M. Kemal Sönmez

I draw tools from statistical signal processing, systems theory, information theory, machine learning, pattern recognition, and neural computation in working on problems in:

• Computational Biology I am interested in discovery and implementation of algorithms that facilitate the understanding of biological processes. Particular applications have included pathway modeling, sequence-to-function analysis of genes and proteins, ontology/schema development for biological databases (http://www.biospice.org), and determination of sleep/arousal states in rodents from electrophysiological recordings.

• Statistical Signal Processing and Automatic Speech/Speaker Recognition My work has concentrated on automatic speech and speaker recognition, especially robust speech recognition, modeling of prosody, and multiresolution speech representations. Earlier work has included information theoretic bases for estimation, and stochastic processes.

• Ph.D. in Electrical Engineering
  UNIVERSITY OF MARYLAND, College Park, Maryland, December 1998
  • Dissertation: Robust Speech Recognition by Topology Preserving Adaptation
  • Advisor: John S. Baras
• M.S. in Electrical and Computer Engineering
  NORTH CAROLINA STATE UNIVERSITY, Raleigh, North Carolina, May 1991
• B.S. in Electrical Engineering
  MIDDLE EAST TECHNICAL UNIVERSITY, Ankara, Turkey, June 1989

• Senior Research Engineer May 2001 - present
  Research Engineer November 1996 - May 2001
  Information and Computing Sciences Division
  SRI INTERNATIONAL (former Stanford Research Institute), Menlo Park, California
• Visiting Research Scientist Summers 1993, 1994, 1995
  Systems and Information Sciences Laboratory
  TEXAS INSTRUMENTS, INC., Dallas, Texas
• Graduate Research Assistant September 1992 to November 1996
  Institute for Systems Research (ISR)
  UNIVERSITY OF MARYLAND, College Park, Maryland
• InstructorSpring 1996
  Department of Computer Science and Electrical Engineering
  UNIVERSITY OF MARYLAND BALTIMORE COUNTY
  ENEE 711: Neural Networks for Signal Processing (graduate course)
• Technical Consultant February 1997 - August 1998
  ENERGY GENERATION TECHNOLOGIES, Annapolis, Maryland
• Technical Consultant October 1993 - September 1995
  S3 TECHNOLOGIES (GSE), Columbia, Maryland
• Research AssistantMay 1990 - July 1992
  Department of Electrical and Computer Engineering
  NORTH CAROLINA STATE UNIVERSITY, Raleigh, North Carolina
• Teaching Assistant August 1989 - May 1990
  Department of Electrical and Computer Engineering
  NORTH CAROLINA STATE UNIVERSITY, Raleigh, North Carolina
• Technical Staff May 1987 - May 1989
  Research and Development Center
  ASELSAN INC., Ankara, Turkey

 RESEARCH GRANTS

• Co-Principal Investigator
  Genomic Discovery of Novel Peptide Hormones
  NIH
  October 2002-October 2005 (pending, with Lawrence Toll, PI)
• Key Technical Member
  Automatic Scoring of Mouse Sleep States
  NIH
  October 2002-October 2006 (pending, with Thomas Kilduff, PI)
• Proposal Manager and Key Technical Member
  BIOSPICE: A Problem-Solving Environment for Biological Network Informatics
  Subcontractors: University of California Berkeley, Lawrence Berkeley National Laboratories, CFD Research Corporation
  DARPA BIOCOMP
  October 2001-October 2006
• Principal Investigator
  Rethinking Acoustic Processing in Speech Recognition
  DARPA EARS-Novel Approaches 
  October 2001-September 2003
• Principal Investigator
  Speaker Recognition and Tracking with Prosodic Features
  Department of Defense
  October 2001-September 2003
• Principal Investigator
  Multiresolution Speech Representation and Modeling for Robust Speech and Speaker Recognition
  DARPA ROAR June 2001-June 2002
• Principal Investigator
  Speaker Style Modeling
  NSF-KDD
  September 2001-August 2003
• Principal Investigator
  Multi-speaker Tracking and Detection
  Department of Defense
  March, 1998 - August 1999
• Principal Investigator
  Nonlinear Discriminant Analysis Feature Design for Speech and Speaker Recognition
  Department of Defense
  March, 1998 - March, 1999
• Co-Principal Investigator
  Bayesian Belief Networks for Effective Troubleshooting Methodology Development
  Maryland Industrial Partnerships and Energy Generation Technologies
  February, 1997 - January, 1999
• Co-Principal Investigator
  Neural Network Simulation of Real-Time Dynamic Systems
  Maryland Industrial Partnerships and S3 Technologies (GSE), Columbia, MD
  Phase I: October, 1993 - September, 1994
  Phase II: October, 1994 - September, 1995

