Ph.D. in Robotics08/2008 to 06/2013
Thesis: Interactive Learning for Sequential Decisions and Predictions. [pdf]
Carnegie Mellon University, Pittsburgh, PA, USA

M.Sc. in Robotics08/2008 to 07/2011
For completing the course, research, writing and speaking qualifiers of the Ph.D. in Robotics program.
Carnegie Mellon University, Pittsburgh, PA, USA

M.Sc. in Computer Science09/2006 to 08/2008
Thesis: Model-Based Bayesian Reinforcement Learning in Complex Domains. [pdf]
McGill University, Montreal, Qc, Canada

B.A.Sc. in Computer Science09/2004 to 05/2006
Specialization in Software Engineering
Laval University, Quebec city, Qc, Canada

Technical Degree in Computer Science08/2001 to 05/2004
CEGEP Francois-Xavier-Garneau, Quebec city, Qc, Canada

• Developing new behavior prediction system applying machine learning to predict the future behaviors of other vehicles, pedestrians and cyclists on the road. (C++)

• Developed new approaches to structured prediction, imitation and reinforcement learning.
• Analyzed theoretical properties of new approaches, including regret and sample complexity.
• Implemented new approaches on several applications in control and computer vision:
• Imitation Learning for a 3D kart racing video game and Super Mario Bros. (C++, Java)
• LADAR 3D point cloud classification and 3D geometry estimation from 2D image. (C++)
• Learning to control a quadrotor through forest environments from camera (C++, Python)
• Implemented inverse kinematics and balance control for a six-leggeg mining robot. (C,C++)

• Implemented my Structured Prediction learning algorithm (DAgger) into John Langford’s open-source learning software Vowpal Wabbit (VW). (C++)
• Implemented contextual bandit algorithms in VW for integration with Structured Prediction approaches and preliminary applications on Bing search engine data. (C++)
• Developed and analyzed a new improved adaptive gradient descent algorithm.
• Implemented improved adaptive gradient descent algorithm in VW. (C++)

• TA for the class: 16-899C Adaptive Control and Reinforcement Learning, taught by Drew Bagnell.
• Helped prepare and grade homeworks, midterm exam and final class projects.
• Answered students’ questions during weekly office hours.
• Maintained class website and message board for announcements and posting homeworks.
• Gave lecture on TD methods in RL.

• Investigated new approaches for model-based Bayesian reinforcement learning in partially observable, continuous and structured environments.
• Developed new efficient planning and inference algorithms for model-based Bayesian reinforcement learning in such domains.

• Implemented various search, planning and scheduling algorithms in the RobocupRescue environment, which is a real time, dynamic and partially observable multiagent disaster management problem. (Java)
• Implemented and compared the performances of different approximate POMDP solvers such as PBVI, Perseus, RTDP-BEL, QMDP and RTBSS in 2 standard benchmarks (Tag and Rock Sample). (Java)
• Developed a new online search algorithm to solve POMDPs.
• Developed new algorithms for equilibrium learning in multiagent incomplete information repeated games.

• Analysed and conceived manufacture management software, such as project, time sheet, expense and JIT inventory management modules. (C#, ASP.NET, VB6)

• Analysed and conceived the career and placement service system and website. (VB6, HTML, Javascript)

NSERC Canada Graduate Scholarship (CGS D)2008
Tenured at Carnegie Mellon University.

FQRNT Doctoral Research Scholarship2008
Not tenured.

Dr. and Mrs. Milton Leong Graduate Student Award2007
Tenured at McGill University.

McGill Graduate Studies Fellowship2007
Tenured at McGill University.

McGill Recruitment Excellence Fellowship2006
Tenured at McGill University.

FQRNT Master's Research Scholarship2006
Tenured at McGill University.

NSERC Postgraduate Scholarship (PGS M)2006
Tenured at McGill University.

NSERC Undergraduate Student Research Award2005
Tenured at Laval University.