Long Tran-Thanh

I am currently a research fellow on the ORCHID project since October 2011. Before that I was a EPSRC funded PhD student, also at AIC (formerly IAM), and was  supervised by Nick Jennings and Alex Rogers. I succesfully defended my thesis on March 8,  2012 (examiners were Nicolò Cesa-Bianchi and Jörg Fliege).  

My PhD thesis focuses on the performance analysis and appilcations of budget-limited multi-armed bandit (MAB) models. In these models, the learner’s actions are costly and constrained by a fixed budget. In this MAB version, an optimal exploitation policy may not be to repeatedly pull the optimal arm, as is the case in other variants of MAB, but rather to pull the combination of arms that maximises the agent’s total reward within the budget. This difference from existing bandit models means that new approaches to maximising the total reward are required. In so doing, I have proposed a number of simple, but efficient, algorithms to tackle this MAB problem. In more detail, I provided upper and lower bounds for their performance regret. In addition, I have applied these results to the domain of wireless sensor networks, in order to achieve efficient long-term information collection from the surrounding environment. 

I was also pursuing another PhD degree in Electrical Engineering at Budapest University of Technology and Economics, Hungary (2007 - ), which is suspended at the moment (but will be hopefully finished by 2012). The research topic here is reliable wireless communictation within ad hoc and mesh networks. 

Another research topic that I am currently investigating is the congestion game models from the domain of game theory. In particular, I analyse the existence of pure strategy profile Nash equlibria (PSNE) in integer splittable congestion games. This is a joint work with Maria Polukarov from University of Southampton and Archie Chapman from University of Sidney Business School. So far, we have provided sufficient  conditions to the existence of PSNE in these games. Possible future work includes complexity and price of anarchy analysis of the model. 

Recently, I am interested in applying online learning techniques to a variety of research areas, such as: decentralised coordination under uncertainty, expert crowdsourcing, and incentive engineering.

MSc in Computer Science at Budapest University of Technology and Economics, Hungary (2002 - 2007).

Highschool diploma at Fazekas Mihaly Grammar School, Budapest, Hungary (1996 - 2002) 

AAMAS 2009, WCNC 2009, AAAI 2010, ICC 2010, COLT 2011, SAGT 2011