报告题目：Fitness Landscape Analysis for Supervised Learning
报告专家：Ferrante Neri 教授
报告时间：7月6日（周四）下午15:00
报告地点：信息科技大楼A101-102会议室
主 持 人：刘 琦 教授
专家简介：
Prof. Neri received a PhD in Electrical Engineering from the Technical University of Bari, Italy. He also received a PhD in Scientific Computing and Optimisation from University of Jyvaskyla, Finland. Prof. Neri subsequently worked for the Academy of Finland, De Montfort University and the University of Nottingham. He is currently a professor of Machine Learning and Artificial Intelligence, and the Head of the Nature Inspired Computing and Engineering (NICE) research group, at the University of Surrey. He is also a Distinguished Professor of Jiangsu Province. His research interests include hybrid heuristic-exact optimisation, memetic computing, differential evolution, and membrane computing. Prof. Neri published over 200 items including two editions of the textbook “Linear Algebra for Computational Sciences and Engineering” and is an Associate Editor of multiple journals including Information Sciences and Integrated Computing Engineering.
报告简介：
The term Fitness Landscape Analysis (FLA) refers to a set of techniques to analyse black-box optimisation problems with the purpose of informing the design of an appropriate solver. This talk focuses on FLA in the training of neural networks and provides the details of two case studies. The first is the training of a Convolutional Neural Network for image classification while the second is the training of a Multi-perceptron Network for Reinforcement Learning in the context of robotics. One example of an ad-hoc design of a learning algorithm is provided. It is highlighted how popular learning algorithms can be outperformed by FLA-based optimisers.
软件学院
2023年7月4日
