A Neural Architecture Search Method using Auxiliary Evaluation Metric based on ResNet Architecture

TL;DR

This paper introduces a neural architecture search method leveraging ResNet with an auxiliary evaluation metric, optimizing both accuracy and validation loss to find competitive models on multiple datasets.

Contribution

It presents a novel search space based on ResNet and incorporates validation loss as a secondary objective for improved architecture optimization.

Findings

Achieved competitive accuracy on MNIST, Fashion-MNIST, CIFAR100

Demonstrated effectiveness of auxiliary validation loss in NAS

Validated the approach across multiple datasets

Abstract

This paper proposes a neural architecture search space using ResNet as a framework, with search objectives including parameters for convolution, pooling, fully connected layers, and connectivity of the residual network. In addition to recognition accuracy, this paper uses the loss value on the validation set as a secondary objective for optimization. The experimental results demonstrate that the search space of this paper together with the optimisation approach can find competitive network architectures on the MNIST, Fashion-MNIST and CIFAR100 datasets.