Machine Learning

Towards a Competitive 3-Player Mahjong AI using Deep Reinforcement Learning
Presented at the 2022 IEEE Conference on Games (CoG).
24 Aug 2022 6:50 PM — 7:00 PM Virtual
Xiangyu Zhao, Sean B. Holden
Towards a Competitive 3-Player Mahjong AI using Deep Reinforcement Learning
Task-Agnostic Graph Neural Network Evaluation via Adversarial Collaboration
Master’s Dissertation Presentation.
8 Jun 2022 9:10 AM — 1:00 PM Computer Laboratory, University of Cambridge
Xiangyu Zhao
Task-Agnostic Graph Neural Network Evaluation via Adversarial Collaboration
Task-Agnostic Graph Neural Network Evaluation via Adversarial Collaboration
Master’s Dissertation supervised by Prof Pietro Liò, Dr Dominique Beaini and Hannes Stärk – We propose GraphAC (Graph Adversarial Collaboration), a conceptually novel, principled, task-agnostic, and stable framework for evaluating GNNs through contrastive self-supervision, without the need of handcrafted augmentations.
Xiangyu Zhao, Hannes Stärk, Dominique Beaini, Yiren Zhao, Pietro Liò
27 May 2022
Task-Agnostic Graph Neural Network Evaluation via Adversarial Collaboration
Master’s Dissertation supervised by Prof Pietro Liò, Dr Dominique Beaini and Hannes Stärk – We propose GraphAC (Graph Adversarial Collaboration), a conceptually novel, principled, task-agnostic, and stable framework for evaluating GNNs through contrastive self-supervision, without the need of handcrafted augmentations.
Xiangyu Zhao, Hannes Stärk, Dominique Beaini, Yiren Zhao, Pietro Liò
Towards a Competitive 3-Player Mahjong AI using Deep Reinforcement Learning
We present Meowjong, an AI for 3-player Mahjong (Sanma) using deep reinforcement learning. We define an informative and compact 2-dimensional data structure for encoding the observable information in a Sanma game. We pre-train 5 CNNs for Sanma’s 5 actions, and enhance the major action’s model via self-play RL using the Monte Carlo policy gradient method.
Xiangyu Zhao, Sean B. Holden
Improving Graph Generative Models via Expressive Graph Neural Networks
MEng Graph Representation Learning Mini-Project.
Xiangyu Zhao, Chaitanya K. Joshi, Cristian Bodnar
26 Apr 2022
Improving Graph Generative Models via Expressive Graph Neural Networks
MEng Graph Representation Learning Mini-Project.
Xiangyu Zhao, Chaitanya K. Joshi, Cristian Bodnar
Multi-Agent Deep Q-Learning for the Berry Poisoning Game
MEng Advanced Topics in Machine Learning Coursework.
Xiangyu Zhao
Last updated on 4 Apr 2022
Multi-Agent Deep Q-Learning for the Berry Poisoning Game
MEng Advanced Topics in Machine Learning Coursework.
Xiangyu Zhao
Building a 3-Player Mahjong AI using Deep Reinforcement Learning
We present Meowjong, an AI for 3-player Mahjong (Sanma) using deep reinforcement learning. We define an informative and compact 2-dimensional data structure for encoding the observable information in a Sanma game. We pre-train 5 CNNs for Sanma’s 5 actions, and enhance the major action’s model via self-play RL using the Monte Carlo policy gradient method.
Xiangyu Zhao, Sean B. Holden