 HONORS AND AWARDS

• Invited panelist at the Fall 2000 Institute for Mathematics and its Applications (IMA) Workshop: Mathematical Foundations of Speech Processing and Recognition, University of Minnesota, September 18-22, 2000.
• Pi Mu Epsilon (Mathematics honorary)
• Third Place, International Mathematica Programming Competition, Rotterdam, Netherlands, September 1992
• Turkish Education Foundation Award, May 1985
• Parlar Foundation Scholarship, Middle East Technical University (awarded to the top three undergraduates in Electrical Engineering), September 1987 - May 1989
• TÜBITAK Scholarship, September 1982 - December 1984
• First Place, TÜBITAK National Mathematics Competition, September 1981

 PROFESSIONAL ACTIVITIES

• Program committee member for 2001, 2002, 2003 IEEE Workshops on Neural Networks for Signal Processing
• Book reviewer for Addison-Wesley (Subject: Neural computation)
• Reviewer for IEEE Transactions on Speech and Audio Processing
• Reviewer for IEEE Signal Processing Letters
• Reviewer for Journal on VLSI Signal Processing Systems
• Reviewer for International Journal for Modeling and Simulation

 COMPUTATIONAL SKILLS

• Experimentation and prototype development in MATLAB, Mathematica, S-Plus, ESPS
• Large scale code development in Unix/C/Perl/Java environment
• RDBMS schema/loader development with Oracle 8.0, Perl DBI

 PATENTS

1. M. K. Sönmez and P. K. Rajasekaran, ``Method and system for compensating speech signals using vector quantization codebook adaptation,'' United States Patent no. 5,745,872.
2. H.L. Bratt, M.K. Sönmez, E.E. Shriberg, ``Method and Apparatus for Performing Prosody Based Endpointing of a Speech Signal'', Patent aplied for April 10, 2001
3. M. K. Sönmez, L.R. Toll, P.D. Lincoln, P. Karp, ``Method of Identifying Processed Peptides,'' provisional application filing date 12/1/2000.
4. T. Adali, A. Mishra, and M. K. Sönmez, ``Adaptive Troubleshooting with Bayesian Belief Networks,'' provisional application filing date 7/12/99, Application no. 60/143,997.

 WORKING PAPERS

1. M. K. Sönmez, L.D. Toll, ``Computational Identification of Processed Peptides''.
2. M. K. Sönmez, ``Information Geometry of Self-organization''.
3. M. K. Sönmez, ``Robust Long-term Statistics of via a Lognormal Tied-Mixture Model''.

 JOURNAL PUBLICATIONS

1. M. Graciarena, H. Franco, M. K. Sönmez, H. Bratt, ``Combining Standard and Throat Microphones for Robust Speech Recognition,'' IEEE Signal Processing Letters, forthcoming.
2. M. K. Sönmez, L. P. Heck, M. Weintraub, ``Multiple Speaker Tracking and Detection: Handset Normalization and Duration Scoring,'', Digital Signal Processing: A Review Journal, vol. 10, no. 1-3, pp. 133-142, 2000.
3. L. P. Heck, Y. Konig, M. K. Sönmez, M. Weintraub, ``Robustness to Telephone Handset Distortion in Speaker Recognition by Discriminative Feature Design,'', Speech Communication, vol. 31 no. 2-3 pp. 181-192, 2000.
4. T. Adali, X. Liu, M. K. Sönmez, ``Channel Equalization by Conditional Distribution Learning: A Partial Likelihood Approach,'' IEEE Transactions on Signal Processing, vol. 45, no. 4, pp. 1051-1064, Apr. 1997.
5. Z. Rivlin, D. Appelt, R. Bolles, A. Cheyer, D. Hakkani-Tür, D. Israel, L. Julia, D. Martin, G. Myers, K. Nitz, B. Sabata, A. Sankar, E. E. Shriberg, M. K. Sönmez, A. Stolcke, G. Tür, ``MAESTRO: Conductor of Multimedia Analysis Technologies'', Communications of the ACM, vol. 43, no. 2, pp. 57-63, 2000.
6. T. Adali, B. Bakal, M. K. Sönmez, and R. Fakory, ``NOx and CO prediction with variable tapped delay line neural networks,'' Journal of Integrated Computer-Aided Engineering, Special Issue on Neural Techniques for Industrial Applications, vol. 6, no. 1, pp. 27-39, 1999.
7. T. Adali, B. Bakal, M. K. Sönmez, R. Fakory, and O. Tsaoi, ``Modeling core neutronics with recurrent neural networks,'' Neurocomputing, Special Issue on Recurrent Networks for Sequence Processing, vol. 15, pp. 363-381, 1997.

