BaLeNAS: Differentiable Architecture Search via the Bayesian Learning Rule

Abstract

Differentiable Architecture Search (DARTS) has received massive attention in recent years, mainly because it significantly reduces the computational cost through weight sharing and continuous relaxation. However, more recent works find that existing differentiable NAS techniques struggle to outperform naive baselines, yielding deteriorative architectures as the search proceeds. Rather than directly optimizing the architecture parameters, this paper formulates the neural architecture search as a distribution learning problem through relaxing the architecture weights into Gaussian distributions. By leveraging the natural-gradient variational inference (NGVI), the architecture distribution can be easily optimized based on existing codebases without incurring more memory and computational consumption. We demonstrate how the differentiable NAS benefits from Bayesian principles, enhancing exploration and improving stability. The experimental results on NAS-Bench-201 and NAS-Bench-1shot1 benchmark datasets confirm the significant improvements the proposed framework can make. In addition, instead of simply applying the argmax on the learned parameters, we further leverage the recently-proposed training-free proxies in NAS to select the optimal architecture from a group architectures drawn from the optimized distribution, where we achieve state-of-the-art results on the NAS-Bench-201 and NAS-Bench-1shot1 benchmarks. Our best architecture in the DARTS search space also obtains competitive test errors with 2.37\%, 15.72\%, and 24.2\% on CIFAR-10, CIFAR-100, and ImageNet datasets, respectively.

References in corpus (19)

• DARTS: Differentiable Architecture Search
• SNIP: Single-shot Network Pruning based on Connection Sensitivity
• PC-DARTS: Partial Channel Connections for Memory-Efficient Architecture Search
• SNAS: Stochastic Neural Architecture Search
• Understanding and Robustifying Differentiable Architecture Search
• DARTS+: Improved Differentiable Architecture Search with Early Stopping
• NAS-Bench-101: Towards Reproducible Neural Architecture Search
• Single Path One-Shot Neural Architecture Search with Uniform Sampling
• One-Shot Neural Architecture Search via Self-Evaluated Template Network
• A Comprehensive Survey of Neural Architecture Search: Challenges and Solutions
• Picking Winning Tickets Before Training by Preserving Gradient Flow
• Practical Deep Learning with Bayesian Principles
• BayesNAS: A Bayesian Approach for Neural Architecture Search
• NAS-Bench-1Shot1: Benchmarking and Dissecting One-shot Neural Architecture Search
• Neural Architecture Search on ImageNet in Four GPU Hours: A Theoretically Inspired Perspective
• Neural Architecture Search without Training
• Rethinking Architecture Selection in Differentiable NAS
• Blockwisely Supervised Neural Architecture Search with Knowledge Distillation
• Training Binary Neural Networks using the Bayesian Learning Rule