Machine Learning with a Year in Industry MSc

Royal Holloway, University of London

This module will describe the key principles of academic integrity, focusing on university assignments. Plagiarism, collusion and commissioning will be described as activities that undermine academic integrity, and the possible consequences of engaging in such activities will be described. Activities, with feedback, will provide you with opportunities to reflect and develop your understanding of academic integrity principles.

In this module you will develop an understanding of the principal methods of the theory of stochastic processes, and probabilistic methods used to model systems that exhibit random behaviour. You will look at methods of conditioning, conditional expectation, and how to generate functions, and examine the structure and concepts of discrete and continuous time Markov chains with countable state space. You will also examine the structure of diffusion processes. Students will take either Inference or Applied Probability.

In this module you will develop an understanding of the basics of algorithmic thinking and problem solving using programming. You will become familiar with using the Java programming language, examining particular features and constructs as well as basics of object-oriented programming. You will use these to solve specific algorithmic tasks and evaluate programming solutions.

This module covers algorithm-independent machine learning; unsupervised learning and clustering; exploratory data analysis; Bayesian methods; Bayes networks and causality; and applications, such as information retrieval and natural language processing. You will develop skills in data analysis, including data mining and statistics.

This course is designed to enhance your awareness of the many ethical implications of working with advanced technology. The course recognises that the ethical issues in computing and AI come to the forefront through developments in technology, bringing new responsibility for novel ethical, social, and legal implications of technology almost on a daily basis.

The individual project provides you will the opportunity to demonstrate independence and originality, to plan and organise a large project over a long period, and to put into practice some of the techniques you have been taught throughout the programme.

In this module you will develop an understanding of the mathematical theory underlying the main principles and methods of statistics, in particular, parametric estimation and hypotheses testing. You will learn how to formulate statistical problems in rigorous mathematical terms, and how to select and apply appropriate tools of mathematical statistics and advanced probability. You will construct mathematical proofs of some of the main theoretical results of mathematical statistics and consider the asymptotic theory of estimation. Students will take either Inference or Applied Probability.

In this module you will develop an understanding of modern machine learning techniques and gain practical experience in developing machine learning systems. You will look at the main advantages and limitations of the various approaches to machine learning and examine the features of specific machine-learning algorithms. You will also consider how the ideas and algorithms of machine learning can be applied in other fields, including medicine and industry.

In this module you will develop an understanding of the on-line framework of machine learning for issuing predictions or decisions in real-time. You will learn about protocols, methods and applications of on-line learning, covering probabilistic models based on Markov chains and their applications, such as PageRank and Markov Chain Monte-Carlo. You will examine the time series models, exploring their connections with Kalman filters, and learning models based on the prequential paradigm, including prediction with expert advice, aggregating algorithm, sleeping and switching experts. You will also consider universal algorithms, their application to portfolio theory, and how prediction within a confidence framework is achieved.

In this module you will learn how to use MATLAB (Matrix Laboratory) and WEKA (Waikato Environment for Knowledge Analysis) as tools for machine learning and data mining. For MATLAB, you will develop an understanding of how to input and output data using vectors, arrays and matrics; learn techniques in data visualization, including plots in 2 and 3 dimensions, scatter plots, barplots, and histograms; and learn how to implement concepts from linear algebra and statistics, including probability and matrix decompositions. For WEKA, you will develop an understanding of how to use the software as a tool for training and testing, predicting generalisation performance, and cross-validation; and learn how to implement decision trees, Bayes classifiers, and clustering methods.