Fisher Task Distance and Its Application in Neural Architecture Search
This work addresses the challenge of accelerating neural architecture search for machine learning practitioners by leveraging task similarity, though it is incremental as it builds on existing gradient-based methods.
The authors tackled the problem of efficiently transferring knowledge between tasks in neural architecture search by introducing an asymmetric Fisher task distance, which reduced search space complexity and improved performance, achieving better results and fewer parameters compared to methods like ENAS, DARTS, and PC-DARTS on datasets including MNIST, CIFAR-10, CIFAR-100, and ImageNet.
We formulate an asymmetric (or non-commutative) distance between tasks based on Fisher Information Matrices, called Fisher task distance. This distance represents the complexity of transferring the knowledge from one task to another. We provide a proof of consistency for our distance through theorems and experiments on various classification tasks from MNIST, CIFAR-10, CIFAR-100, ImageNet, and Taskonomy datasets. Next, we construct an online neural architecture search framework using the Fisher task distance, in which we have access to the past learned tasks. By using the Fisher task distance, we can identify the closest learned tasks to the target task, and utilize the knowledge learned from these related tasks for the target task. Here, we show how the proposed distance between a target task and a set of learned tasks can be used to reduce the neural architecture search space for the target task. The complexity reduction in search space for task-specific architectures is achieved by building on the optimized architectures for similar tasks instead of doing a full search and without using this side information. Experimental results for tasks in MNIST, CIFAR-10, CIFAR-100, ImageNet datasets demonstrate the efficacy of the proposed approach and its improvements, in terms of the performance and the number of parameters, over other gradient-based search methods, such as ENAS, DARTS, PC-DARTS.