Abstract

The Machine Learning book reading club is a new addition to the ML journal club series. It is aimed primarily at PhD students with an interest in the related fields of Machine Learning, Probability Theory, Neuroscience and Information Theory.

Gaining a broad overview over one’s own and neighbouring fields is part of a PhD. Much of such knowledge can be gained from graduate text books, but it is often hard to sustain the motivation necessary to read through the whole of these often massive tomes. The ML book reading club can help to stay on track. Its goal is to read, at a comfortable, sustained pace, through a set of the standard text books. We meet once a week to discuss the contents of last week’s chapter(s).

Currently, the long-term plan is to read through the following books

1. Christopher M. Bishop: Pattern Recognition & Machine Learning
2. E.T. Jaynes: Probability Theory – The Logic of Science
3. Peter Dayan & L.F. Abbot: Theoretical Neuroscience
4. David J.C. MacKay: Information Theory, Inference and Learning Algorithms

• This is a long term project. We aim to have finished Chris Bishop’s book by the start of Michaelmas term 2008/09, and it will take about a year to read through the first three books on the list. Some people may want to join later, when we start on a book that’s more interesting to them. We will try to keep starting dates for new books to convenient points in time (start of term, etc.).

• The point is not to perfectly understand each and every sentence in a book, but also not to skip through half of it. A reasonable level to aim for could be being able to solve the intermediate example problems given.

We will start off with the discussion of the “Introduction” chapter in Chris Bishop’s book (pp. 1 – 67). We will also use the first meeting to discuss good times for consecutive meetings and the general modus operandi. Over the summer, meetings will have to be in the early evening to accommodate those working outside of the University.