Fine-Tuning DARTS for Image Classification

TL;DR

This paper enhances DARTS, a neural architecture search method, by fine-tuning with fixed operations, leading to improved accuracy in image classification tasks across multiple datasets.

Contribution

The paper introduces a fine-tuning approach for DARTS that mitigates approximation issues, achieving better accuracy and a favorable parameter-performance trade-off.

Findings

Improved top-1 accuracy on Fashion-MNIST, CompCars, MIO-TCD datasets.

Enhanced performance over DARTS on CIFAR-10 and CIFAR-100.

Achieved state-of-the-art results with fine-tuned DARTS.

Abstract

Neural Architecture Search (NAS) has gained attraction due to superior classification performance. Differential Architecture Search (DARTS) is a computationally light method. To limit computational resources DARTS makes numerous approximations. These approximations result in inferior performance. We propose to fine-tune DARTS using fixed operations as they are independent of these approximations. Our method offers a good trade-off between the number of parameters and classification accuracy. Our approach improves the top-1 accuracy on Fashion-MNIST, CompCars, and MIO-TCD datasets by 0.56%, 0.50%, and 0.39%, respectively compared to the state-of-the-art approaches. Our approach performs better than DARTS, improving the accuracy by 0.28%, 1.64%, 0.34%, 4.5%, and 3.27% compared to DARTS, on CIFAR-10, CIFAR-100, Fashion-MNIST, CompCars, and MIO-TCD datasets, respectively.