ProxyBO: Accelerating Neural Architecture Search via Bayesian Optimization with Zero-cost Proxies
This work addresses the efficiency problem in neural architecture design for machine learning practitioners, offering a significant speed improvement over existing NAS methods.
The paper tackles the slow speed of neural architecture search (NAS) by proposing ProxyBO, a Bayesian optimization framework that uses zero-cost proxies to accelerate the process, achieving up to 5.41x and 3.86x speedups over state-of-the-art methods.
Designing neural architectures requires immense manual efforts. This has promoted the development of neural architecture search (NAS) to automate the design. While previous NAS methods achieve promising results but run slowly, zero-cost proxies run extremely fast but are less promising. Therefore, it is of great potential to accelerate NAS via those zero-cost proxies. The existing method has two limitations, which are unforeseeable reliability and one-shot usage. To address the limitations, we present ProxyBO, an efficient Bayesian optimization (BO) framework that utilizes the zero-cost proxies to accelerate neural architecture search. We apply the generalization ability measurement to estimate the fitness of proxies on the task during each iteration and design a novel acquisition function to combine BO with zero-cost proxies based on their dynamic influence. Extensive empirical studies show that ProxyBO consistently outperforms competitive baselines on five tasks from three public benchmarks. Concretely, ProxyBO achieves up to 5.41x and 3.86x speedups over the state-of-the-art approaches REA and BRP-NAS.