Research

• I work on practical algorithms for NP-hard search and optimization problems that arise in the decision, control, and design aspects of computational systems.
• I focus on topics that can benefit from the combinatorial perspectives of automated reasoning but are posed in numerical or statistical forms.
• The end goal is to build useful forms of automation and autonomy that are fundamentally reliable yet aggressively optimized.
• Colloquially, I spend much time dealing with the curse of dimensionality and finding needles in haystacks.

Teaching

• UCSD CSE150: Introduction to AI: Search and Reasoning (Spring 2025, Spring 2024, Spring 2023, Spring 2022, Fall 2020, Spring 2020, Winter 2019, Winter 2018)
• UCSD CSE257: Search and Optimization (Winter 2025, Winter 2024, Winter 2023, Fall 2021, Spring 2021)
• UCSD CSE191: Undergraduate Thesis Research Seminar (Fall 2021)
• UCSD CSE291: Topics in Search and Optimization (Winter 2020, Spring 2019, Fall 2017)

• PhD Students: Ya-Chien Chang, Yejin Jeon, Janna Lin (with Sylvia Herbert), Matthew Taylor, Tao Wang, Hongzhan Yu, Ruipeng Zhang
• Alumni: Yongsoo Song (Postdoc '18-'19, then Assistant Professor at Seoul National University), Jacek Cyranka (Postdoc '17-'19, then Assistant Professor at University of Warsaw), Armaiti Ardeshiricham (PhD '20, then Researcher at Apple), Nima Roohi (Postdoc '18-'20, then Applied Scientist at Amazon), Yaoguang Zhai (PhD '24, then Applied Scientist at Amazon), Yuda Song (BS '20, then PhD student at CMU), Milan Ganai (BS/MS '24, then PhD student at Stanford), Eric Yu (BS/MS '24, then PhD student at MIT), Chenning Yu (PhD '25, then ML Engineer at Nvidia), Zhizhen Qin (PhD '25, then Applied Scientist at Amazon)
• Our work is funded by the following sources: Air Force Young Investigator Award, Amazon Research Award, DARPA Assured Autonomy, Department of Energy, NSF AI Institute, NSF Career Award, NSF DASS, NSF National Robotics Initiative, NSF Cyber-Physical Systems.

• dReal [tool page]: Automated reasoning in nonlinear theories over the reals

• When Maximum Entropy Misleads Policy Optimization [arXiv]
  Ruipeng Zhang, Ya-Chien Chang, and Sicun Gao
  ICML (International Conference on Machine Learning) 2025
• Improving Value Estimation Critically Enhances Vanilla Policy Gradient [arXiv]
  Tao Wang, Ruipeng Zhang, and Sicun Gao
  ICML (International Conference on Machine Learning) 2025
• Improving Compositional Generation with Diffusion Models Using Lift Scores [project page]
  Chenning Yu and Sicun Gao
  ICML (International Conference on Machine Learning) 2025
• Estimating Control Barriers from Offline Data [arXiv]
  Hongzhan Yu, Seth Farrell, Ryo Yoshimitsu, Zhizhen Qin, Henrik Christensen, and Sicun Gao
  ICRA (IEEE International Conference on Robotics and Automation) 2025
• Patching Approximately Safe Value Functions Leveraging Local Hamilton-Jacobi Reachability Analysis [arXiv]
  Sander Tonkens, Alex Toofanian, Zhizhen Qin, Sicun Gao, and Sylvia Herbert
  CDC (IEEE Conference on Decision and Control) 2024
• SEEV: Synthesis with Efficient Exact Verification for ReLU Neural Barrier Functions [arXiv]
  Hongchao Zhang, Zhizhen Qin, Sicun Gao, and Andrew Clark
  NeurIPS (Conference on Neural Information Processing Systems) 2024
• Hamilton-Jacobi Reachability in Reinforcement Learning: A Survey [arXiv]
  Milan Ganai, Sicun Gao, and Sylvia Herbert
  IEEE Open Journal of Control Systems 2024
• Activation-Descent Regularization for Input Optimization of ReLU Networks [arXiv]
  Hongzhan Yu and Sicun Gao
  ICML (International Conference on Machine Learning) 2024
• Mollification Effects of Policy Gradient Methods [arXiv]
  Tao Wang, Sylvia Herbert, and Sicun Gao
  ICML (International Conference on Machine Learning) 2024
• Breaking the Barrier: Enhanced Utility and Robustness in Smoothed DRL Agents
  Chung-En Sun, Sicun Gao, and Tsui-Wei Weng
  ICML (International Conference on Machine Learning) 2024
• Extremum-Seeking Action Selection for Accelerating Policy Optimization [project page]
  Ya-Chien Chang and Sicun Gao
  ICRA (International Conference on Robotics and Automation) 2024
• Efficient Motion Planning for Manipulators with Barrier-Induced Neural Controller
  Mingxin Yu, Chenning Yu, Mohammad Mahdi Naddaf Shargh, Devesh Upadhyay, Sicun Gao, and Chuchu Fan
  ICRA (International Conference on Robotics and Automation) 2024
• Sample-and-Bound for Non-Convex Optimization [arXiv]
  Yaoguang Zhai, Zhizhen Qin, and Sicun Gao
  AAAI (AAAI Conference on Artificial Intelligence) 2024
• Iterative Reachability Estimation for Safe Reinforcement Learning [project page]
  Milan Ganai, Zheng Gong, Chenning Yu, Sylvia Herbert, and Sicun Gao
  NeurIPS (Conference on Neural Information Processing Systems) 2023
• Fractal Landscapes in Policy Optimization [arXiv]
  Tao Wang, Sylvia Herbert, and Sicun Gao
  NeurIPS (Conference on Neural Information Processing Systems) 2023
• Sequential Neural Barriers for Scalable Dynamic Obstacle Avoidance [project page]
  Hongzhan Yu, Chiaki Hirayama, Chenning Yu, Sylvia Herbert, and Sicun Gao
  IROS (International Conference on Intelligent Robots and Systems) 2023 [IROS RoboCup Best Paper Award]
  
