Debarun Kar

About Me


PhD Completed! Joining Goldman Sachs in New York as a Big Data Scientist (Associate) starting July, 2017.


Hi, I successfully defended my PhD thesis on April 24, 2017, after spending four amazing years as a doctoral student in the Department of Computer Science at the Viterbi School of Engineering at USC. I was a member of the TEAMCORE Research group led by Prof. Milind Tambe, and the USC Center for AI in Society (CAIS).


My research interests lie at the intersection of Computer Science and Economics, with an emphasis to solving real-world problems in "AI for Social Good", for example, in domains such as Security and Sustainability. More specifically, my PhD research is focused on modeling and learning the boundedly rational behavior of adaptive human adversaries in the following real-world domains: Wildlife Crime and Fisheries Protection. My most recent research on developing anti-poaching decision aids using game theory and machine learning has culminated in initial tests being conducted at the Queen Elizabeth National Park in Uganda and has led to successful results. This has been highlighted in this recent news article. A full-fledged deployment at QENP is currently ongoing. My past research in similar security and sustainability problems with other students has also resulted in deployments in various places around the globe: see here, here, here, here, and here.


Currently, I am working on projects in deep learning to solve challenging social problems that the world is facing. In one of the projects, I am using Convolutional Neural Nets for wildlife conservation using UAVs. Please see Projects section for more details.


I am also interested in applying game theory and machine learning to design effective defender strategies to enhance Cybersecurity. I have previously worked on developing Multiagent Systems for Sustainability.


Before joining USC, I was an MS (Research) student in the Computer Science and Engineering department at the Indian Institute of Technology (IIT) Madras. I was part of two labs : Artificial Intelligence and Databases (AIDB) and the Interactive Intelligence Laboratory (IIL), which is part of the Reconfigurable and Intelligent Systems Engineering (RISE) group. There, I was advised by (in last-name lexicographic order) Dr. Sutanu Chakraborti and Dr. Balaraman Ravindran.


During my MS, I worked on machine learning and case based reasoning to develop and improve the effectiveness of a real-world soil nutrient management system to aid farmers in India to make better decisions about crop cultivation. This involved learning a predictive model of soil nutrient contents based on various soil properties extracted using inexpensive image data. This was a significant advance over the earlier expensive and lengthy chemical testing processes used. The technical aspects of my MS thesis focused on feature weighting, confidence augmented predictions, and similarity learning through interactive visualization of high-dimensional spaces.


A more detailed overview of my projects can be found in the Projects section.

Projects


Publications


2017

D. Kar, "When AI Helps Wildlife Conservation: Learning Adversary Behaviors in Green Security Games", PhD Thesis.


2017

D. Kar, S. Sengupta, E. Kamar, E. Horvitz, M. Tambe, "Believe It or Not: Modeling Adversary Belief Formation in Stackelberg Security Games with Varying Information", (ACS'17).


2016

D. Kar*, B. Ford*, S. Gholami, F. Fang, A. Plumptre, M. Tambe, M. Driciru, F. Wanyama, A. Rwetsiba, "Cloudy with a Chance of Poaching: Adversary Behavior Modeling and Forecasting with Real-World Poaching Data", (AAMAS'17). * indicates joint first author.


2016

D. Kar, F. Fang, F. Delle Fave, N. Sintov, M. Tambe, A. Lyet, "Comparing Human Behavior Models in Stackelberg Security Games: An Extended Study", (Artificial Intelligence, Elsevier Journal).


2016

C. Zhang, S. Gholami, D. Kar, A. Sinha, M. Jain, R. Goyal, M. Tambe, "Keeping Pace with Criminals: An Extended Study of Designing Patrol Allocation against Adaptive Opportunistic Criminals", (Games Journal'16).


2016

Y. Abbasi, N. Ben-Asher, C. Gonzalez, D. Kar, D. Morrison, N. Sintov, M. Tambe, "Know Your Adversary: Insights for a Better Adversarial Behavioral Model", (CogSci'16).


2016

A. Sinha, D. Kar, M. Tambe, "Learning Adversary Behavior in Security Games: A PAC Model Perspective", (AAMAS'16).


2016

N. Sintov, D. Kar, T. Nguyen, F. Fang, K. Hoffman, A. Lyet, M. Tambe, "From the Lab to the Classroom and Beyond: Extending a Game-Based Research Platform for Teaching AI to Diverse Audiences", (Symposium on Educational Advances in Artificial Intelligence (EAAI) 2016).


2015

A. Sinha, T. Nguyen, D. Kar, M. Brown, M. Tambe, A. Jiang, "From Physical Security to Cyber Security", (Journal of Cybersecurity, 2015).


2015

T. Nguyen, D. Kar, M. Brown, A. Sinha, M. Tambe, A. Jiang, "Towards a Science of Security Games", (Book Chapter in New Frontiers of Multi-Disciplinary Research in STEAM-H, 2015).


2015

T. Nguyen, F. Delle Fave, D. Kar, A. Srinivas. L, A. Yadav, M. Tambe, N. Agmon, A. Plumptre, M. Driciru, F. Wanyama, A. Rwetsiba, "Making the most of Our Regrets: Regret-based Solutions to Handle Payoff Uncertainty and Elicitation in Green Security Games", (GameSec'15).


2015

D. Kar, F. Fang, F. Delle Fave, N. Sintov, M. Tambe, " 'A Game of Thrones': When Human Behavior Models Compete in Repeated Stackelberg Security Games", (AAMAS'15).


