A Genetic Programming Approach to Designing Convolutional Neural Network Architectures

Masanori Suganuma, Shinichi Shirakawa, Tomoharu Nagao
4/3/2017
cs.NE

Official code: sg-nm/cgp-cnn (76 stars, MIT, PyTorch, last commit 2023-11-30).

Abstract

The convolutional neural network (CNN), which is one of the deep learning models, has seen much success in a variety of computer vision tasks. However, designing CNN architectures still requires expert knowledge and a lot of trial and error. In this paper, we attempt to automatically construct CNN architectures for an image classification task based on Cartesian genetic programming (CGP). In our method, we adopt highly functional modules, such as convolutional blocks and tensor concatenation, as the node functions in CGP. The CNN structure and connectivity represented by the CGP encoding method are optimized to maximize the validation accuracy. To evaluate the proposed method, we constructed a CNN architecture for the image classification task with the CIFAR-10 dataset. The experimental result shows that the proposed method can be used to automatically find the competitive CNN architecture compared with state-of-the-art models.

View on arXivView PDF

Code Implementations(1)

sg-nm/cgp-cnnOfficial100%
PyTorchMIT

A Genetic Programming Approach to Designing CNN Architectures, In GECCO 2017 (oral presentation, Best Paper Award)

7627PythonJul 13, 20172 years agoMIT
chainerdeep-learningevolutionary-algorithmgenetic-algorithmpython+1 more

Similar Papers

Continuous Cartesian Genetic Programming based representation for Multi-Objective Neural Architecture Search

85% match1 repo22023

Neural Architecture Search Using Genetic Algorithm for Facial Expression Recognition

82% match2023

Automatically designing CNN architectures using genetic algorithm for image classification

82% match1 repo412018

SoftReMish: A Novel Activation Function for Enhanced Convolutional Neural Networks for Visual Recognition Performance

82% match2025

Cartesian Genetic Programming Approach for Designing Convolutional Neural Networks

82% match2024

ConvVitMamba: Efficient Multiscale Convolution, Transformer, and Mamba-Based Sequence modelling for Hyperspectral Image Classification

82% match1 repo22026

Cite this paper

@article{suganuma2017genetic,
  title  = {A Genetic Programming Approach to Designing Convolutional Neural Network Architectures},
  author = {Masanori Suganuma and Shinichi Shirakawa and Tomoharu Nagao},
  year   = {2017},
  eprint = {1704.00764},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1704.00764}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1704.00764.svg)](https://codeofpaper.com/paper/1704.00764)

Discussion