Jaeyoung Lee

Research Associate

Prior affiliations 

  • 2015.09 - 2017.12  Postdoctoral Fellow at the Reinforcement Learning and Artificial Intelligence Lab. at the University of Alberta, AB, Canada.

Selected publications

  • Constrained RL for safety-critical systems
    • Lee, J., Sedwards, S. & Czarnecki, K. (2021). Uniformly constrained reinforcement learning, submitted to Journal of Autonomous Agents and Multi-Agent Systems (JAAMAS): Special Issue on Multi-Objective Decision Making (MODeM) — under review.
    • Lee, J., Sedwards, S. & Czarnecki, K. (2021). Recursive constraints to prevent instability in constrained reinforcement learning. In: Proc. 1st Multi-Objective Decision Making Workshop (MODeM 2021), Hayes, Mannion, Vamplew (eds.). Online at http://modem2021.cs.nuigalway.ie.
  • Distillation and imitation of deep Q-Network by decision tree, for formal verification
    • Abdelzad, V., Lee, J., Sedwards, S., Soltani S. & Czarnecki, K. (2021). Non-divergent imitation for verification of complex learned controllers. In: 2021 International Joint Conference on Neural Networks (IJCNN). doi:10.1109/IJCNN52387.2021.9533410. Shenzhen, China (virtual).

    • Jhunjhunwala, A., Lee, J., Sedwards S., Abdelzad V. & Czarnecki K. (2020). Improved policy extraction via online Q-value distillation. In: 2020 International Joint Conference on Neural Networks (IJCNN, in 2020 IEEE WCCI). doi:10.1109/IJCNN48605.2020.9207648. Glasgow, U.K. (cited:12, including 8 citations of arXiv)

  • Deep RL for ADS
    • Lee, J., Balakrishnan, A., Gaurav, A., Czarnecki, K. & Sedwards, S. (2019). WISEMOVE: a framework to investigate safe deep reinforcement learning for autonomous driving. In: Parker D., Wolf V. (eds) Quantitative Evaluation of Systems. QEST 2019. Lecture Notes in Computer Science, vol. 11785. Springer, Cham. doi:10.1007/978-3-030-30281-8 20. Glasgow, U.K.
    • Balakrishnan, A., Lee, J., Gaurav A., Czarnecki, K. & Sedwards, S. (2021). Transfer reinforcement learning for autonomous driving: from WISEMOVE to WISESIM. ACM Transactions on Modeling and Computer Simulation (TOMAC), 31 (3), article no. 15, 26 pages. doi:10.1145/3449356.
  • Deep RL with advanced prioritization methods
    • Lee, S., Lee, J. & Hasuo I. (2020 & 2021). Predictive PER: balancing priority and diversity towards stable
      deep reinforcement learning. In: (a) Deep Reinforcement Learning Workshop in 2020 NeurIPS. Virtual; (b) 2021 International Joint Conference on Neural Networks (IJCNN). doi:10.1109/IJCNN52387.2021.9534243. Shenzhen, China (virtual). (cited: 2)

  • RL, Dynamic Programming and optimal control in continuous domain
    • Lee, J. & Sutton, R.S. (2021). Policy iterations for reinforcement learning problems in continuous time and space: fundamental theory and methods. Automatica, 126, 109421, 15 pages. doi:10.1016/j.automatica.2020.109421 (cited: 7)

    • Lee, J.Y., Park, J.B. & Choi, Y.H. (2012). Integral Q-learning and explorized policy iteration for adaptive optimal control of continuous-time linear systems. Automatica, 48 (11), pp. 2850–2859. doi:10.1016/j.automatica.2012.06.008 (cited: 140)

    • Lee, J.Y., Park, J.B. & Choi, Y.H. (2014). Integral reinforcement learning for a class of nonlinear systems with invariant explorations. IEEE Transactions on Neural Networks and Learning Systems, 26 (5), pp. 916–932. doi:10.1109/TNNLS.2014.2328590. (cited: 78).

  • Multi-agent consensus techniques and their applications to vehicles’ formation control
    • Lee, J.Y., Choi, Y.H. & Park, J.B. (2014). Inverse optimal design of the distributed consensus protocol for formation control of multiple mobile robots. In: Proc. 53rd IEEE Conference on Decision and Control (CDC), pp. 2222–2227. doi:10.1109/CDC.2014.7039728. Los Angeles, CA, USA. (cited: 4).