2015

D. Kar, F. Fang, F. Delle Fave, N. Sintov, M. Tambe, "Conducting Longitudinal Experiments with Behavioral Models in Repeated Stackelberg Security Games on Amazon Mechanical Turk", (Human Agents Interaction and Decision Making (HAIDM) Workshop at AAMAS'15).


2015

D. Kar, F. Fang, F. Delle Fave, N. Sintov, A. Sinha, A. Galstyan, B. An, M. Tambe, "Learning Bounded Rationality Models of the Adversary in Repeated Stackelberg Security Games", (Adaptive Learning Agents (ALA) Workshop at AAMAS'15).


2015

D. Kar, F. Fang, F. Delle Fave, N. Sintov, M. Tambe, A. Van Wissen, "Effectiveness of Probability Perception Modeling and Defender Strategy Generation Algorithms in Repeated Stackelberg Games: An Initial Report", (Computational Sustainability Workshop at AAAI'15)


2014

W. B. Haskell, D. Kar, F. Fang, M. Tambe, S. Cheung and Lt. E. Denicola, "Robust Protection of Fisheries with COmPASS", (IAAI'14).


2014

B. Ford, D. Kar, F. Delle Fave, R. Yang, M. Tambe, "PAWS: Adaptive Game-theoretic Patrolling for Wildlife Protection", (AAMAS'15 - Demo Paper).


2014

J. Kwak, D. Kar, W. B. Haskell, P. Varakantham, M. Tambe, "Building THINC: User Incentivization and Meeting Rescheduling for Energy Savings", (AAMAS'14)


2013

D. Kar, A. Kumar, S. Chakraborti, B. Ravindran, "iCaseViz : Learning Case Similarities through Interaction with a Case Base Visualizer", (ICCBR'13).


2012

D. Kar, S. Chakraborti, B. Ravindran, "Feature Weighting and Confidence based Prediction for Case based Reasoning Systems", (ICCBR'12)


2009

P. P. Choudhury, S. Sahoo, S. Basu, D. Ghosh, D. Kar, A. Ghosh, A. Ghosh, A. K. Ghosh, "Classification of Cellular Automata Rules based on their Properties", (ICONTOP'09).

Curriculum Vitae


News


  • I successfully defended my PhD Thesis on April 24, 2017!


  • I have been organizing tutorials on deep learning this semester. Some slides can be found here.


  • I am co-organizing a workshop on Adversarial Reasoning in Multi-agent Systems (ADVERSE) at AAMAS 2017. Please consider submitting to this workshop. Paper submission deadline is Feb 15, 2017.


  • July 2016: Gave a talk on "Human Behavior Models to Protect Wildlife Crime" at the 4th International Conference on Computational Sustainability held at Cornell University, Ithaca, NY.


  • We won the Best AI Application Video Award at AAAI 2016 held in Phoenix (Arizona, US), for our work on PAWS: Protection Assistant for Wildlife Security.


  • I am organizing a workshop on Security and Multi-agent Systems (SecMAS) at AAMAS 2016, along with Yevgeniy Vorobeychik and Long Tran-Thanh. Please consider submitting to this workshop. Paper submission deadline is Feb 1, 2016.


  • July 2015: Gave an invited talk on "Human Behavior Modeling for Physical and Cyber Security" at the Army Research Office (ARO) Workshop on Cyber Deception at George Mason University.


  • July 2015: Gave an invited talk on "SHARP: A Novel Human Behavior Model in Repeated Stackelberg Security Games" at the 22nd International Symposium on Mathematical Programming (ISMP) in Pittsburgh.


  • May 2015: Presented my AAMAS main paper " 'A Game of Thrones': When Human Behavior Models Compete in Repeated Stackelberg Security Games" at AAMAS 2015 held in Istanbul, Turkey.


  • May 2015: Presented my AAMAS workshop paper "Conducting Longitudinal Experiments with Behavioral Models in Repeated Stackelberg Security Games on Amazon Mechanical Turk" at the Human Agents Interaction and Decision Making (HAIDM) Workshop at AAMAS 2015 held in Istanbul, Turkey.


  • May 2015: Presented my AAMAS workshop paper "Learning Bounded Rationality Models of the Adversary in Repeated Stackelberg Security Games" at the Adaptive Learning Agents (ALA) Workshop at AAMAS 2015 held in Istanbul, Turkey.


  • February 2015: Gave a talk on "Game Theory and Machine Learning in Green Security Games" at the Indian Institute of Technology (IIT) Madras.


  • January 2015: Presented our AAAI workshop paper "Effectiveness of Probability Perception Modeling and Defender Strategy Generation Algorithms in Repeated Stackelberg Games: An Initial Report" at the AAAI Workshop on Computational Sustainability 2014 held in Austin, Texas.


  • July 2014: Presented our IAAI main paper "Robust Protection of Fisheries with COmPASS" at IAAI 2014 held in Quebec, Canada.


  • June 2014: Gave an invited talk on "Game Theoretic Applications in Computational Sustainability" at the USC Civil Engineering Department Seminar.


  • May 2014: Presented our AAMAS main paper "Building THINC: User Incentivization and Meeting Rescheduling for Energy Savings" at AAMAS 2014 held in Paris, France.


  • 2013: Presented my ICCBR main paper "Feature Weighting and Confidence based Prediction for Case based Reasoning Systems" at ICCBR 2013 held in Lyon, France.