Chelsea Sidrane
Postdoctoral Fellow
chelse@kth.se
Hi! Nice to meet you.
I am a Digital Futures Postdoctoral Fellow at the KTH Royal Institute of Technology in Stockholm. I am working with Jana Tumova in the Divison of Robotics, Perception and Learning. I'm also a PhD graduate from the Stanford Intelligent Systems Lab (SISL) in the Stanford University Department of Aeronautics and Astronautics, advised by Mykel Kochenderfer.
I have been very lucky to be able to collaborate with a talented postdogtoral scholar, Jazz.
I'm currently on the faculty / group leader job market.
Research Interests
In my current position at KTH, I am applying my skills -- a combination of dynamical systems and control theory, machine learning, and formal methods -- to problems of control and planning for underwater robotic vehicles with an eye towards environmental monitoring and conservation.
My PhD thesis focused on methods for assuring the safety and reliability of machine learning systems. This is especially important when machine learning is used in safety-critical applications like cars, planes, and spacecraft. In particular, I spent four years developing methods to verify neural networks used within nonlinear systems.
Broadly, I am interested in applying control, planning, machine learning and formal methods to socially-relevant, climate-relevant and human-focused problems such as (but not limited to) disaster prediction (floods, earthquakes, fires), pre-screening for disease diagnosis, development of prosthetic devices and other assistive technology, and increased access to quality education.
If you are interested in these areas as well and would like to connect please reach out to me at chelse@kth.se.
Publications
Under Review
Chelsea Sidrane and Jana Tumova. TTT: A Temporal Refinement Heuristic for Tenuously Tractable Discrete Time Reachability Problems. 2024. preprint.
Chelsea Sidrane, Sydney Katz, Anthony Corso, Mykel Kochenderfer. Verifying Inverse Model Neural Networks. 2022. https://arxiv.org/abs/2202.02429
Peer-Reviewed Journal Papers
Nicholas Rober, Sydney Katz, Chelsea Sidrane, Esen Yel, Michael Everett, Mykel J. Kochenderfer, Jonathon P. How. Backward Reachability Analysis of Neural Feedback Loops: Techniques for Linear and Nonlinear Systems. 2023.
Chelsea Sidrane*, Amir Maleki*, Ahmed Irfan, Mykel Kochenderfer. OVERT: An Algorithm for Safety Verification of Neural Network Control Policies for Nonlinear Systems. Journal of Machine Learning Research (JMLR), 2022. http://jmlr.org/papers/v23/21-0847.html (* denotes equal contribution)
Peer-Reviewed Conference Papers
Michael Kelly, Chelsea Sidrane, Katherine Driggs-Campbell and Mykel J. Kochenderfer. HG-DAgger: Interactive Imitation Learning with Human Experts. International Conference on Robotics and Automation (ICRA), 2019. https://arxiv.org/abs/1810.02890
Chelsea Sidrane, Mykel J. Kochenderfer. Closed-Loop Planning for Disaster Evacuation with Stochastic Arrivals. IEEE International Conference on Intelligent Transportation Systems (ITSC), 2018. https://drive.google.com/open?id=1l4Quv2XuYloIRcbrATVhYax8eJVku2ip
Peer-Reviewed Workshop Papers
Chelsea Sidrane. Safety assurance for systems with machine learning components. Doctoral Consortium, AAAI Conference on Artificial Intelligence (AAAI), 2021.
Taylor T. Johnson, Diego Manzanas Lopez, Patrick Musau, Hoang-Dung Tran, Elena Botoeva, Francesco Leofante, Amir Maleki, Chelsea Sidrane, Jiameng Fan, Chao Huang. ARCH-COMP20 Category Report: Artificial Intelligence and Neural Network Control Systems (AINNCS) for Continuous and Hybrid Systems Plants. International Workshop on Applied Verification of Continuous and Hybrid Systems, 2020. https://easychair.org/publications/paper/Jvwg
Chelsea Sidrane, Dylan J. Fitzpatrick, Andrew Annex, Diane O'Donoghue, Yarin Gal, and Piotr Biliński. Machine Learning for Generalizable Prediction of Flood Susceptibility. NeurIPS 2019 Artificial Intelligence for Humanitarian Assistance and Disaster Response Workshop. arXiv:1910.06521 (2019).
Chelsea Sidrane and Mykel J. Kochenderfer. OVERT: Verification of Nonlinear Dynamical Systems with Neural Network Controllers via Overapproximation. SafeML ICLR 2019. https://drive.google.com/uc?export=download&id=17dj7HXXzjGymoh1VUm4sfb7ynQ_R0Xem
Michael Kelly, Chelsea Sidrane, Katherine Driggs-Campbell and Mykel J. Kochenderfer. Safe Interactive Imitation Learning from Humans. NeurIPS 2018 Imitation Learning and its Challenges in Robotics. https://drive.google.com/open?id=18rJr7eXPa9rEvapyrQOfjgyQtSlMd03y
Open Source Software
OVERT.jl
Nonlinear Overapproximation
https://github.com/sisl/OVERT.jl
OVERTVerify.jl
Reachability for Nonlinear Systems with Neural Network Control Policies
Background
I was born and raised in the suburbs of New York City. I received my Bachelor's degree in Mechanical Engineering from Cornell University in 2016 and my Master's and PhD degrees from Stanford University in 2022. My interests include singing, comedy, cooking, skiing, hiking, ultimate frisbee, and memes.