Systems Research

Reinforcement Learni9ng

Few-shot Neural Architecture Search

June 30, 2021

Abstract

Efficient evaluation of a network architecture drawn from a large search space remains a key challenge in Neural Architecture Search (NAS). Vanilla NAS evaluates each architecture by training from scratch, which gives the true performance but is extremely time-consuming. Recently, one-shot NAS substantially reduces the computation cost by training only one supernetwork, a.k.a. supernet, to approximate the performance of every architecture in the search space via weight-sharing. However, the performance estimation can be very inaccurate due to the co-adaption among operations. In this paper, we propose few-shot NAS that uses multiple supernetworks, called sub-supernet, each covering different regions of the search space to alleviate the undesired co-adaption. Compared to one-shot NAS, few-shot NAS improves the accuracy of architecture evaluation with a small increase of evaluation cost. With only up to 7 sub-supernets, few-shot NAS establishes new SoTAs: on ImageNet, it finds models that reach 80.5% top-1 accuracy at 600 MB FLOPS and 77.5% top-1 accuracy at 238 MFLOPS; on CIFAR10, it reaches 98.72% top-1 accuracy without using extra data or transfer learning. In Auto-GAN, few-shot NAS outperforms the previously published results by up to 20%. Extensive experiments show that few-shot NAS significantly improves various one-shot methods, including 4 gradient-based and 6 search-based methods on 3 different tasks in NasBench-201 and NasBench1-shot-1.

Download the Paper

AUTHORS

Written by

Yiyang Zhao

Linnan Wang

Yuandong Tian

Rodrigo Fonseca

Tian Guo

Publisher

ICML 2021

Research Topics

Reinforcement Learning

Systems Research

Core Machine Learning

Related Publications

December 05, 2020

Robotics

Reinforcement Learni9ng

Neural Dynamic Policies for End-to-End Sensorimotor Learning

Deepak Pathak, Abhinav Gupta, Mustafa Mukadam, Shikhar Bahl

December 05, 2020

December 07, 2020

Reinforcement Learni9ng

Joint Policy Search for Collaborative Multi-agent Imperfect Information Games

Yuandong Tian, Qucheng Gong, Tina Jiang

December 07, 2020

March 13, 2021

Reinforcement Learni9ng

On the Importance of Hyperparameter Optimization for Model-based Reinforcement Learning

Baohe Zhang, Raghu Rajan, Luis Pineda, Nathan Lambert, Andre Biedenkapp, Kurtland Chua, Frank Hutter, Roberto Calandra

March 13, 2021

October 10, 2020

Computer Vision

Reinforcement Learni9ng

Active MR k-space Sampling with Reinforcement Learning

Luis Pineda, Sumana Basu, Adriana Romero,Roberto CalandraRoberto Calandra, Michal Drozdzal

October 10, 2020

December 05, 2020

Reinforcement Learni9ng

An Asymptotically Optimal Primal-Dual Incremental Algorithm for Contextual Linear Bandits

Andrea Tirinzonin, Matteo Pirotta, Marcello Restelli, Alessandro Lazaric

December 05, 2020

Help Us Pioneer The Future of AI

We share our open source frameworks, tools, libraries, and models for everything from research exploration to large-scale production deployment.