# Theory-Inspired Path-Regularized Differential Network Architecture Search

Source: [Theory-Inspired Path-Regularized Differential Network Architecture Search](<https://www.sota2.com/research/paper/68fa57751e209131ce9a5983>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-25T14:44:29.431Z

Authors: Pan Zhou, Caiming Xiong, Richard Socher, Steven C.H. Hoi

Published: 2020-06-30T05:28:23.000Z

## Abstract

Despite its high search efficiency, differential architecture search (DARTS) often selects network architectures with dominated skip connections which lead to performance degradation. However, theoretical understandings on this issue remain absent, hindering the development of more advanced methods in a principled way. In this work, we solve this problem by theoretically analyzing the effects of various types of operations, e.g. convolution, skip connection and zero operation, to the network optimization. We prove that the architectures with more skip connections can converge faster than the other candidates, and thus are selected by DARTS. This result, for the first time, theoretically and explicitly reveals the impact of skip connections to fast network optimization and its competitive advantage over other types of operations in DARTS. Then we propose a theory-inspired path-regularized DARTS that consists of two key modules: (i) a differential group-structured sparse binary gate introduced for each operation to avoid unfair competition among operations, and (ii) a path-depth-wise regularization used to incite search exploration for deep architectures that often converge slower than shallow ones as shown in our theory and are not well explored during the search. Experimental results on image classification tasks validate its advantages.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Image Classification on ImageNet](<https://www.sota2.com/research/sota/image-classification-on-imagenet.md>): Image Classification; ImageNet; Top-1 Accuracy: 77.9.

- [Image Classification on CIFAR-10 (test)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test.md>): Image Classification; CIFAR-10 (test); Accuracy: not recorded.

- [Image Classification on ImageNet (test)](<https://www.sota2.com/research/sota/image-classification-on-imagenet-test.md>): Image Classification; ImageNet (test); Top-1 Acc: not recorded.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2006.16537>)

[PDF](<https://arxiv.org/pdf/2006.16537.pdf>)
