Expectation Propagation Decoding

Contact: Masoud Alipour
Room: BC 150
Tel: 021 6937529
Email: masoud [dot] alipour [at] epfl [dot] ch

Expectation Propagation is a method for approximate inference in Graphical Models. The aim of this project is to treat the tanner graph of LDPC code as a graphical model and to use EP technique for decoding. The proportion of theoretical and empirical (programming, simulation) work in this project is 40-60 percent.

Requirement:

• Good familiarity with basic probability theory
• Good programming skills in Matlab and/or C/C++

What will you learn

• LDPC codes
• Learning and Inference in Graphical Models
• Expectation Propagation
• Fast implementation of approximate inference in graphical models

Suggested Readings and References

• Thomas Minka's EP page
• A Completed Information Projection Interpretation of Expectation Propagation