51²è¹Ý

Abstract: I will introduce in broad terms the work I and my collaborators have been doing in an attempt to advance science through AI methodologies and to advance AI methodologies through building AI models to answer focused science-driven questions on data. On the science front, I will concentrate on astronomy and cognitive neuroscience. On the AI side I will focus on interpretable and transparent methodologies, ranging from probabilistic modelling of multi-manifold structures in extremely noisy environments (astronomy), to metric learning methods and probabilistic latent variable modelling of spatial-temporal structures (cognitive neuroscience). I will try to relate the message that if we care enough, the interplay between sciences and AI can be healthy and mutually beneficial.

About the Speaker: Peter holds a Chair position in Complex and Adaptive Systems at the University of Birmingham, UK. He is fascinated by the possibilities of cross-disciplinary blending of machine learning, mathematical modelling and domain knowledge in a variety of scientific disciplines ranging from astrophysics to bio-medical sciences. He led an EPSRC-funded consortium of six UK universities on developing a new mathematical framework for personalised healthcare and was awarded three Outstanding Paper of the Year Awards from the IEEE Transactions on Neural Networks and the IEEE Transactions on Evolutionary Computation. Peter (co-)chaired Task Force on Mining Complex Astronomical Data and Neural Networks Technical Committee (IEEE  Computational Intelligence Society).

We look forward to your active participation.