In depth - MRes Machine Learning and Big Data in the Physical Sciences

This core module will provide the foundational understanding of the statistics behind large scale data analysis the physical sciences. The module will cover key concepts in statistical modeling, and statistical inference that are essential for understanding the methodologies behind machine learning applications to the experimental sciences. This core course is intended to provide the statistical background to the applied core course taught in the same term. 

Module spec can be found here.  

For further information see this page.

This core module will provide the hands on experience of techniques required to analyse large data sets. The course will be taught in the Python computing language and will use standard packages such as numpy, scipy, matplotlib, pandas, Scikit-Learn, Keras and Tensorflow. The course assumes no prior knowledge of Python or any of these packages. You will learn how to implement the different techniques required to analyse data (statistical techniques and machine learning techniques) through working through examples and then analysing different data sets.

Module spec can be found here.

For further information see this page.

This module will cover:
 - Introduction and overview of accelerators in computing (covering typical architectures, GPU, FPGA... their features and differences)
 - Mapping applications to architecture (what architecture best suits the problem/algorithm)
 - Performance modelling (making a C++/Excel based model of the algorithm to understand in detail and model implementation in accelerator)
 - Implementation (examples of implementation of algorithms, including typical AI cases)

Module spec can be found here.

For further information see this page.

Elective modules:

either

Accelerated processing for Big Data analysis + another 7.5 ECTS level 6 or 7 module offered within Physics (or Maths/Engineering if appropriate) 

or 

one 7.5 ECTS and one 5 ECTS Level 6 or Level 7 module offered within Physics (or Maths/Engineering if appropriate) 

or 

two 7.5 ECTS Level 6 or Level 7 modules offered within Physics (or Maths/Engineering if appropriate) 

UG level 6 and 7 courses are listed here

Modern problems in statistics require sampling from complicated probability distributions defined on a variety of spaces and setups. In this module we will visit popular advanced sampling techniques, such as Importance Sampling, Markov Chain Monte Carlo, Sequential Monte Carlo. We will consider the underlying principles of each method as well as practical aspects related to implementation, computational cost, and efficiency. By the end of the module the students will be familiar with these sampling methods and will have applied them to popular models, such as Hidden Markov Models, which appear ubiquitous in many scientific disciplines.

Introduction to Statistical Finance presents the fundamental concepts of quantitative finance and statistical methods that are widely used to analyse financial data. It starts off with a concise introduction to financial markets, proceeding then to the basic principles of derivatives pricing and risk measurement. Subsequently, it develops statistical models for financial time series. Finally, it explains how such models are estimated, how their goodness of fit can be assessed and how they can be used in forecasting.

Nonparametric Statistics

Nonparametric methods aim to provide inference under weaker assumptions than conventional parametric methods. In this module, students will apply modern techniques to a variety of problems, such as estimating distribution functions, density estimation and nonparametric regression.

In this module we will develop models and tools to analyse complex and high dimensional datasets as these arise in different application fields as for example the genetics field. This will include statistical and machine learning techniques for multiple testing, penalised regression, clustering, dimensionality reduction and visualisation. The module will cover both Frequentist and Bayesian statistical approaches as well as some modern machine learning unsupervised and supervised approaches.