Task Scheduling for Heterogeneous Systems Using a Hybrid Deep Neural Network and Genetic Algorithm Approach

Yutao Han

2024

Abstract

Task scheduling in heterogeneous computing systems is a highly complex and challenging problem due to the diverse architectures and varying computational capabilities of different hardware resources. Efficiently allocating tasks to these resources to optimize performance is a significant challenge in such environments. This study addresses this challenge by combining the deep neural network with the genetic algorithm to create an efficient task scheduling approach. The research focuses on constructing a deep neural network that progressively learns from the task scheduling schemes generated by the genetic algorithm, aiming to accelerate the scheduling process. The method involves using the genetic algorithm to generate initial scheduling solutions and training a Deep Neural Networks (DNN) to learn from these solutions. The results show that it is difficult for the network to fully reproduce the performance of genetic algorithm-based scheduling, but the network significantly reduces the time required to generate effective scheduling plans. This hybrid model highlights the potential of leveraging machine learning techniques to enhance the efficiency of task scheduling in heterogeneous computing systems.

Paper Citation

in Harvard Style

Han Y. (2024). Task Scheduling for Heterogeneous Systems Using a Hybrid Deep Neural Network and Genetic Algorithm Approach. In Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML; ISBN 978-989-758-754-2, SciTePress, pages 137-143. DOI: 10.5220/0013510800004619

in Bibtex Style

@conference{daml24,
author={Yutao Han},
title={Task Scheduling for Heterogeneous Systems Using a Hybrid Deep Neural Network and Genetic Algorithm Approach},
booktitle={Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML},
year={2024},
pages={137-143},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013510800004619},
isbn={978-989-758-754-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML
TI - Task Scheduling for Heterogeneous Systems Using a Hybrid Deep Neural Network and Genetic Algorithm Approach
SN - 978-989-758-754-2
AU - Han Y.
PY - 2024
SP - 137
EP - 143
DO - 10.5220/0013510800004619
PB - SciTePress