Dr. Khalid DAOUDI

Senior research scientist (Chargé de Recherches) at INRIA.
Now on leave at CNRS with the SAMOVA team of IRIT.

Address:                              
IRIT-CNRS                                                         
118, route de Narbonne                                     
31062 Toulouse - cedex9                                   
France                                                                
                                                                          
Tel    : (+33) 5 61 55 74 32                     
Fax   : (+33) 5 61 55 62 58
Email: daoudi@irit.fr

 

 

 

   

Brief Biography

I received both the D.E.A and the Ph.D. degrees in applied mathematics from University Paris 9 Dauphine in 1993 and 1996, respectively. My Ph.D. dissertation was prepared at the Fractals Group of INRIA Rocquencourt, France. During 1997, I held a post-doctoral position at the Department of Mathematics, Ecole Polytechnique de Montréal, Canada. From December 1997 to July 1999, I held a post-doctoral position at the Stochastic Systems Group (SSG) of the Laboratory for Information and Decision Systems (LIDS), Massachusetts Institute of Technology (MIT), Cambridge, USA. Since October 1999, I have a permanent position at INRIA Lorraine within the Speech Group. Since October 2003, I am on leave at CNRS with the SAMOVA team of IRIT.
My research interests include Statistical Modeling and Estimation, Machine Learning, Bayesian Networks, Speech and Speaker Recognition (and previously Wavelets, and Multifractal Analysis).
 

 

Ph.D. Students

  

European Projects

 
 

Awards


Publications

 

International Journals

¤ "Unsupervised Stream-Weights Computation in Classification and Segmentation Tasks"
    Eduardo Sanchez-Soto, Alexandros Potamianos and Khalid Daoudi 
    To appear in IEEE Transactions on Audio, Speech and Language Processing.
2008.

¤ "Feature Space Mahalanobis Sequence Kernels: Application to SVM Speaker Verification"
    Jérôme Louradour, Khalid Daoudi and Francis Bach
    IEEE Transactions on Audio, Speech and Language Processing.
Vol. 15, Iss. 8, 2007. pp.2465 - 2475 .

¤ "Dynamic Bayesian Networks for Multi-Band Automatic Speech Recognition"
     Khalid DAOUDI, Dominique FOHR and Christophe ANTOINE
     Computer Speech and Language.
Vol 17, 2003. pp.263-285.

¤ "Signal Representation and Segmentation based on Multifractal Stationarity" (postscript)
     Khalid DAOUDI and Jacques LEVY VEHEL
     Signal Processing. Vol. 82. Num. 12. 2002. pp.2015-2024.

¤ "Multiscale Autoregressive Models and Wavelets" (zipped postscript)
     Khalid DAOUDI, Austin B. FRAKT and Alan S. WILLSKY
     IEEE Transactions on Information Theory.
Vol. 45. Num. 3. 1999. pp.828-845.

¤ "Construction of Continuous Functions with Prescribed Local Regularity" (postscript)
     Khalid DAOUDI, Jacques LEVY VEHEL and Yves MEYER
     Constructive Approximation.
Vol. 14. Num. 3, 1998. pp.349-386

¤ "Fractal Modelling of Speech Signals" (postscript)
     Jacques LEVY VEHEL, Khalid DAOUDI and Evelyne LUTTON
     Fractals.
Vol. 2. Num. 3. 1994. pp.379-382
 

 

Book Chapters

¤ "Continuous Speech Recognition Using Dynamic Bayesian Networks : A Fast Decoding Algorithm"
    
Murat DEVIREN and Khalid DAOUDI
     Chapter in the book "Advances in Bayesian Networks".
Physica Verlag, Springer, 2003.

¤ "Généralisations des systèmes de fonctions itérées: Analyse de la régularité locale et modélisation multifractale des signaux"
     Khalid DAOUDI
     Chapter (in French) in the book "Loi d'échelle, Fractales et Ondelettes".  Traité IC2, Hermes Editions, 2002.

