Published on: 21-Jul-2017
SCSE applauds its team of SCSE PhD students Chen Haipeng (IGS-LILY scholarship), Xiong Yanhai (IGS-LILY scholarship) and Zhao Mengchen (SCSE Startup scholarship) led by Dr Bo An on their recent outstanding achievement at the Microsoft Azure for Research.
A team of three PhD candidates led by Dr Bo An, Assistant Professor with SCSE, was recently awarded the first place of Microsoft Azure for Research grant prize winners in the Project Malmo Collaborative AI Challenge, which is designed to encourage research relating to various problems in Collaborative AI. The Challenge takes the form of a collaborative mini-game in which players need to work together to achieve a common goal.
The team, which is named HogRider, built an excellent agent to cooperate with the challenge agent and achieve high score in the mini-game. HogRider can detect the type of its cooperator with high accuracy based on Bayesian belief update. It then makes efficient movements with a delicately designed and tuned decision tree. The average game score of Hogrider is remarkably higher than the competitors.
HogRider is selected as the first place winner for the Microsoft Azure for Research grant prize among more than 80 teams comprising postgraduate students from 26 countries. The winner selection criteria include the ability of the agent to achieve high game scores consistently, and the novelty and creativity of the team’s approach. This prize provides additional computing resources worth up to $20,000 USD to the team for their future research efforts.
Said Prof An: “The Malmo testbed reflects all the important characteristics of collaborative AI: incomplete information, partial/noisy observation, co-existence of cooperation and competition, sequential decision making, and more. To solve the challenge, we needed to coherently integrate techniques from different sub-fields together. We’re looking forward to applying our Azure for Research grant to future efforts.”
A take away message from the PhD students: "Like any cooperation in human world, knowing the type of the cooperating agent is essential in an ad-hoc collaboration."
Back to listing