Abstract:Differentiable ARchiTecture Search, i.e., DARTS, has drawn great attention in neural architecture search. It tries to find the optimal architecture in a shallow search network and then measures its performance in a deep evaluation network. The independent optimization of the search and evaluation networks, however, leaves a room for potential improvement by allowing interaction between the two networks. To address the problematic optimization issue, we propose new joint optimization objectives and a novel Cyclic Differentiable ARchiTecture Search framework, dubbed CDARTS. Considering the structure difference, CDARTS builds a cyclic feedback mechanism between the search and evaluation networks with introspective distillation. First, the search network generates an initial architecture for evaluation, and the weights of the evaluation network are optimized. Second, the architecture weights in the search network are further optimized by the label supervision in classification, as well as the regularization from the evaluation network through feature distillation. Repeating the above cycle results in a joint optimization of the search and evaluation networks and thus enables the evolution of the architecture to fit the final evaluation network. The experiments and analysis on CIFAR, ImageNet and NATS-Bench [95] demonstrate the effectiveness of the proposed approach over the state-of-the-art ones. Specifically, in the DARTS search space, we achieve 97.52% top-1 accuracy on CIFAR10 and 76.3% top-1 accuracy on ImageNet. In the chain-structured search space, we achieve 78.2% top-1 accuracy on ImageNet, which is 1.1% higher than EfficientNet-B0. Our code and models are publicly available at https://github.com/microsoft/Cream.

PD-DARTS - Progressive Discretization Differentiable Architecture Search.

$\alpha$ DARTS Once More: Enhancing Differentiable Architecture Search by Masked Image Modeling

Making Differentiable Architecture Search less local

DARTS+: Improved Differentiable Architecture Search with Early Stopping

D-DARTS: Distributed Differentiable Architecture Search

Partially-Connected Neural Architecture Search for Reduced Computational Redundancy

Cyclic Differentiable Architecture Search

Progressive DARTS: Bridging the Optimization Gap for NAS in the Wild

Progressive Differentiable Architecture Search: Bridging the Depth Gap Between Search and Evaluation.

Differentiable Neural Architecture Search Via Proximal Iterations.

Idarts: Improving DARTS by Node Normalization and Decorrelation Discretization

Single-DARTS: Towards Stable Architecture Search

EPC-DARTS: Efficient partial channel connection for differentiable architecture search

OStr-DARTS: Differentiable Neural Architecture Search based on Operation Strength

Enhancing the Robustness, Efficiency, and Diversity of Differentiable Architecture Search

Improving Differentiable Architecture Search via self-distillation

PC-DARTS: Partial Channel Connections for Memory-Efficient Architecture Search

Differentiable Architecture Search with Random Features

RAPDARTS: Resource-Aware Progressive Differentiable Architecture Search

A new differentiable architecture search method for optimizing convolutional neural networks in the digital twin of intelligent robotic grasping

Semantic-DARTS: Elevating Semantic Learning for Mobile Differentiable Architecture Search