Project Title

Learning for materials discovery and design


Researcher

Isaac Tamblyn, University of Ontario Institute of Technology


Supported by

SOSCIP, IBM Canada Ltd., FedDev

Advanced Manufacturing

A UOIT researcher is combining machine learning with computational modelling to accelerate the discovery of new materials important for renewable energy.

Prof. Isaac Tamblyn is using computer vision and AI to reduce the cost and complexity associated with modelling ‘designer’ materials on a computer. Doing so may improve the efficiency and enhance the quality of materials vital to aerospace and healthcare industries.

“Knowing only the elemental composition of a material, it’s now possible to simulate many of its properties (e.g. electrical conductivity, colour, how strong it is),” he explained.

“If we really want to do bottom-up design of a material, we need to be able to find a cheap, easy way to model them.”

Using the BGQ and Cloud, he is able to train computers to make accurate approximations of difficult problems in material science while building an extensive material database.

The training approach is similar to Google’s object recognition algorithms – where computers have been trained to recognize the faces of users in photographs. 

“If you want to teach a computer the difference between cats and dogs, you need a lot of pictures of each. The same is true with materials. In order to develop approximate models of materials, we need to generate really accurate data first.”

“If we really want to do bottom-up design of a material, we need to be able to find a cheap and easy way to model them.”

Isaac Tamblyn
UOIT

The project could support important breakthroughs by others in materials design, which is critical since not everyone has access to supercomputers and some
calculations take as many as five million computing hours.

His team is now investigating how to make carbon a better catalyst for water-splitting, a process that could be used to develop clean, renewable energy capable of powering electronic devices and fueling vehicles.

Nataliya Portman, a former post-doctoral fellow for the project, gained valuable skills as a data scientist through her approach combining mechanics and machine learning.

“This project allowed me to develop technical skills in HPC, GPU programming and big data management, and advance my knowledge of deep learning methods,” she said. Portman now works as a lead data scientist for a company in Toronto. Shiva Gholami is currently working as the team’s post-doctoral fellow.

When the research is complete, Prof. Tamblyn is hoping to perfect a validated machine learning model capable of rapidly determining the electronic structure and physical characteristics of catalytic surfaces. 

Prof. Tamblyn’s students will use the model to launch a start-up company in 2018 targeted at industries that can benefit from high-tech material design, such as defence, aerospace and health.