CURRICULUM VITAE
DENVER H. DASH
PERSONAL DATA
|
Work Address Intel Research Voice: (412) 297-4027 E-mail: denver.h.dash@intel.com |
|
RESEARCH interests
· Machine learning and security.
· Large-scale bio and cyber-surveillance.
· Scaling up learning and inference in Bayesian networks.
· Causal discovery and reasoning in dynamic systems.
· Bayesian and constraint-based learning of probabilistic graphical models.
Academic background
Doctor of Philosophy April 2003
Intelligent Systems Program, University of Pittsburgh, Pittsburgh, Pennsylvania
Dissertation: Caveats for Causal Reasoning with Equilibrium Models
Master of Science May 1997
Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, Pennsylvania
Completed all PhD qualifying exams.
Bachelor of Science May 1994
Department of Physics, Case Western Reserve University, Cleveland, Ohio
RESEARCH EXPERIENCE
Research Scientist 2003 – Present
Intel Research, Santa Clara, California / Pittsburgh, Pennsylvania
n Conducting basic and applied research on machine learning with probabilistic graphical models. Specific areas include large-scale distributed intrusion detection in networked devices and applying machine learning techniques to improve the manufacturing process. Received an Intel Corporate Technology Group Division award for reducing test costs in silicon manufacturing, resulting in estimated savings of millions of dollars per year.
Postdoctoral Research Fellow and Consultant 2003 – 2004
Center for Biomedical Informatics, RODS Laboratory, University of Pittsburgh, Pittsburgh, Pennsylvania
n Conducted research on syndromic biosurveillance using Bayesian network models of an entire population of 1.4 million people in Allegheny County, Pennsylvania to predict when a disease outbreak is occurring in the population.
NASA Graduate Student Research Fellow 1999 – 2003
Intelligent Systems Program, University of Pittsburgh, Pittsburgh, Pennsylvania
n Conducted research on learning and reasoning with causal models. Dissertation addresses how causal relations change over the observation time-scale of systems, examining the relationship between causality in equilibrium models and that in dynamic models. Learning techniques have focused on developing robust and efficient algorithms for supervised and unsupervised classification using Bayesian networks and for learning causal models using a mixture of constraint-based and Bayesian learning approaches.
Summer Intern Summer 2002
HRL Laboratories, Malibu, California
n Simulating the troubleshooting cycle using Monte-Carlo sampling with Bayesian networks to evaluate and refine sequential diagnosis, including analyzing model performance, developing a tractable and accurate utility model, and quantifying domain separability. Work performed in collaboration with Dr. Wojtek Przytula.
Summer Intern Summer 2001
Microsoft Research, Cambridge, United Kingdom
n Worked in the Machine Learning and Perception group supervised by Dr. Chris Bishop. Research focused on developing a class of Bayesian networks that allow efficient exact and approximate model averaging.
Summer Intern Summer 2000
HRL Laboratories, Malibu, California
n Worked with Dr. Wojtek Przytula developing a scheme for updating the parameters of time-varying Bayesian network models with (incomplete) data; specifically, examining the tradeoffs between maintaining confidence in a model based on data versus degrading confidence due to time-variation in the real world.
Graduate Student Researcher 1997 – 1999
Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, Pennsylvania
n Performed theoretical research on manipulation and reversibility in equilibrium and dynamic causal models with Professor Marek Druzdzel at the Department of Information Sciences and Telecommunications. Studied causality in structural equations and Bayesian belief network representations. Work involved modifying these representations to allow the prediction of direct manipulations on endogenous variables in the systems.
Graduate Student Researcher 1995 – 1996
Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, Pennsylvania
n Performed experimental condensed matter research with Professor Xiao-Lun Wu. Implemented computer-control of experiments, video imaging, and particle tracking techniques.
Undergraduate Intern Summer 1993
Loral Defense Systems. Akron, Ohio
n Co-designed and implemented object-oriented software in C++ to simulate modular, arbitrary undersea vehicles in six degrees of freedom.
TEACHING experience
Visiting Lecturer Spring 1998
Computer Science Department, Technical University of Bialystok, Bialystok, Poland
n Taught the fourth-year computer science course: “Learning Bayesian Networks from Data and Other Machine-Learning Techniques”. I was fully responsible for all aspects of the course. 8-hour per week teaching load.
Graduate Teaching Assistant 1994 – 1995
Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, Pennsylvania
n Taught recitations for introductory undergraduate physics courses. Responsibilities included: conducting weekly review sessions; creating, administering and grading quizzes, assigning and grading homework; grading exams.
Undergraduate Teaching Assistant Spring 1994
Department of Physics, Case Western Reserve University, Cleveland, Ohio
n Assisted a laboratory course for sophomore-level physics majors. Course dealt primarily with advanced instrumentation and computer control of experiments. Responsibilities included setting up and assisting students with equipment, and grading quizzes and exams.
