报告人：Thomas Bäck 院士
工作单位：莱顿大学（荷兰）
报告题目：自动生成优化启发式方法：从算法配置到大型语言模型
报告时间：2024年12月11日（周三）15:30
报告链接：Teams Link
https://teams.microsoft.com/l/meetup-join/19%3ameeting_ZDg4ZWYxNjctMWRlNi00NzI4LTlhYzMtMGYxZjk1NDc2ZmQx%40thread.v2/0?context=%7b%22Tid%22%3a%2222804ebb-30d5-47df-942f-f3a3722f0225%22%2c%22Oid%22%3a%2216a60c03-ad7a-4b85-a403-8ebd947e010c%22%7d
内容摘要：
大量的自然启发式优化算法（例如，进化算法EA、粒子群算法PSO、差分进化算法DE、蚁群算法ACO）及其变体使得选择最佳算法极具挑战性。在本次报告中，首先我们将介绍并讨论自动优化优化启发式算法的思路，举例说明如何在所谓的模块化协方差矩阵自适应进化策略框架中搜索由数千种进化策略配置变体构成的组合设计空间，以及如何利用数据挖掘来分析结果，并概述向诸如粒子群优化和差分进化等其他算法设计空间的拓展情况。其次，我们将讨论该方法在实际工程问题中的应用，并以汽车碰撞为例，展示其性能提升。最后，我们将介绍LLaMEA，一种利用大型语言模型自动生成高性能元启发式算法的新方法。
个人简介：
Thomas Bäck 院士（IEEE会士）分别于1990年和1994年获计算机科学学士学位和博士学位（导师：H.-P. Schwefel 教授），现为荷兰莱顿大学莱顿高等计算机科学研究所（LIACS）教授，他的研究兴趣包括进化计算、机器学习及其在现实世界中的应用，特别是在可持续智能产业和健康领域。Thomas Bäck教授当选为荷兰皇家艺术与科学院院士（KNAW，2021年）、IEEE 会士（2022年）和欧洲科学院院士（2022年）。Thomas Bäck院士于1995年获德国计算机科学学会（GI）颁发的最佳博士论文奖，于2003年当选国际遗传和进化计算学会会士，并于2015年获IEEE计算智能学会（CIS）进化计算先驱奖。Thomas Bäck院士现担任《Evolutionary Computation》主编、《IEEE Transactions on Evolutionary Computation》和《Artificial Intelligence Review 》期刊的副编辑，以及ACM《Transactions on Evolutionary Learning and Optimization.》的区域主编，曾任《Handbook of Evolutionary Computation》（CRC出版社/Taylor & Francis 1997）和《Handbook of Natural Computing》（Springer，2013）的联合主编，也是《Evolutionary Computation in Theory and Practice》的作者（OUP，纽约，1996）及《Contemporary Evolution Strategies》的合著者（Springer，2013）。
Academic Report Notice of Thomas Bäck : Automated Generation of Optimization Heuristics: From Algorithm Configuration to Large Language Models
Speaker: Academician Thomas Bäck
Title: Automated Generation of Optimization Heuristics: From Algorithm Configuration to Large Language Models
Time: December 11, 2024 (Wednesday)15:30
Website: Teams Link
https://teams.microsoft.com/l/meetup-join/19%3ameeting_ZDg4ZWYxNjctMWRlNi00NzI4LTlhYzMtMGYxZjk1NDc2ZmQx%40thread.v2/0?context=%7b%22Tid%22%3a%2222804ebb-30d5-47df-942f-f3a3722f0225%22%2c%22Oid%22%3a%2216a60c03-ad7a-4b85-a403-8ebd947e010c%22%7d
Abstract:
A large number of nature-inspired optimization algorithms (such as evolutionary algorithms (EA), particle swarm optimization (PSO), differential evolution (DE), and ant colony optimization (ACO)) and their variants make it challenging to choose the best algorithm. In this presentation, we will firstly introduce and discuss the idea of automatically optimizing optimization heuristics. We will illustrate how to search the combination design space consisting of thousands of evolution strategy configuration variants using a modular covariance matrix adaptive evolution strategy framework and how to analyze the results using data mining. We will also provide an overview of the expansion of the design space to other algorithms such as particle swarm optimization and differential evolution. Secondly, we will discuss the application of this method to real-world engineering problems, showcasing performance improvements using a car crash simulation as an example. Finally, we will introduce LLaMEA (Large Language Model Evolutionary Algorithm), a new method for automatically generating high-performance meta-heuristic algorithms using large language models.
Personal Introduction:
Thomas Bäck (Fellow, IEEE) received the Diploma degree in Computer Science in 1990 and the Ph.D. degree in Computer Science in 1994 (under supervision of H.-P. Schwefel), both from the University of Dortmund, Germany. He is Professor of Computer Science with the Leiden Institute of Advanced Computer Science (LIACS), Leiden University, Netherlands. His research interests include evolutionary computation, machine learning, and their real-world applications, especially in sustainable smart industry and health. Dr. Bäck has been elected as member of the Royal Netherlands Academy of Arts and Sciences (KNAW, 2021), as IEEE Fellow (class of 2022), and as a member of Academia Europaea (2022). Dr. Bäck has been elected as member of the Royal Netherlands Academy of Arts and Sciences (KNAW, 2021), as IEEE Fellow (class of 2022), and as a member of Academia Europaea (2022). He was a recipient of the IEEE Computational Intelligence Society (CIS) Evolutionary Computation Pioneer Award in 2015, was elected as Fellow of the International Society of Genetic and Evolutionary Computation in 2003, and received the best Ph.D. thesis award from the German society of Computer Science (GI) in 1995.He currently serves as an Editor in Chief of the Evolutionary Computation Journal (MIT Press), Associate Editor of the IEEE Transactions on Evolutionary Computation and Artificial Intelligence Review journals and area editor of the ACM Transactions on Evolutionary Learning and Optimization. He was also co-editor-in-chief of the Handbook of Evolutionary Computation (CRC Press/Taylor & Francis 1997), co-editor of the Handbook of Natural Computing (Springer, 2013), author of Evolutionary Computation in Theory and Practice (OUP, New York, 1996) and co-author of Contemporary Evolution Strategies (Springer, 2013).