Evolving neural architecture using one shot model

Nilotpal Sinha, Kuan Wen Chen

研究成果: Conference contribution同行評審

14 引文 斯高帕斯(Scopus)

摘要

Previous evolution based architecture search require high computational resources resulting in large search time. In this work, we propose a novel way of applying a simple genetic algorithm to the neural architecture search problem called EvNAS (Evolving Neural Architecture using One Shot Model) which reduces the search time significantly while still achieving better result than previous evolution based methods. The architectures are represented by architecture parameter of one shot model which results in the weight sharing among the given population of architectures and also weight inheritance from one generation to the next generation of architectures. We use the accuracy of partially trained architecture on validation data as a prediction of its fitness to reduce the search time. We also propose a decoding technique for the architecture parameter which is used to divert majority of the gradient information towards the given architecture and is also used for improving the fitness prediction of the given architecture from the one shot model during the search process. EvNAS searches for architecture on CIFAR-10 for 3.83 GPU day on a single GPU with top-1 test error 2.47%, which is then transferred to CIFAR-100 and ImageNet achieving top-1 error 16.37% and top-5 error 7.4% respectively.

原文English
主出版物標題GECCO 2021 - Proceedings of the 2021 Genetic and Evolutionary Computation Conference
發行者Association for Computing Machinery, Inc
頁面910-918
頁數9
ISBN(電子)9781450383509
DOIs
出版狀態Published - 26 6月 2021
事件2021 Genetic and Evolutionary Computation Conference, GECCO 2021 - Virtual, Online, France
持續時間: 10 7月 202114 7月 2021

出版系列

名字GECCO 2021 - Proceedings of the 2021 Genetic and Evolutionary Computation Conference

Conference

Conference2021 Genetic and Evolutionary Computation Conference, GECCO 2021
國家/地區France
城市Virtual, Online
期間10/07/2114/07/21

指紋

深入研究「Evolving neural architecture using one shot model」主題。共同形成了獨特的指紋。

引用此