Personal Tutor 1991 – 1997
Department of Physics, Case Western Reserve University, Cleveland, Ohio
Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, Pennsylvania
n Tutored students in undergraduate mathematics and physics courses, including Classical Mechanics, Electricity and Magnetism, Calculus and Differential Equations.
publications
Peer-reviewed Conferences & Journals
Denver Dash, Branislav Kveton, John Mark Agosta, Eve Schooler, Jaideep Chandrashekar, Abraham Bachrach, and Alex Newman, When gossip is good: distributed probabilistic inference for detection of slow network intrusions. In Proceedings of the Twenty-First National Conference on Artificial Intelligence (AAAI), AAAI Press, Menlo Park, California, 2006.
Branislav Kveton and Denver Dash, Automatic excursion detection in manufacturing: Preliminary results. Proceedings of the Eighteenth International Florida Artificial Intelligence Research Society Conference (FLAIRS), 2005.
Weng-Keen Wong, G. F. Cooper, D. H. Dash, J. D. Levander, J. Dowling, W. R. Hogan, M. M. Wagner, Bayesian biosurveillance using multiple data streams, in Morbidity and Mortality Weekly Report: Supplement, 2005.
Denver Dash, Restructuring dynamic causal systems in equilibrium. In Robert Cowell and Zoubin Ghahramani, editors, Proceedings of the Tenth International Workshop on Artificial Intelligence and Statistics (AIStats), 2005.
Greg Cooper, Weng-Keen Wong, Denver Dash, John Levander, Bill Hogan, and Michael Wagner. Bayesian biosurveillance using multiple data streams. In Proceedings of the Third Annual Syndromic Surveillance Conference, 2004.
Denver Dash and Gregory F. Cooper. Model averaging for prediction with discrete Bayesian networks. Journal of Machine Learning Research, 5:1177 – 1203, September 2004.
John Mark Agosta, Denver Dash, Christian Shelton and Krishna Arvind. Exploiting parametric test dependencies for selective avoidance of sort tests. Intel Design and Test Technology Conference (DTTC), 2004.
Gregory F. Cooper, Denver H. Dash, John D. Levander, Weng-Keen Wong, William R. Hogan, and Michael M. Wagner. Bayesian biosurveillance of disease outbreaks. In Proceedings of the 20th annual conference on Uncertainty in artificial intelligence (UAI), pages 94 – 103. AUAI Press, 2004.
Denver Dash and Gregory Cooper, Model averaging with discrete Bayesian network classifiers. In Proceedings of the Ninth International Workshop on Artificial Intelligence and Statistics (AIStats), 2003. Available online: http:/denverdash.com/docs/AIS_03.pdf.
Denver Dash and Marek Druzdzel. A robust independence test for constraint-based learning of causal structure. In Proceedings of the 19th Annual Conference on Uncertainty in Artificial Intelligence (UAI), pages 167 – 174. Morgan Kaufmann, 2003.
K. Wojtek Przytula, Denver Dash, Don Thompson. Evaluation of Bayesian networks used for diagnostics. In Proceedings of the IEEE Aerospace Conference, pages 3177 – 3187. 2003.
Denver Dash and Gregory Cooper, Exact model averaging with naïve Bayesian classifiers. In Proceedings of the Nineteenth International Conference on Machine Learning (ICML), 91-98, 2002. Available online: http://denverdash.com/docs/icml_02.pdf.
Haiqin Wang, Denver Dash and Marek Druzdzel, A method for evaluating elicitation schemes for probabilistic models. IEEE Transactions on Systems, Man and Cybernetics—Part B: Cybernetics, 32:1, 38-43, 2002.
Denver Dash and Marek Druzdzel, Caveats for causal reasoning with recursive equilibrium models. In Proceedings of the Sixth European Conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty (ECSQARU), 2001. Abstract available online: http://www.irit.fr/ECSQARU-2001/Ecsqaru-2001.html.
Haiqin Wang, Denver Dash and Marek Druzdzel, A method for evaluating elicitation schemes for probabilities. In Proceedings of the Fourteenth International Florida Artificial Intelligence Research Symposium Conference (FLAIRS), 2001. Available online: http://denverdash.com/docs/wang_dash_druzdzel_flairs2001.pdf.
Denver Dash and Marek Druzdzel, A hybrid algorithm for the construction of causal models from sparse data. In Proceedings of the Fifteenth Annual Conference on Uncertainty in Artificial Intelligence (UAI), 1999. Available online: http://denverdash.com/docs/uai99.ps.zip.
Denver Dash and Xaio-Lun Wu, A sheer-induced instability in freely suspended Smectic-A liquid crystal films. Physical Review Letters, 28:1483-1486, 1999.
