15 November 2021. Monday. 3PM. London Time. Hybrid (zoom and face-to-face). Room Colln CG15.
Title Explainable AI and Its Application to Medical Diagnosis
Speaker Xiaohong Gao, Department of Computer Science, Middlesex University, London
Abstract While recent advances in AI, in particular deep learning (DL) technology, have achieved state of the art results in many fields, the number of parameters involved in a convolutional neural network (CNN) (> 10 millions) has reached a limit beyond human’s comprehension, arriving at a black box issue. This talk will address what can or cannot a deep learning system do, aiming at clearing some misconceptions on DL while maximising the benefit this technique brings to the mankind. Specifically, this talk will present a few applications of transparent/explainable/interpretable deep learning systems in medical domain.
Bio Xiaohong (Sharon) Gao has education background in Applied Mathematics (BSc) and Computer Graphics (MSc). Her PhD work conducted in Loughborough University has led to the standardisation of a colour appearance model CIECAM02 by the International Commission on Illumination ( Commission Internationale de l’Eclairage) to predict colours from human colour perception point of view. Since then, she worked as a post-doc at two NHS hospitals for 4 years, St. Marys hospital at Imperial College and Addenbrookes Hospital at the University of Cambridge. After joining in Middlesex University as an academic, she has led several research projects on medical applications funded by EPSEC, EU and JISC. Currently she is working on projects of AI in early cancer detection (CRUK) , AI in Pathology (CRUK) and AI in super-resolution microscopic images (The Royal Society).
Meeting ID: 668 413 8396