Skip to content

[AAAI 2025] CDM-PSL: Expensive Multi-Objective Bayesian Optimization Based on Diffusion Models

License

Notifications You must be signed in to change notification settings

aaron-di/CDM-PSL

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 

Repository files navigation

CDM-PSL: Expensive Multi-Objective Bayesian Optimization Based on Diffusion Models

This repository includes the code for the paper “Expensive Multi-Objective Bayesian Optimization Based on Diffusion Models”, which has been accepted at AAAI 2025. The preprint is at: https://arxiv.org/abs/2405.08674

💥 News 💥

  • 🎉🎉🎉 [Dec 10, 2024] Our paper is accepted at AAAI 2025! The camera ready version is coming soon.

Abstract

CDM-PSL (Composite Diffusion Model-Based Pareto Set Learning) is a novel algorithm designed to address the challenges of expensive multi-objective Bayesian optimization (EMOBO). While MOBO has shown great potential in solving expensive multi-objective optimization problems (EMOPs), it struggles to model complex distributions of the Pareto optimal solutions when only a limited number of function evaluations are available.

Existing Pareto set learning methods often exhibit instability in expensive scenarios, leading to significant deviations between the obtained solution set and the true Pareto set (PS). To overcome these challenges, CDM-PSL introduces Composite Diffusion Model to enhance optimization performance.

Overview

🚧 Code Coming Soon! 🚧

The source code for CDM-PSL will be released after the AAAI 2025 conference presentation.

Citation

If you find our work helpful, please cite the following BibTeX entry:

@article{li2024expensive,
  title={Expensive Multi-Objective Bayesian Optimization Based on Diffusion Models},
  author={Li, Bingdong and Di, Zixiang and Lu, Yongfan and Qian, Hong and Wang, Feng and Yang, Peng and Tang, Ke and Zhou, Aimin},
  journal={accepted by AAAI},
  year={2025}
}

About

[AAAI 2025] CDM-PSL: Expensive Multi-Objective Bayesian Optimization Based on Diffusion Models

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published