Hyper-parameter optimization of Deep Convolutional Networks for object recognition

TL;DR

This paper explores the use of sequential model-based optimization (SMBO) to automatically tune hyper-parameters of deep convolutional networks for object recognition, achieving competitive results on benchmark datasets.

Contribution

It introduces a simple SMBO strategy for architecture search in deep convolutional networks, demonstrating its effectiveness in finding high-performing models.

Findings

SMBO can identify architectures with performance comparable to state-of-the-art.

The proposed method effectively automates hyper-parameter tuning for DCNs.

Results are validated on standard object recognition benchmarks.

Abstract

Recently sequential model based optimization (SMBO) has emerged as a promising hyper-parameter optimization strategy in machine learning. In this work, we investigate SMBO to identify architecture hyper-parameters of deep convolution networks (DCNs) object recognition. We propose a simple SMBO strategy that starts from a set of random initial DCN architectures to generate new architectures, which on training perform well on a given dataset. Using the proposed SMBO strategy we are able to identify a number of DCN architectures that produce results that are comparable to state-of-the-art results on object recognition benchmarks.