Project Manager / Business Analyst

Jørgen Harmse, Ph.D.

****B Clay Avenue Austin, Texas 78756

512-***-**** ********@*****.***

PROFILE

Accomplished Research & applied scientist with experience in the application of mathematics to complex real-world problems. Expert in tracking, signal processing, machine learning, reasoning under uncertainty, clustering, data fusion, decision support, speaker recognition, mathematical modeling, and mathematical analysis. Skilled in scientific programming, including object-oriented programming, and in oral & written presentation of ideas & results.

PROFESSIONAL EXPERIENCE

1993-2010: Scientist, BAE Systems (formerly Tracor and Marconi), Austin, TX

Key contributions:

Forensic Automated Speaker Recognition, 2009-2010: Enhanced speech scientists’ ability to investigate terrorists and other criminals by developing unified clustering and score-visualization tool for grouping speech fragments by speaker. Enhanced score normalization for effects of duration and SNR. Investigated minimum-loss channel normalization for speaker recognition.

Acoustic Locating Array System, 2008: Developed preliminary algorithms for acoustic footstep detection.

Multi-Nodal Unattended Ground Sensor, 2008: Developed tracking algorithm and refined Toxic Industrial Chemical detection. Helped secure program continuation by demonstrating successful tracking and ammonia detection.

Laser Warning IR&D, 2007: Developed or refined algorithms to detect threat lasers and achieved desired performance on simulated & laboratory data. Algorithms and results provided starting point for an externally funded program.

Bionic Ear, 2007: Adapted speaker & language recognition to low-power sensor node. Congress allocated funds for program continuation.

Forensic Automated Speaker Recognition, 2002-2006: Established relationships between speaker ID results and SNR & duration, published paper and developed tool for speech analyst; developed pitch estimation algorithms; studied effect of channel & other factors on speaker recognition; studied application of Empirical Mode Decomposition to speech analysis. The SNR & duration results provided the technical motivation for the Bionic Ear project.

Unattended Ground Sensor IR&D, 2005: Developed, implemented and tested tracking algorithm for small low-power nodes that led to externally funded project.

Defense Logistics Agency, 2000-2001: Developed, implemented, tested, and delivered innovative part-clustering algorithm to diversify clusters on some attributes and to use multiple-part comparisons. Tools are used by logistics experts responsible for large numbers of parts, including mission-critical parts required on short notice.

SensIT, 2001: Helped to design general tracking procedure. Implemented Kalman Filter tracker for single-target case that led to further funded unattended ground sensor work.

Intelligent Drilling Monitor (oil & gas), 1998-2000: Developed heave compensation procedure. Investigated qualitative & quantitative differences (from geothermal case) in diagnosis of drilling problems. Implemented & tested modified algorithm. System has potential to provide timely alerts and significantly reduce cost of drilling problems.

Circulation Monitoring System, 1996-2000: Investigated causal connections between problems in geothermal drilling and observable evidence, including allowance for special operations which would cause false alarms. Specified interface between processing engine & components. Implemented, tested & refined processing engine. Effort led to funded work on similar system for oil & gas industry.

Heart Pump Analysis, 1999: Investigated connection between vibro-acoustic features and pump condition. Demonstrated durability of pumps sufficiently for MicroMed to begin implantations.

Agile Manufacturing, 1996: Assessed utility of Activity Based Costing. Analyzed risk assessment procedures. Designed object-oriented framework for simulation of economic ‘battle space’, and coded processing engine. Wrote Backus Naur form description of the processing.

AN/BSY-1 Algorithm Development, 1993–1999: Developed evidence extraction procedures for detection of common frequency dynamics and common SNR dynamics. Achieved order-of-magnitude increase in speed by modifying implementation of Tracor’s greedy clustering algorithm. Suggested changes to prevent cycling. Investigated problems posed by data correlation, showing it could be taken into account without complicating network. Algorithms were deployed in New Attack Submarine.

Data Fusion IR&D, 1993–1996: Wrote paper on conjunction and disjunction in fuzzy logic. Investigated analytic hierarchy process, neural networks, fuzzy logic and genetic algorithms. Assessed several neural network packages. Wrote prototype logistic regression decision aid. Inves¬tigated effect of blockage on traffic flows. Wrote white paper on appli¬cations of neural networks and Bayesian networks to maintenance. Results were used in funded work on acoustic data fusion.

Predictive Maintenance, 1993: Wrote report summarizing all the main automated reasoning tech-nologies and several decision-support technologies, and discussing the application, advantages and disadvantages of each technology in auto¬mated machinery monitoring. Wrote another report suggesting a possible hierarchical design of an automatic machinery monitoring and control system. Reports contributed to proposal win for submarine automation.

