Evolutionary search algorithm for learning activation function of an artificial neural network

Neural networks require careful selection of activation functions to optimize performance. Traditional methods of choosing activation functions through trial and error are time-consuming and resource-intensive. This paper presents a novel approach to automatically design activation functions for art...

Full description

Saved in:
Bibliographic Details
Main Author: Yurshin Viacheslav
Format: Article
Language:English
Published: EDP Sciences 2025-01-01
Series:ITM Web of Conferences
Online Access:https://www.itm-conferences.org/articles/itmconf/pdf/2025/03/itmconf_hmmocs-III2024_05004.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850195541163507712
author Yurshin Viacheslav
author_facet Yurshin Viacheslav
author_sort Yurshin Viacheslav
collection DOAJ
description Neural networks require careful selection of activation functions to optimize performance. Traditional methods of choosing activation functions through trial and error are time-consuming and resource-intensive. This paper presents a novel approach to automatically design activation functions for artificial neural networks using genetic programming combined with gradient descent. The proposed method aims to enhance the efficiency of the search process for optimal activation functions. Our algorithm employs genetic programming to evolve the general form of activation functions, while gradient descent optimizes their parameters during network training. This hybrid approach allows for the exploration of a wide range of potential activation functions tailored to specific tasks and network architectures. The method was evaluated on three datasets from the KEEL repository: Iris, Titanic, and Phoneme. The results demonstrate the algorithm's ability to generate and optimize custom activation functions, although improvements in network accuracy were not observed in this initial study. This work contributes to the ongoing research in neural network optimization and opens avenues for further investigation into the automatic design of activation functions.
format Article
id doaj-art-aafd2a1fecac4ec08568b8248f0de264
institution OA Journals
issn 2271-2097
language English
publishDate 2025-01-01
publisher EDP Sciences
record_format Article
series ITM Web of Conferences
spelling doaj-art-aafd2a1fecac4ec08568b8248f0de2642025-08-20T02:13:44ZengEDP SciencesITM Web of Conferences2271-20972025-01-01720500410.1051/itmconf/20257205004itmconf_hmmocs-III2024_05004Evolutionary search algorithm for learning activation function of an artificial neural networkYurshin Viacheslav0Siberian Federal University, Institute of Space and Information TechnologiesNeural networks require careful selection of activation functions to optimize performance. Traditional methods of choosing activation functions through trial and error are time-consuming and resource-intensive. This paper presents a novel approach to automatically design activation functions for artificial neural networks using genetic programming combined with gradient descent. The proposed method aims to enhance the efficiency of the search process for optimal activation functions. Our algorithm employs genetic programming to evolve the general form of activation functions, while gradient descent optimizes their parameters during network training. This hybrid approach allows for the exploration of a wide range of potential activation functions tailored to specific tasks and network architectures. The method was evaluated on three datasets from the KEEL repository: Iris, Titanic, and Phoneme. The results demonstrate the algorithm's ability to generate and optimize custom activation functions, although improvements in network accuracy were not observed in this initial study. This work contributes to the ongoing research in neural network optimization and opens avenues for further investigation into the automatic design of activation functions.https://www.itm-conferences.org/articles/itmconf/pdf/2025/03/itmconf_hmmocs-III2024_05004.pdf
spellingShingle Yurshin Viacheslav
Evolutionary search algorithm for learning activation function of an artificial neural network
ITM Web of Conferences
title Evolutionary search algorithm for learning activation function of an artificial neural network
title_full Evolutionary search algorithm for learning activation function of an artificial neural network
title_fullStr Evolutionary search algorithm for learning activation function of an artificial neural network
title_full_unstemmed Evolutionary search algorithm for learning activation function of an artificial neural network
title_short Evolutionary search algorithm for learning activation function of an artificial neural network
title_sort evolutionary search algorithm for learning activation function of an artificial neural network
url https://www.itm-conferences.org/articles/itmconf/pdf/2025/03/itmconf_hmmocs-III2024_05004.pdf
work_keys_str_mv AT yurshinviacheslav evolutionarysearchalgorithmforlearningactivationfunctionofanartificialneuralnetwork