• Learning Stabilization Control from Observations by Learning Lyapunov-like Proxy Models [project page]
  Milan Ganai, Chiaki Hirayama, Ya-Chien Chang, and Sicun Gao
  ICRA (International Conference on Robotics and Automation) 2023
• Accelerating Multi-Agent Planning Using Graph Transformers with Bounded Suboptimality [project page]
  Chenning Yu, Qingbiao Li, Sicun Gao, and Amanda Prorok
  ICRA (International Conference on Robotics and Automation) 2023
• Everyone's Preference Changes Differently: A Weighted Multi-Interest Model For Retrieval
  Hui Shi, Yupeng Gu, Yitong Zhou, Bo Zhao, Sicun Gao, and Jishen Zhao
  ICML (International Conference on Machine Learning) 2023
• Safe Control with Learned Certificates: A Survey of Neural Lyapunov, Barrier, and Contraction Methods [arXiv]
  Charles Dawson, Sicun Gao, and Chuchu Fan
  IEEE Transactions on Robotics (T-RO) 2023
• Monte Carlo Tree Descent for Black-Box Optimization [arXiv]
  Yaoguang Zhai and Sicun Gao
  NeurIPS (Conference on Neural Information Processing Systems) 2022
• Learning-based Motion Planning in Dynamic Environments Using GNNs and Temporal Encoding [arXiv]
  Ruipeng Zhang, Chenning Yu, Jingkai Chen, Chuchu Fan, and Sicun Gao
  NeurIPS (Conference on Neural Information Processing Systems) 2022
• Policy Optimization with Advantage Regularization for Long-Term Fairness in Decision Systems [arXiv]
  Eric Yang Yu, Zhizhen Qin, Min Kyung Lee, and Sicun Gao
  NeurIPS (Conference on Neural Information Processing Systems) 2022
• Learning Control Admissibility Models with Graph Neural Networks for Multi-Agent Navigation [project page]
  Chenning Yu, Hongzhan Yu, and Sicun Gao
  CoRL (Conference on Robot Learning) 2022
• Quantifying Safety of Learning-based Self-Driving Control Using Almost-Barrier Functions [arXiv]
  Zhizhen Qin, Tsui-Wei Weng, and Sicun Gao
  IROS (International Conference on Intelligent Robots and Systems) 2022
• Learning Bounded Context-Free Grammar via LSTM and Transformers: Differences and Explanations [arXiv]
  Hui Shi, Sicun Gao, Yuandong Tian, Xinyun Chen, and Jishen Zhao
  AAAI (AAAI Conference on Artificial Intelligence) 2022
• Reducing Collision Checking for Sampling-Based Motion Planning Using Graph Neural Networks [project page]
  Chenning Yu and Sicun Gao
  NeurIPS (Conference on Neural Information Processing Systems) 2021
• A Neural Lyapunov Approach to Transient Stability Assessment of Power Electronics-interfaced Networked Microgrids [pdf]
  Tong Huang, Sicun Gao, and Le Xie
  IEEE Transactions on Smart Grid 2021 [IEEE Power & Enery Society Technical Committee Prize Paper Award]
• Safe Nonlinear Control Using Robust Neural Lyapunov-Barrier Functions [arXiv]
  Charles Dawson, Zengyi Qin, Sicun Gao, and Chuchu Fan
  CoRL (Conference on Robot Learning) 2021
• Stabilizing Neural Control Using Self-Learned Almost Lyapunov Critics [project page]
  Ya-Chien Chang and Sicun Gao
  ICRA (International Conference on Robotics and Automation) 2021
• Provably Efficient Model-based Policy Adaptation [project page]
  Yuda Song, Aditi Mavalankar, Wen Sun, and Sicun Gao
  ICML (International Conference on Machine Learning) 2020
• Active Learning of Many-body Configuration Space: Application to the Cs+-water MB-nrg Potential Energy [chemArxiv]
  Yaoguang Zhai, Alessandro Caruso, Sicun Gao, and Francesco Paesani
  JCP (Journal of Chemical Physics) 2020
• Neural Lyapunov Control [project page]
  Ya-Chien Chang, Nima Roohi, and Sicun Gao
  NeurIPS (Conference on Neural Information Processing Systems) 2019
• VeriSketch: Synthesizing Secure Hardware Designs with Timing-Sensitive Information Flow Properties [pdf]
  Armaiti Ardeshiricham, Yoshiki Takashima, Sicun Gao, and Ryan Kastner
  CCS (ACM Conference on Computer and Communications Security) 2019
• Numerically-Robust Inductive Proof Rules for Continuous Dynamical Systems [pdf]
  Sicun Gao, James Kapinski, Jyotirmoy Deshmukh, Nima Roohi, Armando Solar-Lezama, Nikos Arechiga, and Soonho Kong
  CAV (International Conference on Computer Aided Verification) 2019
• Inverse Abstraction of Neural Networks Using Symbolic Interpolation [pdf]
  Sumanth Dathathri, Sicun Gao, and Richard M. Murray
  AAAI (AAAI Conference on Artificial Intelligence) 2019
• Tight Continuous-Time Reachtubes for Lagrangian Reachability [pdf]
  Jacek Cyranka, Md. Ariful Islam, Scott Smolka, Sicun Gao, and Radu Grosu
  CDC (IEEE Conference on Decision and Control) 2018
• Delta-Decision Procedures for Exists-Forall Problems over the Reals [pdf]
  Soonho Kong, Armando Solar-Lezama, and Sicun Gao
  CAV (International Conference on Computer Aided Verification) 2018
• LEMAX: Learning-based Energy Consumption Minimization in Approximate Computing with Quality Guarantee
  Vahideh Akhlaghi, Sicun Gao, and Rajesh Gupta
  DAC (Design Automation Conference) 2018
• Modular SMT-Based Analysis of Nonlinear Hybrid Systems [pdf]
  Kyungmin Bae and Sicun Gao
  FMCAD (Formal Methods in Computer-Aided Design) 2017
• Learning-Based Abstractions for Nonlinear Constraint Solving [pdf]
  Sumanth Dathathri, Nikos Arechiga, Sicun Gao, and Richard M. Murray
  IJCAI (International Joint Conference on Artificial Intelligence) 2017
• Interpolants in Nonlinear Theories over the Reals [pdf]
  Sicun Gao and Damien Zufferey
  TACAS (International Conference on Tools and Algorithms for the Construction and Analysis of Systems) 2016
• APEX: A Tool for Autonomous Vehicle Plan Verification and Execution [pdf]
  Matthew O'Kelly, Houssam Abbas, Sicun Gao, Shin'ichi Shiraishi, Shinpei Kato, and Rahul Mangharam
  SAE (Society of Automotive Engineers) World Congress and Exhibition 2016
  
