Syllabus for 2012/13

The course starts with the Statistics Foundation Fortnight which starts one week before the official start of term. This is a compulsory two week mathematics refresher and test. Students who subsequently feel uncomfortable with the level of mathematics required may transfer to our sister Machine Learning course.

The course is then split into core and optional modules. Students must take 4 core modules plus 4 additional modules. They will also complete an individual project. The programme is designed to provide a rigorous understanding of Machine Learning and Statistical techniques through the core courses. Optional courses enable the student to utilise their key theoretical skills on relevant applications.

**Please note that this course is only available on a full time study basis.**

Module options are as follows (Statistics modules are taught by the Department of Statistical Science, and Gatsby modules are taught by the Gatsby Computational Neuroscience Unit .)

Core modules
Students must take the following Computer Science core module:

• COMPGI01 - Supervised Learning [CS]

Plus one of either

• COMPGI08 - Graphical Models [CS]

• COMPGI18 - Probabilistic and Unsupervised Learning [Gatsby]

Students must then take the following Statistics core module:

• STATG001 - Statistical Modelling and Data Analysis [Stats]

Plus one of either

• STATG004 - Applied Bayesian Methods [Stats]
• STATG002 - Statistical Design of Investigations [Stats]
• STATG003 - Statistical Computing [Stats]
• STATG012 - Statistical Inference [Stats]

Optional modules
Students should then choose 4 modules from the following list:

• COMPGI06 - Evolutionary and Natural Computation [CS]
• COMPGI07 - Programming & Mathematical Methods for Machine Learning [CS]
• COMPGI09 - Applied Machine Learning [CS]
• COMPGI10 - Bioinformatics [CS]
• COMPGI13 - Advanced Topics in Machine Learning [CS]
• COMPGI14 - Machine Vision [CS]
• COMPGI15 - Information Retrieval and Data Mining [CS]
• COMPGI16 - Approximate Inference and Learning in Probabilistic Models [Gatsby - only available if you have chosen 'Probabilstic and Unsupervised Learning' as a core module.]
• COMPGV08 - Inverse Problems in Imaging [CS]
• STATG004 - Applied Bayesian Methods [Stats]
• STATG002 - Statistical Design of Investigations [Stats]
• STATG003 - Statistical Computing [Stats]
• STATG012 - Statistical Inference [Stats]
• STATG011 - Forecasting [Stats]
• STATG017 - Stochastic Methods in Finance [Stats]
• STATG019 - Advanced Topics in Statistics [Stats]
• STATG020 - Stochastic Methods in Finance 2 [Stats]

Please note that 'Statistical Computing' and 'Programming & Mathematical Methods for Machine Learning' are very similar. Due to their similarity, students should not pick both of these modules.

Project

The project is a major component of the Masters degree and is a chance to research in some detail an area of interest to the student. The project dissertation counts for 33% of the total marks for the programme. The project is chosen in agreement with the course Director, and would nominally be (co)supervised by the Statistics or Computer Science department. More information on the project is available here.

Themes

Whilst not mandatory, by taking a suitable combination of optional and core courses, natural themes of study arise. Depending on timetable constraints, suggested themes that students may wish to follow are given below:

• Bioinformatics Theme: The optional course pertinent to this theme is currently Bioinformatics. Student projects in this theme are supervised by the Bioinformatics unit.
• Machine Vision Theme: Pertinent courses are Machine Vision and Graphical Models. Student projects are supervised by the Vision and Imaging Science group in the Computer Science department.
• Information Retrieval Theme: Pertinent courses are, Probabilistic and Unsupervised Learning, Approximate Inference and Learning in Probabilistic Models, Information Retrieval and Graphical Models. Projects are supervised by Computer Science.
• Finance Theme: Pertinent courses are Stochastic Methods in Finance, Forecasting. Projects are supervised by Computer Science or Statistics.
• Natural Computation Theme: Pertinent courses are Evolutionary Systems and Advanced Topics in Machine Learning, which covers visual information processing in natural systems. Students may be supervised by the Gatsby Computational Neuroscience Unit.