D-DARTS: Distributed Differentiable Architecture Search

TL;DR

D-DARTS introduces a nested architecture search method that enhances diversity and depth of neural networks, improves performance, and offers an alternative to traditional NAS by optimizing existing architectures.

Contribution

It proposes nesting at the cell level, a new algorithm for deeper architectures from few cells, and an alternative search space for optimizing existing models.

Findings

Achieves competitive performance on vision tasks.

Provides a new metric for architecture similarity.

Demonstrates effectiveness of nested search approach.

Abstract

Differentiable ARchiTecture Search (DARTS) is one of the most trending Neural Architecture Search (NAS) methods. It drastically reduces search cost by resorting to weight-sharing. However, it also dramatically reduces the search space, thus excluding potential promising architectures. In this article, we propose D-DARTS, a solution that addresses this problem by nesting neural networks at the cell level instead of using weight-sharing to produce more diversified and specialized architectures. Moreover, we introduce a novel algorithm that can derive deeper architectures from a few trained cells, increasing performance and saving computation time. In addition, we also present an alternative search space (DARTOpti) in which we optimize existing handcrafted architectures (e.g., ResNet) rather than starting from scratch. This approach is accompanied by a novel metric that measures the distance between architectures inside our custom search space. Our solution reaches competitive performance on multiple computer vision tasks. Code and pretrained models can be accessed at https://github.com/aheuillet/D-DARTS.