REFEREED PUBLICATIONS IN CONFERENCE PROCEEDINGS

1. Steven Eker, Merrill Knapp, Keith Laderoute, Patrick Lincoln, José Meseguer, Kemal Sonmez, ``Pathway Logic: Symbolic Analysis of Biological Signaling,'' Proceedings of the Pacific Symposium on Biocomputing, Hawaii, January 2002, pp. 400-412.
2. Lawrence Toll, Kemal Sonmez, ``Computational Methods for Novel Neuropeptide Identification'', NeuroScience 2002, Orlando.
3. M. K. Sönmez, M. Plauche, Elizabeth Shriberg, Horacio Franco, ``Consonant Discrimination in elicited and Spontaneous Speech: A Case for Signal-adaptive Front Ends in ASR,'' Proc. of ICSLP 2000, Beijing, China, October 2000.
4. M. Plauche, M. K. Sönmez, ``Machine Learning Techniques for Identification of Cues for Stop Place'', Proc. of ICSLP 2000, Beijing, China, october 2000.
5. C. Teixeira, H. Franco, E. Shriberg, K. Precoda, K. Sönmez, "Prosodic Features for Automatic Text-Independent Evaluation of Degree of Nativeness for Language Learners," ICSLP'2000 - 6th International Conference on Spoken Language Processing, Beijing, China, October 2000.
6. M. K. Sönmez, "Information Geometric Topology Preserving Adaptation," invited paper to appear in Proc. IEEE Int. Conf. Acoust., Speech, Signal Processing (ICASSP-2000), Istanbul, Turkey, June 2000.
7. M. K. Sönmez, L. Heck, M. Weintraub, ``Speaker Tracking and Detection with Multiple Speakers,'' Proc. EUROSPEECH 99, Volume 5, Page 2219-2222 Budapest, Hungary, September 1999, Volume 5, pp. 2219-2222.
8. M. K. Sönmez, E. Shriberg, L. Heck, M. Weintraub, ``Modeling Dynamic Prosodic Variation for Speaker Verification,'' Proc. Intl. Conf. on Spoken Language Processing, Sydney, Australia, 1998, Vol. 7, pp. 3189-3192.
9. M. K. Sönmez, L. Heck, M. Weintraub, E. Shriberg, ``A Lognormal Model of Pitch for Prosody-based Speaker Recognition,'' Proc. EUROSPEECH 97, Rhodes, Greece, September 1997, Volume 3, pp. 1391 - 1394.
10. M. K. Sönmez, P. K. Rajasekaran, J. S. Baras, ``Robust Recognition of Cellular Telephone Speech by Adaptive Vector Quantization,'' Proc. IEEE Int. Conf. Acoust., Speech, Signal Processing (ICASSP-96), Atlanta, Georgia, May 1996, Volume 1, pp. 503-506.
11. T. Adali, M. K. Sönmez, and X. Liu, ``Partial Likelihood Estimation for Real-time Signal Processing,'' Proc. IEEE Int. Conf. Acoust., Speech, Signal Processing (ICASSP-96), Atlanta, Georgia, May 1996, Volume 6, pp. 3562-3565.
12. M.K. Sönmez, J.S. Baras, ``The Multiple Lag Process and Its Estimation,'' Proc. Int. Conf. on Acoustics, Speech and Signal Processing (ICASSP-95), June 95, Detroit, Michigan, Volume 3, pp. 1697-1700.
13. T. Adali, M.K. Sönmez, K.Patel, ``On the Dynamics of the LRE Algorithm: A Distribution Learning Approach to Adaptive Equalization,'' Proc. Int. Conf. on Acoustics, Speech and Signal Processing (ICASSP-95), June 95, Detroit, Michigan, Volume 2, pp. 929-932.
14. M.K. Sönmez, J.S. Baras, ``A Statistical Complexity Framework for Topology Preserving Adaptive Vector Quantization,'' invited paper in Proc. of the 30th Annual Conference on Information Sciences and Systems, March 96, Princeton University, New Jersey.
15. M.K. Sönmez, T. Adali, ``Least Relative Entropy Channel Equalization '' invited paper in Proc. of the 27th Annual Conference on Information Sciences and Systems, March 93, Johns Hopkins University, Baltimore, Maryland.
16. M. K. Sönmez and T. Adali, ``Channel Equalization by Distribution Learning: the Least Relative Entropy Algorithm,'' in Applications of Neural Networks to Telecommunications, J. Alspector, R. Goodman, and T. X. Brown editors, 1993, pp. 218-224.
17. T. Adali, X. Liu, M. K. Sönmez, and N. Li, ``A Partial Likelihood Framework for the Analysis of LRE Algorithm: A Distribution Learning Approach for Adaptive Equalization,'' Proc. IEEE Int. Workshop on Neural Networks for Signal Processing, Boston, MA, Sep. 1995.
18. T. Adali, M.K. Sönmez, K. Patel, ``Convergence and Tracking Properties of the Least Relative Entropy Algorithm,'' Proc. Int. Conf. on Acoustics, Speech and Signal Processing (ICASSP-95), June 95, Detroit, Michigan.
19. M.K. Sönmez, ``Multiresolution Finite-State Vector Quantization: Robust Source Coding with Memory '', Proc. of the 28th Annual Conference on Information Sciences and Systems, March 94, Princeton University, Princeton, New Jersey.
20. B. Bakal, T. Adali, R. Fakory, D. Komo, M. K. Sönmez, and O. Tsaoi, ``Modeling Core Neutronics by Recurrent Neural Networks,'' in Proc. World Congress on Neural Networks, (Washington, DC), July 1995, vol. 2, pp. 504-508.
21. B. Bakal, T. Adali, M. K. Sönmez, and R. Fakory, ``Time Delay Neural Networks for NOx and CO Prediction in Fossil Fuel Plants,'' in Proc. of World Congress on Neural Networks, (Washington, DC), July 1995, vol. 3, pp. 111-115.
22. B. Bakal, T. Adali, R. Fakory, D. Komo, M. K. Sönmez and C. O. Tsaoi, ``Simulation of Nuclear Reactor Core Dynamics by Recurrent Neural Networks,'' in Computer-Based Human Support Systems: Technology, Methods and Future, (Philadelphia, PA), June 1995, pp. 475-479.
23. B. Bakal, T. Adali, R. Fakory, D. Komo, M. K. Sönmez, and O. Tsaoi, ``Neural Network Simulation of Real-time Core Neutronic Model,'' in Proc. SCS Simulation Multiconference, (Phoenix, AZ), Apr. 1995, pp. 160-165.
24. T. Adali, M.K. Sönmez, ``Channel Equalization with Information Theoretic Criteria,'' Proc. Int. Conf. on Acoustics, Speech and Signal Processing, (ICASSP-94), May 94, Adelaide, Australia.
25. M.K. Sönmez, J.S. Baras, ``Maximum Partial Likelihood Estimation with Perceptrons,'' Proc. of the 27th Annual Conference on Information Sciences and Systems, March 93, Johns Hopkins University, Baltimore, Maryland.
26. M.K. Sönmez, J.S. Baras, ``Time Series Modeling with Perceptrons: A Likelihood Approach,'' World Congress on Neural Networks, July 93, Portland, Oregon.
27. M.K. Sönmez, M.W. White, R.J. Trew,``An Efficient Neural Network Development Environment via MathLink,'' Mathematica Conference, February 92, Boston.
28. M.K. Sönmez, R.J. Trew,``Front-end Topologies for Phased Array Radiometry,'' Proceedings of the 22nd European Microwave Conference, August 1992, Helsinki, Finland.
29. M.K. Sönmez,``Automated Geometry Theorem Proving,'' Mathematica Conference, February 92, Boston.
30. M.K. Sönmez,``Geometry Theorem Proving with Wu's Method,'' Mathematica Conference, September 92, Rotterdam, Netherlands.