                                                The English version of this book is under press

 
 

Ph.D. Thesis

  "Généralisations des systèmes de fonctions itérées: Applications au traitement du signal"
   University Paris IX Dauphine.  Date of defense: November 29th, 1996.  
                                                                            Committee:  
   President: Prof. Benoît B. MANDELBROT
   Supervisors: Dr. Jacques LEVY VEHEL and Prof. Yves MEYER
   Referees: Prof. Bruno FORTE, Prof. Stéphane JAFFARD, Prof.  Claude TRICOT, Prof. Edward R. VRSCAY
 

 

International Conferences and Workshops

¤ "Unsupervised Stream Weight Estimation Using Anti-Models"
     Eduardo Sanchez-Soto, Alexandros Potamianos and Khalid Daoudi
      IEEE ICASSP'2007.
Apr 15-20, 2007. Hawaii, USA.

¤ "A Novel Strategy for Speaker Verification based on SVM Classification of Pairs of Speech Sequences"
     Khalid Daoudi and Jérôme Louradour
      International Symposium on Signal Processing and its Applications (ISSPA). Feb 12-15, 2007. Sharjah, U.A.E.

¤ "SVM Speaker Verification using an Incomplete Cholesky Decomposition Sequence Kernel"
    Jérôme Louradour, Khalid Daoudi and Francis Bach  
    IEEE Odyssey'2006.
June 28-30, 2006. San Juan, Puerto-Rico.

¤ "Stream Weight Computation for Multi-Stream Classifiers"
   
A. Potamianos, E. Sánchez-Soto and K. Daoudi
    ICASSP'2006.
May 14-19, 2006. Toulouse, France.

¤ "Evaluation of the SPACE denoising Algorithm on Aurora2"
   
Christophe CERISARA and Khalid DAOUDI
    ICASSP'2006.
May 14-19, 2006. Toulouse, France.

¤ "An improved version of the SPACE algorithm for noise robust speech recognition"
    Khalid DAOUDI and Christophe CERISARA
    IEEE-EURASIP ISCCSP'2006.
March 13-15, 2005. Marrakech, Morroco

¤ "The MAP-SPACE denoising algorithm for noise robust speech recognition" (postscript)
   
Khalid DAOUDI and Christophe CERISARA
    IEEE ASRU Workshop.
Nov 27th-Dec 1st, 2005. San Jaun, Puerto Rico.

¤ "Conceiving a new Sequence Kernel and Applying it to SVM Speaker Recognition" (pdf)
   
Jérôme LOURADOUR and Khalid DAOUDI
    Interspeech, Lisboa, Portugal.
Sept 4-8, 2005.

¤ "SVM Speaker Verification using a new Sequence Kernel" (pdf)
   
Jérôme LOURADOUR and Khalid DAOUDI
    European Signal Processing Conference (EUSIPCO), 2005.

¤ "Discriminative Power of Transient Frames in Speaker Recognition"
   
Jérôme LOURADOUR and Khalid DAOUDI
    Int. Conf. on Acoustics, Speech and Signal Processing (ICASSP), 2005.

¤ "Rethinking Language Models Within the Framework of Dynamic Bayesian Networks"
   
Murat DEVIREN, Khalid DAOUDI and Kamel SMAILI
    18th Canadian Conference on Artificial Intelligence (AI'05). Victoria, Canada, May 9-11, 2005.

¤ "Segmentation and Relevance Measure for Speaker Verification"
     Jérôme LOURADOUR, Régine ANDRE-OBRECHT and Khalid DAOUDI
     International Conference on Spoken Language Processing (
ICSLP). Jeju Island, South-Korea, Oct 4-8, 2004.

¤ "Language Modeling using Dynamic Bayesian Networks"
    
Murat DEVIREN, Khalid DAOUDI and Kamel SMAILI
     Int. Conference on Language Resources and Evaluation  (
LREC). Lisbon, Portugal, June 26-28, 2004.

¤ "A New Supervised-Predictive Compensation Scheme for Noisy Speech Recognition"
    
Khalid DAOUDI and Murat DEVIREN
     
EUROSPEECH.
Geneva, Switzerland, Sept 1-4, 2003.

¤ "Frequency Filtering or Wavelet Filtering?"
    
Murat DEVIREN and Khalid DAOUDI
     Joint 13th Int. Conf. on Artificial Neural Networks and 10th Int. Conf. on Neural Information Processing (
ICANN/ICONIP) . Istanbul, Turkey, June 26-29, 2003.

¤ "Continuous Multi-Band Speech Recognition using Bayesian Networks: a Fast Decoding Algorithm"
    
Murat DEVIREN and Khalid DAOUDI
     Probalistic Graphical Models Workshop (
PGM'02). Cuenca, Spain, Nov 6-8, 2002.

¤ "Automatic Speech Recognition: the New Millennium" (postscript)
     Khalid DAOUDI
     International Conference on Industrial and Engineering Application of Artificial Intelligence and Expert Systems (
IEA/AIE). Cairns, Australia, June 17-20th, 2002.
     This paper recieved a "Best Paper Award".

¤ "Dynamic Bayesian Networks for Robust Automatic Speech Recognition"
     Khalid DAOUDI
     (Poster presentation at the) International Meeting on Bayesian Statistics (
Valencia 7).
Tenerife, Spain, June 2 - 6th, 2002.

¤ "Continuous Speech Recognition using Structural Learning of Dynamic Bayesian Networks"(postscript)
     Murat DEVIREN and Khalid DAOUDI
     European Signal Processing Conference (
EUSIPCO).
Toulouse, France, September 3-6th, 2002.

¤ "Continuous Multi-Band Speech Recognition using Bayesian Networks" (postscript)
     Khalid DAOUDI, Dominique FOHR and Christophe ANTOINE
     IEEE Automatic Speech Recognition and Understanding Workshop (ASRU).
Trento, Italy, December 9 -13th, 2001.

¤ "Structural Learning of Dynamic Bayesian Networks in Speech Recognition" (postscript)
     Murat DEVIREN and Khalid DAOUDI
     EUROSPEECH. Alborg, Denemark, September 3-7th, 2001.

¤ "Modeling Dependency between Regression Classes in MLLR using Multiscale Autoregressive Models" (pdf)
     Christophe CERISARA and Khalid DAOUDI
     ISCA Workshop on Adaptation methods for speech recognition.
Sophia-Antipolis, France, August 29 - 30th, 2001.

¤ "A Bayesian Network for Time-Frequency Speech Modeling and Recognition" (postscript)
     Khalid DAOUDI, Dominique FOHR and Christophe ANTOINE
     International Conference on Artificial Intelligence and Soft Computing (ASC).
  Cancun, Mexico, May 21-24th, 2001.

¤ "A New Approach for Multi-Band Speech Recognition based on Probabilistic Graphical Models" (zipped postscript)
     Khalid DAOUDI, Dominique FOHR and Christophe ANTOINE
     International Conference on Spoken Language Processing (ICSLP). Beijing, China, October 16-20th, 2000.

¤ "A New Approach for Multifractal Analysis of Turbulence Signals"
     Khalid DAOUDI
     Fractal'98. Valetta, Malta. October 25-28, 1998.

¤ "Statistical Analysis and Representation of TCP Traffic"
     Khalid DAOUDI and Jacques LEVY VEHEL
     Wavelet Applications in Signal and Image Processing (part of SPIE Annual Meeting). San Diego, USA, July 20-24th, 1998.

¤ "Multifractal Representation of Turbulence Signals : A Wavelet Based Approach"
     Khalid DAOUDI
     International Wavelets Conference. Tangier, Morocco, April 13-17th, 1998.

¤ "Weak Self-Similarity in Speech Signals"
     Khalid DAOUDI
     International Conference on Signal and Image Processing (SIP). New Orleans, USA, December 4-7th, 1997.

¤ "Generalized IFS for Signal Processing"
     Khalid DAOUDI and Jacques LEVY VEHEL
     IEEE Digital Signal Processing Workshop (IEEE'DSP). Loen, Norway, September 1-4th, 1996.

¤ "Speech Modeling Based on Local Regularity Analysis"
     Khalid DAOUDI and Jacques LEVY VEHEL
     International Conference on Signal and Image Processing (SIP).
Las Vegas, USA, November 20-23th, 1995.