• SMT-Based Analysis of Virtually Synchronous Distributed Hybrid Systems [pdf]
  Kyungmin Bae, Peter Olveczky, Soonho Kong, and Sicun Gao
  HSCC (ACM International Conference on Hybrid Systems: Computation and Control) 2016
• Automated Vulnerability Analysis of AC State Estimation under Constrained False Data Injection in Electric Power Systems [pdf]
  Sicun Gao, Le Xie, Armando Solar-Lezama, Dimitrios Serpanos, and Howard Shrobe
  CDC (IEEE Conference on Decision and Control) 2015
• dReach: Delta-Reachability Analysis for Hybrid Systems [pdf]
  Soonho Kong, Sicun Gao, Wei Chen, and Edmund Clarke
  TACAS (International Conference on Tools and Algorithms for the Construction and Analysis of Systems) 2015
• SMT-Based Nonlinear PDDL+ Planning [pdf]
  Daniel Bryce, Sicun Gao, David Musliner, and Robert P. Goldman
  AAAI (AAAI Conference on Artificial Intelligence) 2015
• Towards Personalized Prostate Cancer Therapy Using Delta-Reachability Analysis [pdf]
  Bing Liu, Soonho Kong, Sicun Gao, Paolo Zuliani, and Edmund Clarke
  HSCC (ACM International Conference on Hybrid Systems: Computation and Control) 2015
• Parameter Synthesis for Cardiac Cell Hybrid Models Using Delta-Decisions [arXiv]
  Bing Liu, Soonho Kong, Sicun Gao, Paolo Zuliani, and Edmund Clarke
  CMSB (Computational Methods in Systems Biology) 2014
• Proof Generation from Delta-Decisions [arXiv]
  Sicun Gao, Soonho Kong, and Edmund Clarke
  SYNASC (International Conference on Symbolic and Numerical Algorithms for Scientific Computing) 2014
• Satisfiability Modulo ODEs [pdf]
  Sicun Gao, Soonho Kong, and Edmund Clarke
  FMCAD (Formal Methods in Computer-Aided Design) 2013
• Floating-point Bugs in the Embedded GNU C Library [pdf]
  Soonho Kong, Sicun Gao, and Edmund Clarke
  CMU SCS Technical Report CMU-CS-13-130 2013
• dReal: An SMT Solver for Nonlinear Theories of Reals [pdf] [tool]
  Sicun Gao, Soonho Kong, and Edmund Clarke
  CADE (International Conference on Automated Deduction) 2013
• Computable Analysis, Hybrid Automata, and Decision Procedures: A New Framework for the Formal Verification of Cyber-Physical Systems [extended abstract]
  PhD Thesis in Logic, Carnegie Mellon University 2012
  Committee: Edmund Clarke (co-chair), Jeremy Avigad (co-chair), Lenore Blum, Randy Bryant, and Jeannette Wing
  
