Computer Science News
Forging Links Between Neuroscience and Artificial Intelligence
On Monday 19 March, UCL’s Centre for Computation, Mathematics and Physics in the Life Sciences and Experimental Biology (CoMPLEX), in association with UCL’s Neuroscience Domain hosted 200 researchers at an afternoon symposium entitled “Neuro-AI - Forging Closer Links between Neuroscience and Machine Learning.”
In his welcome address to an audience of academics, students and researchers CoMPLEX Director, Dr Guy Moss, explained the need for this meeting: “UCL boasts an exceptional community of Neuroscientists, second only to Harvard on the international stage. UCL is also home to the leading-edge of research in artificial intelligence (AI) and Machine Learning via the UK’s top-ranked Department of Computer Science. The jewel in this crown is the collaboration between the Gatsby Institute and Computer Science, but our hope is to build on this significant foundation and to drive further, bi-directional benefits from collaborations between researchers who share common goals for understanding the basic principles of learning.”
The fields of neuroscience and AI have a long history, and the rates of advance in both areas have increased rapidly. The last 5-10 years have seen a step change in the capability of machine learning, particularly deep learning, that has, in part, been informed by our growing understanding of the brain. In turn, new machine learning tools provide a powerful means to explore how information is processed in the brain, mapping sensory data to neural representations, or representations to behavior, for example. Moreover, it appears that deep networks can function as models of the brain and thus generate and test hypotheses.
Talks on various themes followed, with one of the highlights being contrasting talks about adversarial image perturbation - subtle changes made to images which disproportionately impair their classification. The first talk, by Prof. Ken Harris entitled ‘The Kernel function of the visual cortex”, examined the implications of our understanding of visual processing for adversarial image processing. The second talk by Dr. Lewis Griffin from Computer Science considered adversarial images from both the biological and deep network situation with respect to image classification.
In his closing remarks, Professor Trevor Smart, Chair of the UCL Neuroscience Domain thanked attendees for their perceptive contributions: “It is not often that we have the opportunity to connect specialist research areas from across UCL. I am confident that today has shown what we can possibly achieve in the future by collaboration. It has drawn out new questions that will develop our understanding of these areas, and no doubt will act as a catalyst to bring the scientific community at UCL together.”
Launched in 1998 as a virtual centre, CoMPLEX has been part of the Department of Computer Science since 2016. Medical and life scientists work shoulder-to-shoulder with engineers, physical scientists and computer science experts. This active and intellectually stimulating environment pioneers interdisciplinary research and collaboration.
Find out more about CoMPLEX on their website.
Find out more about UCL’S Gatsby Computational Neuroscience Institute on their website.
Find out more about UCL’s Neuroscience Domain here website.
CoMPLEX and the Neuroscience Domain would like to thank all speakers for their valued contributions:
Dr Caswell Barry, Cell & Developmental Biology, Division of Biosciences, UCL
Prof Janaina Mourao-Miranda, Department of Computer Science, UCL
Prof Kenneth Harris, Quantitative Neuroscience, Institute of Neurology and Department of Physiology, Pharmacology, and Neuroscience UCL
Prof Maneesh Sahani, Theoretical Neuroscience and Machine Learning, Gatsby Computational Neuroscience Unit, UCL
Prof Neil Burgess, Institute of Cognitive Neuroscience, UCL
Dr Lewis Griffin, Department of Computer Science, UCL
Dr Shirley Mark, Motor Neuroscience & Movement Disorders, Institute of Neurology, UCL
Dr David Barber, Department of Computer Science, UCL