Heart Valve Analysis, 1993: Investigated the connection between acoustic features and heart valve condition. Performed analysis to demonstrate life-saving potential of condition-based reoperation.

1992-1993: Consultant, Tracor Applied Sciences

Investigated and prepared reports on Bayesian networks and fuzzy logic. Supported concept development for the Intelligent Maintenance Action Planner.

1989-1992: Instructor, University of Texas at Austin

1984-1989: Teaching and Research Assistant, Massachusetts Institute of Technology

EDUCATION

Ph.D. in Mathematics, Massachusetts Institute of Technology

M.Sc. (Distinction) in Mathematics, Auckland University

B.Sc. (Hons.) (First Class) in Mathematics, Auckland University

PUBLICATIONS

Wrote peer-reviewed papers, conference papers, and internal & external technical reports. Topics include pure mathematics, fuzzy logic, Bayesian networks, and data fusion. Published papers and reports with Tracor document numbers are listed in publication addendum.

PUBLICATION ADDENDUM

Machine Learning and Reasoning under Uncertainty

• Harmse, “Reduction of Gaussian mixture models by maximum similarity”, Journal of Nonparametric Statistics, First published on: 18 December 2009 (iFirst) http://dx.doi.org/10.1080/10485250903377293

• T. Leih, E. Giannopoulos & Harmse, “Multiple Source Clustering: A Probabilistic Reasoning Approach”, Proceedings of the First Australian Data Fusion Symposium (1996), 141-146.

• Harmse, “Continuous Fuzzy Conjunctions and Disjunctions”, IEEE Transactions on Fuzzy Systems, Volume 4, No. 3 (August 1996), 295–314.

• Harmse, “Correlated Evidence in Bayesian Networks”, Tracor Applied Sciences, Analysis and Applied Research Division, Document No. T95-01-9522-U, March 1995.

• G. Corser & Harmse, “Techniques in Reasoning Under Uncertainty”, Tracor Report No. T93-01-9543-U, Submitted to ARPA/UWO, Attn. Dr. E. Carrapezza, August, 1993.

• Harmse, “Notes on Fuzzy Theory”, Tracor Applied Sciences, Analysis and Applied Research Division, Document No. T92-01-9554-U, December 1992.

• Harmse, “Notes on Bayesian Networks”, Tracor Applied Sciences, Analysis and Applied Research Division, Document No. T92-01-9553-U, December 1992.

Mathematics

• Harmse, “Extensions of the Cauchy-Goursat Integral Theorem”, Journal of Mathematical Analysis and Applications 339 (2008), 429-437, http://dx.doi.org/10.1016/j.jmaa.2007.06.063.

• D. Berend & Harmse, “On Polynomial-Factorial Diophantine Equations”, Transactions of the American Mathematical Society 358, No. 4 (2005), 1741-1779.

• D. Berend & Harmse, “On some arithmetical properties of middle binomial coefficients”, Acta Mathematica LXXXIV.1, 1998, 31-41.

• D. Berend & Harmse, “Gaps between Consecutive Divisors of Factorials”, Annales Institute Fourier, Tome 43 (1993) Fascicule 3, 569–583.

• D. Berend & Harmse, “Expert Rule versus Majority Rule under Partial Information”, Theory and Decision 35 (1993), 179–197.

• Harmse, “On Lebesgue Space Estimates for the Wave Equation”, Indiana University Mathematics Journal 39, No 1 (1990), 229–247.

• Harmse, “Antipodal Coincidence Sets and Stronger Forms of Connectedness”, Bull. Aust. Math. Soc. 31, No 2 (1985), 271–284.

• M. K. Vamanamurthy & Harmse, “A Note on the Implicit Function Theorem in Euclidean Spaces”, Math Chronicle 12 (1983), 139–149.

Speaker Recognition

• Harmse, S. Beck & H. Nakasone, “Speaker Recognition Score-Normalization to Compensate for SNR and Duration”, Proceedings of 2006 IEEE Odyssey – The Speaker and Language Recognition Workshop.

Drilling Problem Diagnosis

• M. W. Weiss, J. E. Copeland, Harmse, B. A. Corser, R. D. Wallace & G. L. Whitlow, “Intelligent Drilling Monitor for Circulation Problems”, IADC Well Control Conference of the Americas, 1999.

• Harmse, R. D. Wallace, D. A. Glowka & A. J. Mansure, “Automatic Detection and Diagnosis of Problems in Drilling Geothermal Wells”, Geothermal Resources Council Transactions 21 (1997), 107 111.

Contact this candidate