Papers › Noisy Differentiable Architecture Search

Noisy Differentiable Architecture Search

7 May 2020arXiv:2005.03566archive 2025-07-28

Xiangxiang Chu, Bo Zhang

Simplicity is the ultimate sophistication. Differentiable Architecture Search (DARTS) has now become one of the mainstream paradigms of neural architecture search. However, it largely suffers from the well-known performance collapse issue due to the aggregation of skip connections. It is thought to have overly benefited from the residual structure which accelerates the information flow. To weaken this impact, we propose to inject unbiased random noise to impede the flow. We name this novel approach NoisyDARTS. In effect, a network optimizer should perceive this difficulty at each training step and refrain from overshooting, especially on skip connections. In the long run, since we add no bias to the gradient in terms of expectation, it is still likely to converge to the right solution area. We also prove that the injected noise plays a role in smoothing the loss landscape, which makes the optimization easier. Our method features extreme simplicity and acts as a new strong baseline. We perform extensive experiments across various search spaces, datasets, and tasks, where we robustly achieve state-of-the-art results. Our code is available at https://github.com/xiaomi-automl/NoisyDARTS.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

xiaomi-automl/NoisyDARTS officialmentioned in papermentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

AutoMLImage ClassificationNeural Architecture Search

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Classification CIFAR-10 NoisyDARTS-A-t Percentage correct 98.28 #48 of 265 Archive leaderboard report
Image Classification CIFAR-10 NoisyDARTS-a Percentage correct 97.61 #75 of 265 Archive leaderboard report
Neural Architecture Search CIFAR-10 NoisyDARTS FLOPS 534M #12 of 41 Archive leaderboard report
Neural Architecture Search CIFAR-10 NoisyDARTS Parameters 3.25M #12 of 41 Archive leaderboard report
Neural Architecture Search CIFAR-10 NoisyDARTS Search Time (GPU days) 0.25 #12 of 41 Archive leaderboard report
Neural Architecture Search CIFAR-10 NoisyDARTS Top-1 Error Rate 2.39% #12 of 41 Archive leaderboard report
Neural Architecture Search ImageNet NoisyDARTS-A Accuracy 77.9 #54 of 135 Archive leaderboard report
Neural Architecture Search ImageNet NoisyDARTS-A MACs 449M #54 of 135 Archive leaderboard report
Neural Architecture Search ImageNet NoisyDARTS-A Params 5.5M #54 of 135 Archive leaderboard report
Neural Architecture Search ImageNet NoisyDARTS-A Top-1 Error Rate 22.1 #54 of 135 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections