The journal club is designed to provide a forum for discussing relevant and interesting pieces of literature, be it theses, book chapters, journal papers, collections of papers etc.
Friday 30-07-04, 11am-1pm LP290C
Quaid is going to present this one.
• Discovery of Conserved Sequence Patterns Using a a Stochastic Dictionary model M. Gupta and J.S. Liu [pdf].
As yet unscheduled meetings
Papers and schedule from the reading group from last year can be found here.
In reverse chronological order:-Boosted Particle Filters
Friday 16-07-04, LP290C
• A Boosted Particle Filter: Multitarget Detection and Tracking K. Okuma, A. Taleghani, N. deFreitas, J. J. Little and D. G. Lowe [pdf].Stick-Breaking Priors
Friday 18-06-04, 11am-1pm LP290C
David, Yee Whye and Matt.
• Gibbs sampling methods for stick-breaking priors H. Ishwaran and L. F. James [pdf].
• A Constructive Definition of Dirichlet Priors J. Sethuraman [pdf].
• Hierarchical Dirichlet Processes Y. W. Teh, M. I. Jordan, M. J. Beal, D. M. Blei [pdf].
Thu 29-01-04, 10am-12pm LP290C
Jacob is going to present. No reading necessary(!) but do print out this and bring it with you.
To save trees, just print the first 28 pages, as we won't be going any further than that.
Thursday 18-12-03, 11am-1pm PT290C
Rómer is presenting. We will be discussing the following paper:
• Information Geometry of the EM and em Algorithms for Neural Networks Shun-ichi Amari. Neural Networks, Vol.8, No. 9, pp. 1379-1408, 1995. [pdf].Low-Density Parity-Check Codes
Tue 02-12-03, 10am-12pm LP290C
Jacob will present an introduction to LDPC codes. These are state-of-the-art codes, and are used everywhere.
Minimum reading is the very-easy-to-read introductory paper (30 pages):
• An Introduction to Low-Density Parity-Check Codes A. Shokrollahi [pdf].
For keen students, Jacob has recommended a further optional paper for discussion should we have time:
• The Capacity of Low-Density Parity Check Codes under Message-Passing Decoding T. Richardson and R. Urbanke [pdf] [ps.gz] .
Fri 21-11-03, 10am-12pm LP290C
Jacob and Matt will present Tom Minka's EP work from his thesis.
Minimum reading is chapters 1-3 of his thesis (just 25 pages):
• A family of algorithms for approximate Bayesian inference T. P. Minka [pdf] [ps.gz].
We will also try to cover chapter 4 and parts of 5, as these contain exciting applications.
K-SAT and Survey Propagation
Fri 12-09-03, 12-2pm LP290C
NOTE: as this is a continuation of a previous meeting, it is probably suitable only for those already familiar with Survey Propagation. Stark will lead the discussion.
• Survey propagation: an algorithm for satisfiability A. Braunstein, M. Mezard, R. Zecchina.
• Also look at this page for a collection of papers.