Sakyasingha Dasgupta  Sakyasingha Dasgupta photo       

contact information

Research Staff Member - Embodied Cognition & Machine Learning
IBM Research - Tokyo
  +81dash33dash808dash6026

links

Professional Associations

Professional Associations:  Deutsche Physikalische Gesellschaft (DPG)  |  IEEE  |  IEEE Systems, Man, and Cybernetics Society  |  International Neural Network Society  |  Organization for Computational Neurosciences

more information

More information:  LinkedIn  |  ORCID  |  ResearchGate

profile


 

For more upto date information check external personal page.  

Short Bio

Sakyasingha Dasgupta received his Masters in Artificial Intelligence, specializing in Machine Learning from  the University of Edinburgh, U.K., and his doctoral degree (Dr. rer. nat)  in Physics of complex systems from the University of Goettingen, Germany. During this time he was jointly affiliated with the Bernstein Center for Computational Neuroscience, Goettingen and the Max Planck Institute for Dynamics and Self-organization. His doctoral thesis focused on "temporal information processing and memory guided behaviors using recurrent neural networks" and the interaction of various plasticity mechanisms, with application to complex neuro-robotic systems.

He is currently a Research Staff Member at IBM Research - Tokyo, Cognitive Computing group working on deep generative models, energy-based machine learning and reinforcement learning, applied towards finance, IoT and embodied cognition in robotic systems.

His diverse research interests include nonlinear dynamical systems, recurrent neural networks, reinforcement learning, time-series modeling, computational neuro-dynamics, neuro-robotics, optimization & control. Overall, he is interested in statistical machine learning as a field and in searching for novel theoretical frameworks to build adaptive intelligent systems.

He is currently leading an IBM Research-Tokyo Far Reaching Research project on a sequential decision making learning framework called "ACT - Actions with Closed-loop Training", formely known as closed-loop experience building embodied robots (CLEBER).