arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2102.10490cs.LGcs.CVstat.ML
收藏

较强的NAS和较弱的预测因子

Stronger NAS with Weaker Predictors

Texas A&M University(德州农工大学);Microsoft Corporation(微软公司);University of Texas at Austin(德克萨斯大学奥斯汀分校)

Junru Wu, Xiyang Dai, Dongdong Chen, Yinpeng Chen, Mengchen Liu, Ye Yu, Zhangyang Wang, Zicheng Liu, Mei Chen, Lu Yuan

摘要:

神经架构搜索(NAS)经常训练和评估大量的架构。最近基于预测器的NAS方法试图通过两个关键步骤来减轻这种繁重的计算成本:对一些架构-性能对进行采样和拟合代理精度预测器。然而,给定有限的样本,由于难以拟合巨大的搜索空间,这些预测器远不能准确地定位顶级架构。本文思考了一个简单而关键的问题:如果我们的最终目标是找到最好的建筑,我们真的需要很好地建模整个空间吗?我们提出了一个范式的转变,从拟合整个架构空间使用一个强大的预测,逐步拟合搜索路径的高性能子空间,通过一组较弱的预测。作为弱预测器的一个关键属性,它们对更好架构进行采样的概率不断增加。因此,我们只对一些由先前学习的预测器指导的性能良好的架构进行采样,并估计一个新的更好的弱预测器。这个非常简单的框架,被称为WeakNAS,产生由粗到细的迭代,以逐渐细化采样空间的排名。大量的实验表明,WeakNAS成本更少的样本,找到最高性能的架构NAS-Bench-101和NAS-Bench-201。与最先进的(SOTA)基于预测器的NAS方法相比,WeakNAS以显著的优势优于所有方法,例如,在NAS-Bench-101上找到全局最优值所需的样本至少少7.5倍。WeakNAS还可以吸收他们的想法来进一步提高性能。此外,WeakNAS在ImageNet MobileNet搜索空间中获得了81.3%的新SOTA结果。该代码可在https://github.com/VITA-Group/WeakNAS上获得。

英文摘要:

Neural Architecture Search (NAS) often trains and evaluates a large number of architectures. Recent predictor-based NAS approaches attempt to alleviate such heavy computation costs with two key steps: sampling some architecture-performance pairs and fitting a proxy accuracy predictor. Given limited samples, these predictors, however, are far from accurate to locate top architectures due to the difficulty of fitting the huge search space. This paper reflects on a simple yet crucial question: if our final goal is to find the best architecture, do we really need to model the whole space well?. We propose a paradigm shift from fitting the whole architecture space using one strong predictor, to progressively fitting a search path towards the high-performance sub-space through a set of weaker predictors. As a key property of the weak predictors, their probabilities of sampling better architectures keep increasing. Hence we only sample a few well-performed architectures guided by the previously learned predictor and estimate a new better weak predictor. This embarrassingly easy framework, dubbed WeakNAS, produces coarse-to-fine iteration to gradually refine the ranking of sampling space. Extensive experiments demonstrate that WeakNAS costs fewer samples to find top-performance architectures on NAS-Bench-101 and NAS-Bench-201. Compared to state-of-the-art (SOTA) predictor-based NAS methods, WeakNAS outperforms all with notable margins, e.g., requiring at least 7.5x less samples to find global optimal on NAS-Bench-101. WeakNAS can also absorb their ideas to boost performance more. Further, WeakNAS strikes the new SOTA result of 81.3% in the ImageNet MobileNet Search Space. The code is available at https://github.com/VITA-Group/WeakNAS.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