Book Chapters and Workshops
Denver Dash, John Mark Agosta, Jaideep Chandrashekar, Eve Schooler, A Distributed Host-based Worm Detection System. Proceedings of the ACM SIGCOMM Workshop on Large Scale Attack Defense (LSAD06), September, 2006.
Denver Dash, John Mark Agosta, Jaideep Chandrashekar, Eve Schooler, Detecting weak network anomalies with Bayesian models. Workshop on Machine Learning Algorithms for Surveillance and Event Detection, In conjunction with the Twenty-Third International Conference on Machine Learning (ICML), 2006.
Gregory F. Cooper, Denver H. Dash, John D. Levander, Weng-Keen Wong, William R. Hogan, and Michael M. Wagner. Bayesian Biosurveillance. In Handbook of Biosurveillance. eds. M. Wagner, A. Moore, R. Aryel, San Diego:Elsevier, 2006.
Denver Dash, John Mark Agosta, Abraham Bachrach, Branislav Kveton, Alex Newman, Eve Schooler, Learning robust generative models for distributed anomaly detection. Intelligence Beyond the Desktop, In conjunction with the Nineteenth annual conference on Neural Information Processing Systems (NIPS), 2005.
Denver Dash, John Mark Agosta, Eve Schooler, Branislav Kveton, Population-based modeling for distributed detection of network anomalies. Intel Research Conference (IRCON), 2005.
John Mark Agosta, Abraham Bachrach, Denver Dash, Branislav Kveton, Alex Newman, Eve Schooler, Distributed inference to detect a network attack. Adaptive and Resilient Computing Security Workshop (ARCS), 2005.
Weng-Keen Wong, Greg Cooper, Denver Dash, John Levander, William Hogan, Mike Wagner, Population-wide anomaly detection, The Workshop on Data Mining Methods for Anomaly Detection, In conjunction with the Eleventh ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, (2005), Chicago, Illinois.
Denver Dash. Empirical investigation of Manipulation-Equilibration commutability in causal models. The Workshop on Causality and Causal Discovery, In conjunction with the Seventeenth Canadian Conference on Artificial Intelligence, (AI) London, Ontario, Canada, 2004.
Denver Dash and Marek J. Druzdzel. Learning the causal structure of dynamic systems over varying time-scales. The Workshop on Graphical Models, Longitudinal Data, Latent Variables and Causal Interpretations, Wiesbaden, Germany, 2002.
Denver Dash and Marek J. Druzdzel. An inconsistency between causal discovery and causal reasoning. The Workshop on Conditional Independence Structures and Graphical Models and the Workshop on Causal Interpretation of Graphical Models, The Fields Institute for Research in Mathematical Sciences, Toronto, Canada, 1999.
Denver H. Dash and Marek J. Druzdzel. Problems related to causal reasoning in equilibrium models. In Proceedings of the Conference on Theoretical Informatics: Methods of Analysis of Incomplete and Distributed Information, pages 24-26, Bialystok, Poland, 1999.
Work In Progress
Branislav Kveton and Denver Dash, Dynamic Bayesian networks for robust detection of epidemic outbreaks.
Denver Dash, Branislav Kveton, John Mark Agosta, and Eve Schooler, Importance sampling for optimal message-passing in distributed rare-event detection.
Denver Dash and Marek J. Druzdzel. Equilibrium causal systems: Theory and practice.
Denver Dash and Marek J. Druzdzel. A note on the correctness of the Causal Ordering Algorithm, submitted to Artificial Intelligence Journal, 2005.
Invited talks
Exact Bayesian Model Averaging for Prediction. (2003) Intel Research, Santa Clara, California.
Caveats for Causal Reasoning with Equilibrium Models. (2003) The Gatsby Unit, London, United Kingdom.
Learning Causal Graphs: Manipulation and Dynamics. The Machine Learning Spring Lecture series at the Technical University of Bialystok (1998), Computer Science Department.
Patents FILED
K. Wojtek Przytula and Denver Dash. Evaluation of Bayesian Network Models for Diagnosis, Filed 8/2003.
Simon Crosby, John Mark Agosta and Denver Dash. A Method for Collaborative Attack Detection in Networked Computer Systems. Filed 11/2004.
Simon Crosby, John Mark Agosta and Denver Dash, A Method For Distributed Sequential Hypothesis Testing In Autonomic Computing Systems. Filed 12/2004.
Programming experience
§ Over 12 years experience with object-oriented C++ and Visual Studio IDE, proficient with C++ STL, Matlab, C, python, experience with Java, Perl, Pascal (Delphi), Basic, LISP, and Prolog.
