mouse-on-web
 
PEOPLE
PUBLICATIONS
PROJECTS
RESEARCH
          -TOPICS
JOBS
SOFTWARE

STUDENT
       -PROJECTS

CONTACT
 

Here are the main topics that we are interested in:

  • Reinforcement learning (see also the games section)
    - function approximation and RL, algorithms and theory
    - partially observable MDPs
    - applications to portfolio optimization 

  • Games: Multi-players, stochastic games (eg. poker!) and related applications (e.g. bidding)

  • Natural language processing
    - unsupervised learning from large corpora
    - applications to intelligent information retrieval

  • Speech and sound processing
    - speech recognition - dialogue systems, high quality next generation small footprint synthesis (e.g. to be used in PDAs)
    - foreign language teaching (for an inspiring example see Carnegie Speech)
    - CASA - computational scene analysis (understanding complex "acoustical scenes")

  • Large scale machine learning (machine learning on Grids)

  • Applications of machine learning in machine vision
    - handwritten signature verification
    - robust character recognition (e.g. recognizing noisy, corrupted, occluded, etc. characters)

These seemingly distant areas are all closely related to each other. Suprised? Why do we think this? What provides the tie between these areas? The answer should be straightforward: (machine) learning.