Wenwen Li is a Research Assocaite in Machine Learning Group, University of Sheffield. She is currently working on Dr. Haiping Lu’s project – Learning Sparse Features from 4D fMRI Data for Brain Disease Diagnosis (EP/R014507/1). The project aims to develop a new tensor-based machine learning method for dealing with severe cases of “large p, small n” for multidimensional data such as whole-brain fMRI.
She obtained her PhD degree in Computer Science from the University of Nottingham, UK. She was supervised by Dr. Ender Özcan and Prof. Robert John from Automated Scheduling, Optimisation and Planning (ASAP) research group. She received an MSc. with Distinction in Operational Research from the University of Edinburgh, UK in 2013 and a BEng. in Software Engineering from Anhui University, China in 2008.
Her current research interests include regularisation methods such as lasso, fused lasso, group lasso etc. alternating direction method of multipliers (ADMM) and sparse and low-rank tensor decomposition. Wenwen Li also has expertise in multiobjective evolutionary algorithms, hyper-heuristics and the application of machine learning methods in the design of evolutionary algorithms. She has published in conferences and journals including IEEE Congress on Evolutionary Computation (CEC), and the IEEE Transactions on Evolutionary Computation.
PhD in Computer Science, 2018
University of Nottingham, UK
MSc Operational Research, 2013
University of Edinburgh, UK
BEng Software Engineering, 2008
Anhui University, China
This project has two primary foci. First, to gain deep understanding of the pros and cons of different types of multiobjective evolutionary algorithms (MOEAs) and the influences on their performance on both mathematical benchmarks and real-world problems. For example, wind farm layout optimisation and the vehicle crashworthiness problems. The second seeks to design a high-level method to combine the strengths of different MOEAs in an online learning manner. To achieve this goal, a reinforcement learning-based multiobjetive selection hyper-heuristic framework is designed and tested on both benchmark and real-world problems. The experimental results demonstrated the effectiveness and generality of the proposed hyper-heuristic framework.