Mathematical and Statistical Methods for Data Science

The school aims to introduce students to the mathematical and statistical underpinnings of some of the latest Data Science methods that seek to address the challenge of Big Data analysis. There will be an emphasis on matricial methods both in modelling and in numerical computations. The topics covered will include Randomized Numerical Linear Algebra, Deep Generative Models, Bayesian Nonparametric Models and their Asymptotic Properties, Modern Graphical Models and High- Dimensional Statistics Based on Random Matrix Theory.

The courses will start by introducing randomized numerical linear algebra, deep generative and modern graphical models, fundamental for problems of interest in machine learning and statistical data analysis. For all lectures there will be dedicated exercises with practical problems (in particular with Python) to be solved by the students under the experts’ guidance.

There will be also be seminars by workshop participants to showcase their works as well as invited talks by Data Science experts in South Africa to expose participants to real world problems being worked on by academics and industry practitioners.