Hello, and welcome to Paper of the Day (Po'D): Fast Bayesian Pursuit Algorithm for Sparse Linear Models Edition. Today's paper has been sitting on my stack for a while: P. Schniter, L. C. Potter, and J. Ziniel, ``Fast Bayesian Matching Pursuit,'' Proc. Workshop on Information Theory and Applications (ITA), (La Jolla, CA), Jan. 2008. (An expanded version is here: P. Schniter and L. C. Potter and J. Ziniel, "Fast Bayesian matching pursuit: model uncertainty and parameter estimation for sparse linear models.") We reviewed another Bayesian approach here and here.
July 2010 Archives
Hello, and welcome to part 2 of the Po'D from yesterday. We left off yesterday with the solution element activity decision rule, which brings us to state the pursuit in full.
Hello, and welcome to Paper of the Day (Po'D): Bayesian Pursuit Algorithm for Sparse Representation Edition. Today's paper is: H. Zayyani, M. Babaie-Zadeh, and C. Jutten, "Bayesian pursuit algorithm for sparse representation," in Proc. IEEE Int. Conf. Acoust., Speech, Signal Process., (Taipei, Taiwan), pp. 1549-1552, Apr. 2009. I spoke about this paper on briefly in Paper of the Day (Po'D): The Other Probabilistic Matching Pursuits Edition and Paper of the Day (Po'D): The Other Other Probabilistic Matching Pursuits Edition. (To prepare myself more thoroughly for this task, I have been reading A First Course in Bayesian Statistical Methods.)
This just in from Prof. Laurent Daudet:
A 12-month postdoc position will be available at the Langevin Institute, Paris (France), on the topic of Model-based phase reconstruction of audio signals from magnitude spectrograms.
This position is open in the framework of the ANR DREAM collaborative project, that develops new technologies for active music listening, funded by the french Agence Nationale de la Recherche. The goal of this post-doc is to revisit the classic problem of phase reconstruction from magnitude spectrograms, that has attracted a lot of effort since the pioneering work of Griffin and Lim, in view of recent advances in sound modeling : sparse representations, high-resolution sinusoidal models, etc. After a thorough review of the state-of-the-art in phase reconstruction techniques, the post-doctoral student will investigate how some a priori knowledge about sound sources can be used into reconstruction algorithms, in order to improve and accelerate their convergence. Possible applications of this work include audio spectral enhancement, blind source separation, and a link with theories of compressive sensing and Non-negative Matrix Factorization.
Candidates should hold a Ph.D in Signal Processing, or related domain, with some mandatory experience in audio (speech / music) processing. Good programming skills (Matlab) and ability to work independently are required. A prior research experience in sparse representations and/or time-frequency transforms would be desirable.
Applications (detailed CV, list of publications, name and email of 2 references) or informal enquiries should be made before August 31st, 2010, to
Professor Laurent DAUDET
Tel : +33 1 40 79 52 16
Salary : net amount about 1950 Euro / month
Start date : preferably January 1st, 2011 (negotiable).
The Langevin Institute is a physics laboratory that investigates all aspects of waves (acoustics / optics), attached to the prestigious ESPCI engineering school. It is centrally located in the vibrant 5th "arrondissement" of downtown Paris.
Summer has been in full effect, which explains the recent sad lack of posts here. But now that I am back in the office preparing for a new semester, polishing two grant applications, and enjoying the Wittenborg 7100, I look forward to reviewing several of the papers in my growing pile. On top of this, I am looking forward to several papers at EUSIPCO 2010 (the following list is certainly not comprehensive, but these titles caught my eye with respect to my own ongoing pursuits in research):
Bob L. Sturm, Associate Professor
Audio Analysis Lab
Aalborg University Copenhagen
A.C. Meyers Vænge 15
DK-2450 Copenahgen SV, Denmark