Data Science and AI Methodologies - Bristol Turing Fellows Projects Seminar Series
Professor Peter Flach, Dr Song Liu, Dr Patrick Rubin-Delanchy
Data Science and AI Methodologies
Professor Peter Flach, Professor of Artificial Intelligence, Department of Computer Science, University of Bristol (10:30am - 11:00am)
Towards a Measurement Theory for Data Science and Artificial Intelligence
Measurement concepts are underdeveloped in data science and AI, in at least the following senses: (i) a wide-spread under-appreciation of the importance and effects of measurement scales; and (ii) the fact that in most cases the quantity of interest is latent, i.e. not directly observable. This project seeks to make fundamental advances in our understanding of capabilities and skills of models and algorithms in data science and AI , and how to measure those capabilities and skills.
Dr Song Liu, Lecturer, School of Mathematics, University of Bristol (11:00am -11:30am)
Statistical Inferences of Intractable Models using Density Ratio Estimation
Implicit generative models have been successfully applied for simulating artificial examples of images and texts in recent years. Such a generative procedure does not specify an explicit probabilistic model. It is shown to generate samples from highly complicated distributions successfully. The fitting procedure of implicit generative models is closely related to estimating the ratio between two probability density functions. The success of density ratio estimation on fitting implicit generative models gives hope to use this technique to infer other complicated statistical models. This research applies density ratio estimation to solve statistical inference problems and shows how this procedure can work with various intractable models. We also demonstrate how this inference procedure can guide the implicit generative models simulating interesting and interpretable examples.
Dr Patrick Rubin-Delanchy, Associate Professor in Statistical Science, School of Mathematics, University of Bristol (11:30 - 12:00)
Analysis of jointly embedded cyber-security graphs
There is more information in graphs, and other complex discrete data structures than is currently exploited in data-driven approaches to cyber-security. This project aims to set out a principled and tractable framework, at the intersection of Mathematics, Statistics and Computer Science, to allow more effective use of multi-modal cyber-security data and work with industry and government partners to see prototype algorithms and software deployed.