IJCAI-17 Tutorial: Energy-based machine learning       

links

Takayuki Osogami photo

IJCAI-17 Tutorial: Energy-based machine learning - Organizers


Sakyasingha Dasgupta
Research Staff Member
IBM Research - Tokyo
Webpage

Brief Biography: Sakyasingha Dasgupta received his Dr.rer.nat. in Physics of Complex Systems from the International Max Planck Research School, Georg-August University Goettingen, Germany for his work on temporal information processing with recurrent neural networks. Prior to that he obtained his Masters in Artificial Intelligence with a specialisation in Neural Computation from the University of Edinburgh in November 2010.  Based on his expertise in neural networks and statistical machine learning, he is working with energy-based models and their biological motivations for generative learning and reinforcement learning. An area of focus for him as part of a large government funded project at IBM Research are dynamic Boltzmann machines, for learning time-series and sequential decision making in partially observable environments.



Takayuki Osogami
Research Staff Member
IBM Research - Tokyo
Webpage

Brief Biography: Takayuki Osogami received his Ph.D. in Computer Science from Carnegie Mellon University in August 2005 for his work on analysis of stochastic models. Based on his expertise in applied probability, he has recently been working on modeling and learning with Boltzmann machines or energy-based approaches. Two representative streams of his recent work are the dynamic Boltzmann machines for learning time-series and the deep choice model for learning human choice. Most importantly, he has proposed the dynamic Boltzmann machine in Osogami and Otsuka (2015); http://www.nature.com/articles/srep14149.




Tutorial slides

Box folder




Tutorial date & venue

  • Date: August 21
  • Time: PM1 and PM2
  • Venue: Melbourne convention & exhibition center Room 211