The lecture is free to attend and open to all, but registration is required in advance.
Tea, coffee and biscuits will be served prior to the lecture in the SAF ground floor foyer.
Register to attend by visiting: bit.ly/BharathInaugural
Abstract:
The field of artificial intelligence has taken enormous strides over the past few years. Some of this rapid advance can be attributed to the ready availability of computational power, and raw material in the form of data. But we should not forget the role of fundamental neuroscience research in the remarkable demonstrations that have characterised recent progress in AI, partly through using layered networks of artificial neurons.
In this talk, I will return to the source of these ideas: biological neurons, and their responses to stimuli. In so doing, I will draw attention to the early origins of convolutional neural networks and describe several ways in which the study of biological vision has stimulated ideas in machine learning, computer vision, and the representation of spatial and temporal data.
Finally, I will discuss how the technology of visual search either draws on ideas from, or parallels, findings in biological vision.
Biography:
Anil Anthony Bharath (Fellow of the Institute of Engineering & Technology) is Professor of Biologically-Inspired Computation and Inference in the Department of Bioengineering at Imperial College. He holds a first degree from UCL (Elec Eng) and a PhD from Imperial College London. Motivated by discovering parallels between biological and machine intelligence, his current research interests are around architectures of convolutional neural networks; his lab has recently published work on generative adversarial networks, and deep reinforcement learning for visuo-motor control.
He is academic founder of Cortexica Vision Systems, a company that applies biologically-inspired algorithms and deep machine learning for “visual search” – the ability to query databases based on the visual appearance of items. Cortexica was one of the first companies to provide commercial visual search services using cloud-based Graphical Processing Unit (GPU) architectures. Anil has authored or co-authored papers in the areas of pattern recognition, deep learning and computer vision, and has worked on the applications of inference techniques to a wide variety of domains, including medical, non-medical data.