Contextual probability: concepts, implementations and applications
ABSTRACT:
Contextual probability is a method for nonparametric probability estimation without using any model assumption. If we can estimate probability for any data instance, we can classify the instance in a Bayes optimal way. Contextual probability is also a method for reasoning with uncertainty, so it can be used for decision making and potentially general problem solving.
Neighbourhood counting is a methodology for measuring data similarity. which applies to any type of data as long as neighbourhood can be defined. For a given set of data instances, the neighbourhood counting similarity is the number of all neighbourhoods that cover all data instances in the given set. Neighbourhood counting has been applied to multivariate data, sequence, tree and graph, resulting in elegant similarity measures.
This lecture will cover the concept of contextual probability, its relationship with the frequentist probability, how to calculate contextual probability for different types of data through neighbourhood counting. An outlook for future work will also be given.
BIOGRAPHY:
Professor Hui Wang obtained his BSc in Computer Science and MSc in Artificial Intelligence from Jilin University (China), and his PhD in Artificial Intelligence from the University of Ulster. He is currently Professor of Computer Science. His research interests are knowledge representation, reasoning, learning and enumerative combinatorics, and applications in multimodal video search and spectral data analytics for food authentication and virus detection. He has over 270 publications in these areas.
He played a principal role in the development of an algebraic framework for machine learning, Lattice Machine. He proposed the original concept of contextual probability, which can be used for reasoning with uncertainty, probability estimation and machine learning. He also proposed a generic similarity measure, neighbourhood counting, and its specialisations on multivariate data, sequences, tree and graph structures. The contextual probability and neighbourhood counting similarity bear strong similarity to kernel methods.
He is an associate editor of IEEE Transactions on Cybernetics, and an associate editor of The Computer Journal. He is the Chair of IEEE SMCS Ireland Chapter since 2009, and a member of IEEE SMCS Board of Governors (2010-2013). He is principal investigator of a number of regional, national and international projects in the areas of virus detection (EPSRC funded VIPIRS 2020-2022), image/video analytics (EPSRC funded MVSE 2021-2023, EU Horizon 2020 funded ASGARD 2016-20, Horizon 2020 funded DESIREE 2016-19, EU FP7 funded SAVASA 2011-14, Royal Society funded VIAD 2014-16), text analytics (Invest Northern Ireland funded DEEPFLOW 2010-14, Royal Society funded BEACON 2009-11), and intelligent content management (EU FP5 funded ICONS 2002-05); and is co-investigator of several other funded projects.
Venue:
LEJ Computer labs (for physical participants)
Registration deadline: 20th October 2020
Registration Link: https://docs.google.com/forms/d/e/1FAIpQLSdeYpsZXV1G7LKYXdtWuLdT2IcBC0szwfCj8DIooPgjZ4aWxQ/viewform
Zoom Details:
Oct 26, 2020
Topic: Contextual probability: concepts, implementations and applications
Time: Oct 26, 2020, 02:00 PM Islamabad, Karachi, Tashkent
Join Zoom Meeting
https://us02web.zoom.us/j/89297334635?pwd=eG5iWDlYUUhJdTl0cmtWbXpiRHJXZz09
Meeting ID: 892 9733 4635
Passcode: 370623
Oct 27, 2020
Topic: Contextual probability: concepts, implementations and applications
Time: Oct 27, 2020, 02:00 PM Islamabad, Karachi, Tashkent
Join Zoom Meeting
https://us02web.zoom.us/j/84239405514?pwd=cmV5TEZBdzNFY09xTm50YkU3UUhzZz09
Meeting ID: 842 3940 5514
Passcode: 620276
Coordinated By
Prof Dr M Sadiq Ali Khan ,
Chair IEEE Educational Activities Karachi Section
msakhan@uok.edu.pk
03002334578
Dr. Hina Siddiqui
Assistant Professor
ICCBS, Karachi, Pakistan
hinahej@gmail.com
Ms Khazima Muazim
Program Manager
khazima@comstech.org
0092519220681-3