• Delta-Complete Decision Procedures for Satisfiability over the Reals [pdf]
  Sicun Gao, Jeremy Avigad, and Edmund Clarke
  IJCAR (International Joint Conference on Automated Reasoning) 2012
• Delta-Decidability over the Reals [arXiv]
  Sicun Gao, Jeremy Avigad, and Edmund Clarke
  LICS (Logic in Computer Science) 2012
• Quantifier Elimination over Finite Fields with Groebner Bases [arXiv]
  Sicun Gao, Andre Platzer, and Edmund Clarke
  CAI (International Conference on Algebraic Informatics) 2011
• Integrating ICP and LRA Solvers for Deciding Nonlinear Real Arithmetic [pdf]
  Sicun Gao, Malay Ganai, Franjo Ivancic, Aarti Gupta, Sriram Sankaranarayanan, and Edmund Clarke
  FMCAD (Formal Methods in Computer Aided Design) 2010
• A Non-Prenex DPLL-Based QBF Solver with Game-State Learning [pdf]
  William Klieber, Samir Sapra, Sicun Gao, and Edmund Clarke
  SAT (Theory and Applications of Satisfiability Testing) 2010
• Counting Zeros over Finite Fields with Groebner Bases [pdf]
  MS Thesis in Logic, Carnegie Mellon University 2009

The infinite we shall do right away. The finite may take a little longer. -- Stanislaw Ulam