TECHNICAL REPORTS AND ABSTRACTS

1. Consonant Discrimination in Elicited and Spontaneous Speech: A case for signal-adaptive front ends in ASR. Proceedings of NIST Speech Transcription Workshop. College Park, MD. (With M. Plauche, E. Shriberg, and H. Franco.)
2. The SRI March 2000 Hub-5 Conversational Speech Transcription System. Proceedings of NIST Speech Transcription Workshop. College Park, MD. (With A. Stolcke,H. Bratt, J. Butzberger, H. Franco, V.R. Rao Gadde, M. Plauche, C. Richey, E. Shriberg, F. Weng, and J. Zheng.)
3. A. Stolcke, E. Shriberg, D. Hakkani-Tur, G. Tur, Z. Rivlin, M. K. Sönmez, ``Combining Words and Speech Prosody for Automatic Topic Segmentation,'' Proc. DARPA Broadcast News Workshop, Herndon, VA, 1999.
4. M. K. Sönmez, Y. H. Kao, P. K. Rajasekaran, J. S. Baras, ``Comparison Studies of Several Microphone Robustness Techniques,'' Institute for Systems Research Technical Report TR.94-30
5. M. K. Sönmez, P. K. Rajasekaran, ``Studies in Speech Recognition over Telephony,'' International Conference on Acoustics, Norway, July 95.