Presentations

Machine Learning for Modelling and Decision Making in Complex Physical Domains, at Centre for Artificial Intelligence Decision-making and Action (CAIDA), University of British Columbia, Vancouver, BC, Monday, November 4, 2019

Abstract:

My lab at the University of Waterloo carries out work on a variety of topics within Artificial Intelligence and Machine Learning with a focus on using real world problems to discover computational challenges for modelling of uncertainty, dealing with large or streaming datasets, learning predictive models and enabling decision making.  In this talk I will provide a brief overview of a few ongoing projects where these challenges arise from different sources.  In combustion modelling for energy production and engine design, standard practices...

Read more about Machine Learning for Modelling and Decision Making in Complex Physical Domains
Artificial Intelligence in Support of Physics and Other Complex Domains., at National Research Council, Ottawa, Canada, Thursday, October 17, 2019:
ai4design2019talk.pdf
I was invited up to NRC in Ottawa to give a talk on the work I am doing with Dr. Isaac Tamblyn there on using Reinforcement Learning for challenging problem of Material Design and Discovery. The research question from an AI perspective is essentially "Can we train a chemical lab manager using RL if we have simulations of the individual processes and begin with some foundational expert knowledge?". The talk was part of an local conference launching the new AI... Read more about Artificial Intelligence in Support of Physics and Other Complex Domains.
Machine Learning for Modelling and Decision Making in Complex Physical Domains, at University College Dublin, Dublin, Ireland, Wednesday, September 25, 2019
My lab at the University of Waterloo carries out work on a variety of topics within Artificial Intelligence and Machine Learning with a focus on using real world problems to discover computationally hard challenges for modelling of uncertainty, dealing with large or streaming data, learning predictive models and enabling decision making.  The methods we focus on include Deep Reinforcement Learning, Convolutional and Recurrent Neural Networks, Ensemble Tree methods and manifold based data/dimensionality reduction analysis. Motivation for our work comes from domains such as automotive,... Read more about Machine Learning for Modelling and Decision Making in Complex Physical Domains
Compact Representation of a Multi-dimensional Combustion Manifold Using Deep Neural Networks, at European Conference on Machine Learning (ECML 2019), Wurzburg, Germany, Thursday, September 19, 2019:
ecml2019_talk.pdf

The computational challenges in turbulent combustion simulations stem from the physical complexities and multi-scale nature of the problem which make it intractable to compute scale-resolving simulations. For most engineering applications, the large scale separation between the flame (typically sub-millimeter scale) and the characteristic turbulent flow (typically centimeter or meter scale) allows us to evoke simplifying assumptions–such as done for the flamelet model–to pre-compute all the chemical reactions and map them to a low-order manifold. The resulting manifold is then tabulated...

Read more about Compact Representation of a Multi-dimensional Combustion Manifold Using Deep Neural Networks
Into the Depths : Recent and Future directions for Spatial and Collaborative Reinforcement Learning, at National Research Council, Ottawa, Ontario, Canada, Friday, July 27, 2018:
CAI2018slides_forNRC2018Prez.pdf
On July 27, 2018 I visited the National Research Council labs in Ottawa to chat with Isaac Tamblyn about physics, machine learning, reinforcement learning and quantum computing. This presentation I gave highlights the place of Reinforement Learning in the modern AI research landscape and connects some related work on modelling of combustion and forest fire spread. Read more about Into the Depths : Recent and Future directions for Spatial and Collaborative Reinforcement Learning

Pages