Flexible Multi-task Networks by Learning Parameter Allocation
This addresses the challenge of optimizing knowledge transfer in multi-task learning for AI applications, though it is incremental as it builds on existing parameter sharing methods.
The paper tackles the problem of parameter sharing in multi-task neural networks, where sharing between unrelated tasks can hurt performance, by proposing a framework that learns fine-grained patterns of parameter allocation to encourage sharing between related tasks and discourage it otherwise, achieving a 17% relative error reduction on the Omniglot benchmark compared to state-of-the-art.
This paper proposes a novel learning method for multi-task applications. Multi-task neural networks can learn to transfer knowledge across different tasks by using parameter sharing. However, sharing parameters between unrelated tasks can hurt performance. To address this issue, we propose a framework to learn fine-grained patterns of parameter sharing. Assuming that the network is composed of several components across layers, our framework uses learned binary variables to allocate components to tasks in order to encourage more parameter sharing between related tasks, and discourage parameter sharing otherwise. The binary allocation variables are learned jointly with the model parameters by standard back-propagation thanks to the Gumbel-Softmax reparametrization method. When applied to the Omniglot benchmark, the proposed method achieves a 17% relative reduction of the error rate compared to state-of-the-art.