MSc in Advanced Computing MS51
(Machine Learning, Data Mining and High Performance Computing)

This is part of the MSc in Advanced Computing. The course director is Tim Kovacs and the course representative is Long Pham.

Units on offer in 2011-12

The MSc is built out of 3 themes: High Performance Computing, Intelligent Systems, and Data Mining. Students will take any 2 of the 3 themes.

	First semester	Second semester
HPC	Algorithmic and Economic Aspects of the Internet Advanced Algorithms	Server Software High-Performance Computing
Intelligent Systems	Autonomous Systems Image Processing & Computer Vision Mobile and Ubiquitous Computing	Computational Neuroscience Robotic Systems
Data Mining	Introduction to Machine Learning (Mandatory) Computational Genomics (and see also) Uncertainty Modelling	Pattern Analysis and Statistical Learning Artificial Intelligence with Logic Programming

All units in the table above are worth 10 credits. Each student takes units from 2 of the 3 themes (rows in the table).

1st semester: 60 credits

In the first semester all students take 2 units from each of their two rows above (for a total of 40 credits). In addition:

• F(oundation) stream students take COMSM1211 Programming in C (20 credits). F stream students on the Data Mining theme must take Introduction to Machine Learning (the only mandatory unit).
• E(xperienced) stream students take 2 units (20 credits) not already taken, from any of the three rows above.

2nd semester: 60 credits

In the second semester all students take 2 units from each of their two rows (40 credits) and COMSM2100 Project Specification and Design (20 credits) for a total of 60 credits.

Summer: 60 credits

All students take COMSM3100 MSc Advanced Project (60 credits) over the summer.

Semester 1 Unit Codes and Links

Unit Code	CP	Unit Title	MS51
COMS30121	10	Image Processing and Computer Vision	o
COMS30301	10	Introduction to Machine Learning	o
COMS31900	10	Advanced Algorithms	o
COMSM0106	10	Mobile and Ubiquitous Computing	o
COMSM0305	10	Learning in Autonomous Systems	o
COMSM1211	20	Programming in C	o
COMSM2006	10	Algorithmic and Economic Aspects of the Internet	o
EMATM0004	10	Computational Genomics and Bioinformatics Algorithms	o
EMATM1120	10	Uncertainty Modelling for Intelligent Systems	o

Key: o - Option Choice

Semester 2 Unit Codes and Links

Unit Code	CP	Unit Title	MS51
COMSM0005	10	Artificial Intelligence with Logic Programming	o
COMSM2001	10	Server Software	o
COMSM2127	10	Computational Neuroscience	o
COMSM2202	20	Research Skills	c
COMSM2400	10	High Performance Computing	o
COMSM4111	10	Robotic Systems	o
EMATM1400	10	Pattern Analysis and Statistical Learning	o

Key: o - Option Choice, c - Core Choice

Project

Unit Code	CP	Unit Title	MS51
COMSM3100	60	MSc Advanced Project	c

Key: c - Core Choice

Streams

Students are assigned to either the Experienced or Foundation stream following a programming test during the introductory week.

Optional units

You will need to register yourself on your optional units. You can initially register for and take as many optional units from your themes as you like. This is so you can try them out. You must finalise your semester 1 options by unregistering from the extra units by the end of week 2 (specifically, by 09:00 Monday of week 3). Full time students will have exactly 60 credits in semester 1 after deregistering from the extra units. You can check your registration here by replacing the xxxxxx with your login name and you can change your registrations by clicking on the "unit choice" link. Any changes will be listed as "pending" until the faculty approves them. That is ok. You don't need to do anything else. Please note that the mybristol page may take a few days to be updated with any unit changes, so use the link above instead. If you need any help changing your unit registrations please contact the graduate school office in room 2.19 MVB.

You do not need to register for semester 2 options until about January 15, but please make sure you do so by then. Otherwise you may miss emails from the unit organisers. As with semester 1, you will have 2 weeks to try out any optional units and you must unregister from the extra ones by 09:00 on the Monday of week 15. Full time students will have exactly 60 credits in semester 2 after deregistering from the extra units.

