paper-with-me

홈 › Papers

An Analysis of Super-Net Heuristics in Weight-Sharing NAS

2021-10-04 · Kaicheng Yu, René Ranftl, Mathieu Salzmann

Weight sharing promises to make neural architecture search (NAS) tractable even on commodity hardware. Existing methods in this space rely on a diverse set of heuristics to design and train the shared-weight backbone network, a.k.a. the super-net. Since heuristics substantially vary across different methods and have not been carefully studied, it is unclear to which extent they impact super-net training and hence the weight-sharing NAS algorithms. In this paper, we disentangle super-net training from the search algorithm, isolate 14 frequently-used training heuristics, and evaluate them over three benchmark search spaces. Our analysis uncovers that several commonly-used heuristics negatively impact the correlation between super-net and stand-alone performance, whereas simple, but often overlooked factors, such as proper hyper-parameter settings, are key to achieve strong performance. Equipped with this knowledge, we show that simple random search achieves competitive performance to complex state-of-the-art NAS algorithms when the super-net is properly trained.

📄 PDF Abstract BibTeX arXiv:2110.01154

Code (0)

등록된 구현이 없습니다.

Tasks

Neural Architecture Search

Methods 이 논문이 사용한 방법론

Random Search Random Search replaces the exhaustive enumeration of all combinations by selecting them randomly. This can be simply applied to the discrete setting described above, but also…

Similar Papers 제목 키워드 기반

How to Train Your Super-Net: An Analysis of Training Heuristics in Weight-Sharing NAS

2020-03-09 · Kaicheng Yu, Rene Ranftl, Mathieu Salzmann

Weight sharing promises to make neural architecture search (NAS) tractable even on commodity hardware. Existing methods in this space rely on a diverse set of heuristics to design and train the shared-weight backbone net…

Neural Architecture Search

HM-NAS: Efficient Neural Architecture Search via Hierarchical Masking

2019-08-31 · Shen Yan, Biyi Fang, Faen Zhang, Yu Zheng 외

The use of automatic methods, often referred to as Neural Architecture Search (NAS), in designing neural network architectures has recently drawn considerable attention. In this work, we present an efficient NAS approach…

Neural Architecture Search

How Does Supernet Help in Neural Architecture Search?

2020-10-16 · Yuge Zhang, Quanlu Zhang, Yaming Yang

Weight sharing, as an approach to speed up architecture performance estimation has received wide attention. Instead of training each architecture separately, weight sharing builds a supernet that assembles all the archit…

Neural Architecture Search

Institutionally Distributed Deep Learning Networks

2017-09-10 · Ken Chang, Niranjan Balachandar, Carson K Lam, Darvin Yi 외

Deep learning has become a promising approach for automated medical diagnoses. When medical data samples are limited, collaboration among multiple institutions is necessary to achieve high algorithm performance. However,…

Deep Learningimage-classificationImage Classification

AutoDistil: Few-shot Task-agnostic Neural Architecture Search for Distilling Large Language Models

2022-01-29 · Dongkuan Xu, Subhabrata Mukherjee, Xiaodong Liu, Debadeepta Dey 외

Knowledge distillation (KD) methods compress large models into smaller students with manually-designed student architectures given pre-specified computational cost. This requires several trials to find a viable student, …

Inductive BiasKnowledge DistillationNeural Architecture Search