Erik Hemberg, Ph.D

Research Scientist

AnyScale Learning For All (ALFA) Group
MIT Computer Science and Artificial Intelligence Lab


Email: hembergerik (@)
Office: 32-D540 Stata Center










Erik Hemberg is a Research Scientist with ALFA Group at MIT-CSAIL. His work focuses on developing autonomous, proactive cyber defenses that are anticipatory and adapt to counter attacks. Novel methods for: cyber-hunting, automated methods of cyber-attack and defense scenarios in Software Defined and Peer-to-Peer Networks, models of networks and a proof-of-concept for how adversarial modeling can inform adaptive cyber security defenses and pro-actively design better network protocols. Additional work includes: automated semantic parsing of law, predicting stop-out in Massive Open Online Courses (MOOCs) and analyzing neuronal development of autism. Research sponsorships include: cybersecurity research with DARPA XD3, DARPA CHASE program, Lincoln Labs and CyberSecurity @CSAIL. Data science for MOOC with MIT-HKUST alliance. Erik also supervises Ph.Ds, SM, M.Eng students and UROPs.

Most recently, Erik won: Best Student Paper Award at IEEE Learning With MOOCs 2019, Outstanding Poster IBM Research AI Horizons Colloquium 2018; HUMIES Bronze award at GECCO 2017; Co-author of Best Paper at the NIPS 2016 Machine Learning for Health workshop; is a member of the founding team of the YCombinator Fellowship, 2016. Erik's work was featured in the NY Times article, Computer Scientists Wield Artificial Intelligence to Battle Tax Evasion, in 2015.

Erik completed his M.Sc in Industrial Engineering and Management at Chalmers University of Technology, Gothenburg, Sweden in 2003. He went on to get his Ph.D. in Computer Science from the University College Dublin, Computer Science and Informatics Department, Dublin, Ireland in 2010 as a member of the Natural Computing Research & Applications (NCRA) in UCD's Complex and Adaptive Systems Lab (CASL). Erik continued on as a Postdoc with NCRA group in CASL, University College Dublin, where he lead industrial collaboration with Bell Labs Dublin, researched Evolutionary Computation as a method for creating pilot power control algorithms for Femtocells. He worked on developing Grammatical Evolutions in Matlab in addition to hiring and managing a research assistant for 12 weeks, supervising Ph.D. students and final year undergraduate projects.

Erik joined ALFA Group as a Postdoc in 2012 where he lead research in Simulating Tax Evasion and Law with Heuristics with MITRE Corp, collaborated with Sentient Ltd researching Evolutionary Computation on Volunteer Compute for Machine Learning. Erik developed and delivered a "blended" online course in Evolutionary Processes and Systems for Shantou University in China and material to US kids. He assisted grant writing for cybersecurity research projects and supervised S.M., M.Eng. students and UROPs. Erik is the author of a multitude of academic papers, journal articles, and scholarly book chapters.

Awards & Honors

2019   Best Student Paper, IEEE Learning With MOOCs 2019, Milwaukee WI

2018   Outstanding Poster IBM Research AI 2018 AI Horizons Colloquium, Cambridge MA

2017   HUMIES Bronze award at the GECCO Conference, Berlin Germany

2016   - Co-author of Best Paper at the NIPS 2016 Machine Learning for Health workshop
           - YCombinator Fellowship 2016, member of founding team

2015   - MIT Translational Fellow Award Fall 2015, 20% of salary funded for pursuing translation of research to technology
           - Peter Jackson Award for Best Innovative Application Paper at 15th ICAIL for STEALTH paper
           - STEALTH won finalist highlight award at the MIT CSAIL Amazing Research Highlights Competition
           - Supervised best thesis for Computer Aided Tax Avoidance Policy Analysis from the Technology and Policy Program

Top 3 Publications

2015   E. Hemberg, J Rosen, G Warner, S Wijesinghe, UM O'Reilly. "Tax non-Compliance Detection Using Co-Evolution of Tax Evasion Risk and Audit Likelihood." International Conference on AI and Law, 2015, San Diego, (Award for Best Innovative Application Paper)

2013   E. Hemberg, L. Ho, M. O'Neill, and H. Claussen. "A comparison of grammatical genetic programming grammars for controlling femtocell network coverage." Genetic Programming and Evolvable Machines 14 (1), 65-93, 2013

2012   E. Hemberg, K Veeramachaneni, J McDermott, C Berzan, and UM O'Reilly. "An investigation of local patterns for estimation of distribution genetic programming." GECCO, 2012

For additional publications please see the ALFA Publications page

Selected Invited Talks

Analyzing Educational Data With MOOC-Learner-Project, Summer School, Carnegie Mellon University; CITE, Hong Kong University; for MIT-ILP at Kawasaki Heavy Industries, ASICS, Toshiba, Japan, July 2018

Proactive Cyber Security, for MIT-ILP at Hitatchi, Konica Minolta, Japan, July 2018

Algorithmic Simulation of Tax Avoidance and Automated Representation of Law, Institute Conference on Improving Tax System at Northwestern Law, Law & Technology Seminar Series, March 2017

Simulating Tax Evasion And Law Through Heuristics, Institute Conference on Improving Tax System, Washington DC, November 2016

Computing in High-Energy AstroParticle Research, Genetic Programming Workshop at Ohio State University, August 2016

Simulating Tax Evasion And Law Through Heuristics, CodeX Seminar, Stanford University, September 2015