Course Information

• The MS51 timetable.
• Your personal information can be found here (replace xx9999 with your login name in the link).
• You should know the following critical parts of the department handbook. (I'm not giving links as you should browse the handbook and because they change every year and I don't want them to get out of date.)
  • Plagriarism
  • Communication
  • Mitigating circumstances (if e.g. you become ill and it affects your work)
  • Penalties for late submission of coursework
  • The UK marking scale
• See the Advanced Computing page
• All students on this course, with photos
• Introductory talk (Advanced Computing MSc)
• Timetables. Also here.
• Introductory talks (Matlab, Latex, programming tools, etc.)
• Intelligent Systems Research at Bristol
• Intelligent Systems seminars
• Intelligent Systems Group Meetings

Please note that although most units in computer science only have lectures in teaching weeks 1-10, there may be exceptions. In particular, EMATM0004 Computational Genomics and Bioinformatics Algorithms, which is run by the Engineering Mathematics department, has had lectures in weeks 11 and 12 in the past. These weeks are after Christmas. Other first semester units may schedule assignments which are due after Christmas, although this is unusual.

Also note that the official timetable for lectures and lab sessions is not always followed. Lectures are occasionally moved to different times and places if, for example, the lecturer is not available at the scheduled time. Lab sessions may simply be times when you are given priority access to the computer labs. Alternatively, there may be a lab demonstrator who is there to help with the material you are working on. Some units have a demonstrator some weeks but not others.

In short you should check the lecture plan and list of assignments on each unit's home page to make sure you know what is going on. You should also check your email at least once a day as sometimes changes are made at short notice.

Part-time students

We occasionally have part time students and we're fairly flexible about dividing the MSc across years. Intro to ML and AI are done in the first autumn as they're foundational and the project specification and design unit is done the first spring in order to start the project afterwards (which gives you up to 16 months to work on it, assuming you do the degree over two years). The teaching year is basically 20 weeks of teaching plus a couple of weeks of exams. Following the June exams the only requirement is the project which you work on in your own time.

Please note that the MSc is a lot of work and it's not possible to complete it while working 4 days a week on something else.

How you will spend your time

Ten-credit units typically have 2 hours of lectures per week for 10 weeks. With 60 credits per semester that's 12 hours of lectures per week. There are few if any mandatory lab sessions. Most lab sessions are just times when that unit has priority in the lab. Most of your time will be spent reading and coursework.

Jobs and PhD study

Many companies will advertise for employees with a PhD in a relevant area. There are not many specialist MSc programmes like this one and some employers are not aware they exist at all. As a result it may be worth contacting them to ask if a job would be suitable for someone without a PhD.

Some good places to look for jobs and PhD places are:

• The CS department's placements and jobs board in the forum
• jobs.ac.uk
• PhDjobs
• The Connectionists Mailing List
• Machine Learning News
• The Reinforcement Learning Mailing List
• The CS department destinations page.
• You may want to look at the profiles of Ming Li and Anastasia Krithara who both graduated from MS51.

If you are interested in studying for a PhD in the UK you should start making applications in January. (To study in the USA you should apparently start 1 year in advance, i.e. in October.) The two main difficulties are finding funding and finding a suitable supervisor. The MLBC group lecturers can give you advice. You should read the aims, objectives and guidelines for PhD students to get a feel for what a PhD involves. You may also want to read these books:

• The PhD Application Handbook This is specifically for people who are trying to find a PhD.
• How to get a PhD. This is written for students who already have a PhD, but it gives a lot of insight into what a PhD involves and may help you decide whether you want to do one, and may help chose a suitable supervisor.

ECML/PKDD 2012

Bristol will host the European Conference on Machine Learning and Principles and Practices of Knowledge Discovery in Databases (ECML/PKDD) conference at the end of September 2012. This is the major European event in this area and a great opportunity for students to learn about cutting-edge research and to interact with top researchers from around the world.

Providing there is enough space available, MS51 students will be invited to help run the conference (e.g. helping at the registration desk) in exchange for free admission to the technical sessions (where research talks are given).

Preparing for the MSc

Here is some information.