Talks in Optimisation, AI, and Dynamical Systems (Day 1)

In short

Date:

7 March 2022

Time:

14:00 - 15:00

Location:

THUAS, OV K.47 Kennedy Room

Within the programme of THUAS new UNESCO Chair in AI & Data Science, we are delighted to announce a series of talks in Optimisation, AI and Dynamical Systems by world-renown expert Optimization and Data Modelling, Prof. Vassiliadis, taking place in March 2022.

Optimisation Overview & Novel Machine Learning Deep-Learning for huge scale problems

Part 1: Overview of Optimisation

An overview of optimisation theory: applications, models and methods in relation to science and technology. The presentation proceeds from fundamental basic principles to examine cutting-edge, state-of-the-art applications of optimisation from the traditional petrochemical industry to the modern arena of biomedical applications and the life sciences at large.

Part 2: Novel Machine Learning / Deep-Learning training methodology for huge scale problems

Machine learning and multiscale hierarchical ANN decomposition for the efficient implementation of fast-converging block coordinate descent procedures. A global sensitivity measure is introduced for structured Artificial Neural Network models, such as the ones employed in Deep Learning implementations. According to this, entire compartments of an ANN can be lumped together into one sub-structure (submodel) which is the evaluated for its impact, "structurally" by being present, on the objective function value of the fitting (Least Squares error of fit, for example). According to this measure, the algorithm prioritizes the adaptation of the weights of certain layers and neurons over others. It is shown computationally that this approach can lead to significant speed-ups for the solution of very large scale ANN implementations, particularly in situations where unsupervised learning and ANN architecture adaptation may be required in real-time applications.

Next talk

The next talk will be held on Thursday March 10th, 1.00 - 3.00 p.m.

About Professor Vassilios (“Vassilis”) S. Vassiliadis

Professor Vassilios VassiliadisProf. Vassiliadis’ research interests lie in the development and application of optimisation and simulation tools in engineering and scientific domains. His research field is mainly focused in Process Systems Engineering (PSE). PSE is at the cusp of applied mathematics, engineering, science and computing, with theoretical and practical applications ranging from the heavy-duty petrochemical industry to personalized medicine.

His favourite quote is from Carl Sagan’s “Cosmos”, the 1980’s TV series: “to accept the truth over our dearest illusions is the heart of science”. He also adds a bit of his own: “this is the essence of accepting our existence in the here and now, whatever these may be, and to be able to see what lies ahead of us without our own lies blurring the view...!”. He is the leading author of the new book "Optimisation for Chemical and Biomedical Engineering", Cambridge University Press, 2021. See for more information Linkedin.

About the UNESCO Chair in AI & Data Science