Control and Signal Processing Lab

Friday Educational Activities

The following courses are designed for all members of CSP Lab from postdocs through to full staff members, but are open to all to attend.

Presenter Course
Prof Jonathan Manton Luenberger. Optimization by Vector Space Methods
Prof Jonathan Manton Elliot et al. Hidden Markov Models: Estimation and Control
Dr Rika Hagihara An introduction to symbolic dynamics
Dr Pierre-Olivier Amblard Introduction to RKHS and their applications in signal processing
Prof Jonathan Manton “33 Miniatures” Course
Prof Jonathan Manton Differential Topology
Prof Jonathan Manton Differential Geometry

Past Courses

Luenberger. Optimization by Vector Space Methods

Presenter

Jonathan Manton

Time & Venue

1pm to 2pm, Greenwood Theatre, Level 1, EEE Building

Synopsis

This book is a classic in its field. It uses functional analysis to study infinite-dimensional optimisation problems.

Elliot et al. Hidden Markov Models: Estimation and Control

Presenter

Jonathan Manton

Time & Venue

2.30pm to 3.30pm, Greenwood Theatre, Level 1, EEE Building

Synopsis

This book will be taught in a way which makes it an introduction to stochastic filtering. The advantage of Hidden Markov Models is that they are easy to understand yet quite powerful because they are nonlinear. Optimal filters will be derived using the reference probability method.

An introduction to symbolic dynamics

Presenter

Dr Rika Hagihara

Time & Venue
  • 11am to 12pm (noon), Greenwood Theatre, Level 1, EEE Building.
  • 1pm to 2pm, Greenwood Theatre, Level 1, EEE Building.
Synopsis

In symbolic dynamics Markov measures are described by a finite amount of data and can be realized naturally on subshifts of finite type. However, hidden Markov measures, the images under factor maps of Markov measures, do not always inherit properties of their pre-images.

In this lecture series we will first review the setting of symbolic dynamics: subshifts, shift transformation, and maps between subshifts. Our focus will be on subshifts of finite type and we will study their graph representations and topological entropy. We will then introduce Markov measures and consider when a given measure is hidden Markov. Along the way ideas from topological dynamics will be mentioned.

Working knowledge of point-set topology and linear algebra are prerequisites. Concepts and properties from measure theory will be explained as needed.

Introduction to RKHS (Reproducing kernel Hilbert space) and their applications in signal processing

Presenter

Dr Pierre-Olivier Amblard

Topics to be covered:

  • Intro to the math of RKHS
  • Simple applications in Machine learning (the representer theorem, regression, support vector machines)
  • Random variables with values in RKHS, application to dependence (or in-) measures
  • A Bayesian point of view

“33 Miniatures” Course

Presenter

Prof Jonathan Manton

Book: “Thirty-three miniatures: Mathematical and Algorithmic Applications of Linear Algebra”.

The motivation for the course is that sometimes it is a clever trick (that is easy to understand but difficult to think of in the first place) that is required to improve the performance of an engineering algorithm, and becoming familiar with a range of known tricks increases the chances of being able to come up with the right trick for the problem at hand.

Differential Topology

Presenter

Prof Jonathan Manton

Book: “Differential Topology” by Victor Guillemin and Alan Pollack.

Book: “A Topological Introduction to Nonlinear Analysis” by Robert F. Brown

Target Audience: Students wishing to learn some rigorous mathematics relevant to engineering.

Differential Geometry

Presenter

Prof Jonathan Manton
Book: Methods of Information Geometry by Shun-ichi Amari and Hiroshi Nagaoka.


Contact Us

Prof Jonathan Manton

Director, Control and Signal Processing Laboratory

E: