Explainability in finance
Thursday 31 Oct 2019
Time: 13:00 - 18:00
Bio
Chris Russel is a Reader (equivalent to Associate Professor) at the Centre for Vision, Speech and Signal Processing at the University of Surrey in machine learning and computer vision; an ELLIS fellow, and with Sandra Wachter and Brent Mittlestadt co-founder of the GET (Governance of Emerging Technology https://www.oii.ox.ac.uk/research/governance-of-emerging-technologies/) programme a cross-disciplinary and cross institute research group looking at the socio-technical issues arising from new technology and proposing legal, ethical and technical remedies
Research interests
Chris's primary focus lies in mathematical modelling and optimisation, and the application of these tools to a wide range of cross-disciplinary problems, predominately in societal machine learning and interpretability, and computer vision. The key theme underlying his research, from his PhD onwards, is a search to express what was previously inexpressible i.e. to find mathematical models that capture unused cues about the world while still keeping the optimisation tractable.
More recently, he has been focusing on understanding the internal state of algorithms and what this state can tell us about the world and society. Answering this question has led to a range of publications in diverse fields such as 3D reconstruction from images; urban analytics; algorithmic fairness; and explaining black box functions. His recent work which proposed a new approach for explaining black-box functions is cited in the guidelines to the GDPR and has been implemented as part of the what-if tool in Google’s TensorFlow.
