[1] Turing Artificial Intelligence World-Leading Research Fellowship project (PI by Samuel Kaski): Human-AI Research Teams - Steering AI in Experimental Design and Decision-Making.
[2] Engineering and Physical Sciences Research Council (EPSRC) Fellowship Project (PI by Anthony Constantinou): Bayesian Artificial Intelligence for Optimal Decision Making.
Under this project, I contributed to the research on causal discovery algorithms from observational data, especially BN structure learning under various assumptions. For example, I worked on the BN structure learning for large-scale/high-dimensional problems with hundreds or thousands of variables. I also contributed to a survey paper on structure learning algorithms. I also contributed to a large-scale validation of 15 mainstream causal discovery algorithms, on noisy data. We investigated the performance of mainstream causal discovery algorithms on noisy data, such as missing data, measurement error data, and data with latent variables. In addition, I contributed to pruning-based model-averaging, knowledge infusion, measurement error based causal discovery algorithms and a study on impact of prior knowledge on causal discovery.
[3] National Natural Science Foundation of China (NSFC) Project (PI by Xiaoguang Gao): Learning Bayesian Networks from Small Data Sets: Theory and Applications.
Under this project, I was responsible for the theory part of the project, which includes the structure and parameter learning of Bayesian networks using insufficient observational data and domain knowledge. To that end, I explored theories, such as particle swarm algorithm, ant colony algorithm, genetic algorithm, convex optimization, linear programming, mini-max algorithm, maximum entropy models, maximum likelihood models, penalized maximum models, isotonic regression, MCMC sampling, free-energy principle, etc.