§ Technical director of Intel’s open-source Probabilistic Network Library, an open-source C++ library, based on BNT, for building and learning graphical models, including: Bayesian networks, Dynamic Bayesian networks, Markov random fields, and factor graphs, including both discrete and continuous variables. Oversaw the construction of high-level interface, R interface, file i/o, and GUI.
§ Systems Implemented (year, language, approximate fraction implemented):
o Distributed detection and inference simulator (2005, C++, 1). A general simulator to model worm outbreaks in a system of networked devices and to test efficacy of distributed detection.
o Wafer testing application core (2004, C++, 1). A custom simulator to calculate wafer failure histograms and optimize test reduction based on statistical models. Front and back ends implemented by Bob Davies.
o Panda inference core (2003, Delphi, 1). The core inference module for the PANDA software, using the Hugin library. Bayesian network inference was re-implemented in log-space to manage calculation underflow due to large evidence sets. Other components of PANDA (front and back ends, among others) were implemented by John Levander.
o Learning component of SMILE (1998-2003, C++, .9), an object-oriented library for building Bayesian networks and influence diagrams. The learning module includes constraint-based and Bayesian methods for learning structure, supports discrete variables with limited support for continuous variables and can handle missing data using the EM algorithm. Also implements efficient Bayesian model averaging, classification tools, cross-validation, many Bayesian metrics, and online parameter updating.
o Diagnostic troubleshooter evaluation tool (2002, C++, 1). A program to help evaluate and visualize the efficacy of discrete BN diagnostic models. Used forward sampling to build a map of test coverage within the model. Code based on the SMILE library.
o Learning component for locomotive diagnosis (2000, C++, 1). A module to refine discrete BN diagnostic model parameters from historical data. Required customization to allow a tradeoff between historical data and possible changes in locomotive components. Code based on the SMILE library.
o Undersea vehicle simulator (C++, 1993, 0.5). A modular program to simulate an undersea vehicle in six degrees-of-freedom, designed to allow swapping in and out of various control surfaces and propulsion mechanisms.
OTHER proffessional EXPERIENCE
|
Workshops |
|
|
Machine Learning Algorithms for Surveillance and Event Detection, At the Twenty-Third International Conference on Machine Learning, Pittsburgh, Pennsylvania. Co-organized with Dragos Margineantu, Terran Lane and Weng-Keen Wong. |
2006 |
|
Program Committees |
|
|
International Conference on Machine Learning (ICML) |
2006 |
|
National Conference on Artificial Intelligence (AAAI) |
2005, 2007 |
|
Workshop on Artificial Intelligence and Statistics (AIStats) |
2005 |
|
Conference on Uncertainty in Artificial Intelligence (UAI) |
2004 – 2006 |
|
Referee |
|
|
Neural Information Processing Systems Conference (NIPS) |
2005 – 2006 |
|
International Joint Conference on Artificial Intelligence (IJCAI) |
2005 |
|
Machine Learning Journal |
2005 – 2006 |
|
Journal of Machine Learning Research |
2004 – 2005 |
|
Journal of Artificial Intelligence Research |
2004 |
|
Journal of Approximate Reasoning |
2004 |
|
Conference on Uncertainty in Artificial Intelligence (UAI) |
1999 – 2003 |
|
Conference on Artificial Intelligence (AAAI) |
2002 |
|
Online Proceedings Organizer |
|
|
Uncertainty in Artificial Intelligence (UAI) |
1997, 1999 |
|
n Organized and maintained online proceedings for the Thirteenth and Fifteenth Annual Conferences on Uncertainty in Artificial Intelligence (http://uai.sis.pitt.edu). |
|
|
Webmaster |
|
|
Intelligent Systems Program, University of Pittsburgh |
2000 – 2002 |
|
WWW Tutorial Designer |
|
|
Department of Physics, Case Western Reserve University |
1993 – 1994 |
|
n Created animations and text for an online html tutorial for liquid crystalline polymers: http://plc.cwru.edu/tutorial/enhanced/main.htm. |
|
awards & honors
§ Performance Networking Lab Recognition Award, 2005
§ Intel Corporate Technology Group Division Achievement Award, 2005
§ Intel Fab 18 Achievement Award, 2005.
§ NASA Graduate Student Research Fellowship, 1999-2002.
§ Dean’s Tuition Scholarship 1999-2002, University of Pittsburgh
§ Phi Eta Sigma, Alpha Lambda Delta, and Golden Key National Honor Societies.
§ Dean’s High Honors, Case Western Reserve University, Spring 1991, Fall 1994
§ Dean’s Honors, Case Western Reserve University, Spring 1994
§ First Prize Project in Classroom Competition for Decision Analysis and Decision Support Systems, Spring, 1997.
Hobbies & interests
§ Ultimate Frisbee.
§ Latin Jazz and Salsa Dancing.
§ Scuba Diving.