Oliver E Bent  Oliver E Bent photo         

contact information

Research Scientist
IBM Research - Africa
  

links

Professional Associations

Professional Associations:  ACM

more information

More information:  LinkedIn Profile

profile


Oliver completed his MEng and DPhil degrees at the department of Engineering Science, University of Oxford. His masters studies in Information Engineering with the Intelligent Patient Monitoring Group, supervised by Professor Gari Clifford  and his DPhil studies in Autonomous Intelligent Machines and Systems with the Machine Learning Research Group, supervised by Professor Stephen Roberts.

He joined IBM Research in 2013 with the opening of the Africa Research Lab in Nairobi, Kenya.  Initial projects were focussed on the development of technological innovations to compliment the wide-spread adoption of mobile devices and infrastructure in sub Saharan Africa. Working on mobile-based intelligent agents for primary school education and personal management of under supported non-communicable diseases. In these domains developing techniques for Machine learning about learning towards m-health and robust inference from low cost sensors.

Currently Oliver focusses on problems to expand the promise of simulated learning for decision making under uncertainty, specifically with regards to moving discussions past the state of the art in Reinforcement learning, towards demonstrable results in real world domains.

With ongoing work in the following areas

  • An infrastructure to allow for the sharing of dynamical system descriptions;
  • Methods to combine these model descriptions under uncertainty;
  • Machine learning model based approximations or surrogates to computationally expensive simulators; enabling the prototyping of model based reinforcement learning approaches;
  • Distributing the search for optimal policies through Reinforcement learning environment abstractions to global teams of data-scientists in hackathons and competitions.

These approaches are currently being iterated to provide support in high impact national-level decision making tasks under uncertainty, for disease control and planning, with additional extensions to climate impact modelling problems.