Alexey Tregubov

Ongoing research projects

Alexey’s recent research projects include the development of DASH - a large-scale agent-based simulation framework. The purpose of the DASH framework is to provide a high-fidelity computational simulation of the spread and evolution of narratives via social networks. Alexey together with his ISI team developed a framework architecture that combined discrete event simulation with an ensemble of predictive ML models in the agent’s decision process. To achieve a high-fidelity performance of individual agents, the framework aggregates models for time-series prediction of activation time, activity burstiness prediction, and response impact/engagement prediction. This approach won several SocialSim challenges and was acknowledged by field experts in leading conferences and journals (AAMAS, PAAMS, JAAMAS). Additionally, this architecture design proved effective for generating synthetic data for augmented training datasets.

Alexey’s other research projects also include the development of a vulnerability prediction model in Linux kernel patches using signals from social interactions of the contributors in the Linux kernel mailing list (LKML). The project aims to identify patches (potential git commits) that are likely to contain future common vulnerabilities and exposures (CVEs), which helps kernel maintainers stay vigilant and prevent risky patches before they are pushed upstream. Alexey designed a pipeline incorporating LLMs, sentiment analysis, and rule learning models for explainability.

Alexey is also a skilled software engineer with expertise in Python, Java, TensorFlow, PyTorch, Keras, Scikit-learn, XGBoost, and Gurobi.

Selected Publications

Modeling cognitive workload in open-source communities via simulation. Tregubov, A.; Abramson, J.; Hauser, C.; Hussain, A.; and Blythe, J. In AAMAS International Workshop on Multi-Agent-Based Simulation, 2023.

Dynamic graph reduction optimization technique for interdiction games. Blythe, J.; and Tregubov, A. In AAMAS Workshop on Optimization and Learning in Multiagent Systems, 2022.

Large-scale agent-based simulations of online social networks. Murić, G.; Tregubov, A.; Blythe, J.; Abeliuk, A.; Choudhary, D.; Lerman, K.; and Ferrara, E. Autonomous Agents and Multi-Agent Systems, 36(2): 38. 2022. Top conference: Google Scholar H5-index: 25, acceptance rate 24%

Optimization of Large-scale Agent-based Simulations through Automated Abstraction and Simplification, Tregubov A., Blythe J., In Proceedings of International Conference on Autonomous Agents and Multi-Agent Systems, May 2020. Top conference: Google Scholar H5-index: 25, acceptance rate 24%

The DARPA SocialSim Challenge: Cross-platform Multi-Agent Simulations, Muric G., Tregubov A., Blythe J., Ferrara E., In Proceedings of International Conference on Autonomous Agents and Multi-Agent Systems, May 2020. Top conference: Google Scholar H5-index: 25, acceptance rate 24%

The DARPA SocialSim Challenge: Massive Multi-Agent Simulations of the GitHub Ecosystem, Blythe J., Ferrara E., Lerman K., Tregubov A., Muric G., In Proceedings of International Conference on Autonomous Agents and Multi-Agent Systems, 13th-17th of May 2019. Top conference: Google Scholar H5-index: 25, acceptance rate 25%

Massive Multi-Agent Data-Driven Simulations of the GitHub Ecosystem, Blythe J., Ferrara E., Lerman K., Tregubov A., Muric G., In Proceedings of International Conference on Practical Applications of Agents and Multi-Agent Systems, 26th-28th June, 2019.

FARM: Architecture for Distributed Agent-based Social Simulations, Blythe J., Tregubov A., In Proceedings of International Workshop on Massively Multi-Agent Systems, July 14th, 2018.

Impact of Task Switching and Work Interruptions on Software Development Processes, Tregubov A., Boehm B., Rodchenko N., Lane, J.A.; In Proceedings of International Conference on Software and Systems Process (ICSSP’17), Paris, France, 5-7 July, 2017. Top conference: Google Scholar H5-index: 15, acceptance rate 21.9%

Evaluation of cross-project multitasking in software projects, Tregubov A., Lane, J.A., Boehm B.; Conference on Systems Engineering Research (CSER’17), Los Angeles, CA, 23-25 March, 2017.

What does it mean to be Lean in SoSE environment? Tregubov A., Lane, J.A.; 26th Annual INCOSE International Symposium (IS’16) Edinburgh, Scotland, UK, July 18-21, 2016.

Simulation of Kanban-based scheduling for systems of systems: initial results, Tregubov A., Lane, J.A.; Conference on Systems Engineering Research (CSER’15), Hoboken, NJ, 17-19 March, 2015.