Current Graduate Student
Hao Cui is currently a PhD student studying with Roni Khardon. He got his M.Sc. and B.Sc. from Beijing Institute of Technology (China). His main interest lies in large action space MDP planning and machine learning.
Associated Publications: [+]
Authors: H. Cui and R. Khardon
International Joint Conference on Artificial Intelligence (IJCAI)
Authors: H. Cui, R. Khardon, A. Fern, and P. Tadepalli.
Proceedings of the AAAI Conference on Artificial Intelligence (AAAI)
Current Research Topics:
Description: Markov decision processes give a mathematical model for agents acting in a dynamic environment. The actions of the agent affect the world, its own state, and whether it is "rewarded" or not. However, the results of actions are not deterministic.The basics of this framework are well understood but the main challenge is in scalability to problems with large state spaces and/or action spaces. Our main interest is in developing efficient agents that learn and act in such domains. Our solutions take advantage of structure (relational or propositionally factored) in the state and action space to yield effective solutions.
This work is partly supported by NSF grants IIS-0936687, IIS 0964457 and IIS-1616280
Code for stochastic planning in large factored state and action spaces is provided through Hao Cui's github site.
For context and description please see: H. Cui and R. Khardon, Online Symbolic Gradient-Based Optimization for Factored Action MDPs, International Joint Conference on Artificial Intelligence (IJCAI), 2016
Download File: https://github.com/hcui01/SOGBOFA