementally, either by binding with layers of a similar task or by expanding layers which are more likely to conflict between tasks. Central to our approach is a novel, interpretable, parameterization of the shared, multi-task architecture space, which then enables computing globally optimal architectures using Bayesian optimization. Experiments on continual learning benchmarks show that our framework performs comparably with earlier expansion based approaches and is able to flexibly compute multiple optimal solutions with performance-size trade-offs.