Pores for Thought: Designing Electrodes with Machine Learning and Sorcery

Samuel Cooper, Imperial College London
Deep convolutional generative adversarial networks


The performance of batteries and fuel cells depends on the morphology of their porous electrodes. Characterization of these structures is hard enough, but then using this information to inform improved designs harder still. In this talk I will start by giving an overview of the work going on in the battery space both at Imperial College and the UK more broadly. Then I will describe in detail a few of the projects currently being pursued by my team, including the application of deep convolutional (and now conditional) generative adversarial networks (see figure) to explore the space of possible microstructures.

Link to a recent paper: https://www.nature.com/articles/s41524-020-0340-7

Please use this link to attend the virtual seminar: