Alexandre Megretski

Alexandre Megretski is a professor in MIT’s Department of Electrical Engineering and Computer Science, and a member of the Laboratory for Information and Decision Systems. His research interests include nonlinear system identification and model reduction, nonlinear dynamical system analysis, design and validation of hybrid control algorithms, optimization of nonlinear robust controllers, optimization, and applications such as nonlinear circuits, live neuron modeling, analog-digital conversion, and digital pre-distortion. Previously, he was a researcher at the Royal Institute of Technology in Sweden; University of Newcastle, Australia; and a faculty member at Iowa State University, Ames. He was also a co-founder of NanoSemi Inc.

Selected Publications

• Zhang, W.,Nguyen, L. M., Weng, L., Das, S., Megretski, A., Daniel, L. (2023). ConCerNet: A Contrastive Learning Based Framework for Automated Conservation Law Discovery and Trustworthy Dynamical System Prediction. International Conference on Machine Learning (ICML).
• Das, S., Daniel, L., Weng, L., Nguyen, L. M., Zhang, W., Megretski, A. (2022). Fast Convergence for Unstable Reinforcement Learning Problems by Logarithmic Mapping. International Conference on Machine Learning (ICML) Workshop on Decision Awareness in Reinforcement Learning.
• Weng, L., Oikarinen, T., Daniel, L., Megretski, A., Zhang, W. (2021). Robust Deep Reinforcement Learning through Adversarial Loss. Conference on Neural Information Processing Systems (NeurIPS).

Media

• July 3, 2019: MIT News, Making wireless communication more energy efficient.