Paper on Optimization Autoformalism that uses large language models to craft optimization solutions for decision-making
Paper on certified robustness for neural ODE accepted in IEEE Control Systems Letters (L-CSS)
Paper on derivative-free meta blackbox (nonconvex) optimization on manifold at L4DC 2023 (oral presentation)
One paper adversarial ML (sifting out clean data from poisoned data) at USENIX Security 2023
Three papers on meta-safe reinforcement learning (spotlight), model-agnostic data valuation (spotlight), and adversarial ML (certified robustness against UAP/backdoors) at ICLR 2023
Three papers at AAAI on winning the CityLearn Challenge (oral), approximation/statistical properties of solution functions (oral), and nonstationary risk-sensitive RL (oral)
Paper on dynamic regret for online optimization "Dynamic Regret Bounds for Online Nonconvex Optimization" to appear in IEEE Transactions on Control of Network Systems
Paper on general bi-level optimization "Iterative Implicit Gradients for Nonconvex Optimization with Variational Inequality Constraints"
Thanks The Commonwealth Cyber Initiative (CCI) for supporting our research
Presentation at PMS 406 Autonomy MRE workshop on assured RL for dynamical systems
Paper on learning under specifications "Learning Neural Networks under Input-Output Specifications" (arXiv) to appear in ACC 2022
Paper on adversarial ML "Adversarial Unlearning of Backdoors via Implicit Hypergradient" (arXiv) accepted to ICLR 2022
Paper on safe RL "Recurrent Neural Network Controllers Synthesis with Stability Guarantees for Partially Observed Systems" (arXiv) to appear in AAAI 2022
Paper on control by proxy "Controlling Smart Inverters using Proxies: A Chance-Constrained DNN-based Approach" (arXiv) to appear in IEEE Transactions on Smart Grid
Two teams that I led (ROLEVT & ZoRL) jointly won the 1st place in the CityLearn Challenge 2021. Congrats to team members: Vanshaj Khattar, Qasim Wani, Zhiyao Chang, and Mingyu Kim
New paper on adversarial ML "Adversarial Unlearning of Backdoors via Implicit Hypergradient" (arXiv)
Talk on assured RL for energy systems at C3.ai Digital Transformation Institute (video)
My group will give an oral presentation on implicit RL in SECC 2021
New paper on implicit RL "Zeroth-Order Implicit Reinforcement Learning for Sequential Decision Making in Distributed Control Systems"
New paper on assured learning of neural networks "Learning Neural Networks under Input-Output Specifications"
New paper on dynamic regret for online optimization "Dynamic Regret Bounds for Online Nonconvex Optimization"
New paper on safe reinforcement learning "Recurrent Neural Network Controllers Synthesis with Stability Guarantees for Partially Observed Systems" (arXiv)
I will serve as an Associate Editor for IEEE Systems Journal
New paper on data quality management "A Unified Framework for Task-Driven Data Quality Management" (arXiv)
I received the SCEEE Research Initiation Grant to study safe and resilient AI for networked dynamical systems
New paper on control by proxy "Controlling Smart Inverters using Proxies: A Chance-Constrained DNN-based Approach" (arXiv)
Paper on safe imitation learning "Imitation Learning with Stability and Safety Guarantees" (arXiv) to appear in IEEE Control Systems Letters
Paper on online nonconvex optimization "Diminishing Regret for Online Nonconvex Optimization" to appear in 2021 American Control Conference
I will teach ECE5984 Special topic: Optimization Theory for ML (spring, 2021)
New paper on safe imitation learning "Imitation Learning with Stability and Safety Guarantees" (arXiv)
Paper on power grid cybersecurity "Boundary Defense Against Cyber Threat for Power System State Estimation" (pdf) to appear in IEEE Transactions on Information Forensics & Security
Paper on reinforcement learning "Stability-certified reinforcement learning: A control-theoretic perspective" to appear in IEEE Access
Paper on robust graph convolutional network "Power up! Robust Graph Convolutional Network via Graph Powering" accepted to AAAI 2021
See here a 5-minutes introduction to my research presented at the Engineering Faculty organization (EFO) faculty meeting, Virginia Tech
Paper on reinforcement learning for building control to appear in ACM International Workshop on Reinforcement Learning for Energy Management (RLEM'20) "Towards Off-policy Evaluation as a Prerequisite for Real-world Reinforcement Learning in Building Control"
Invited presentation at the 2020 CPES & PEC conference on "Adversarially robust learning and control for power grid"
Paper on deep learning for smart inverter control to appear in IEEE SmartGridComm'20 "Deep Learning for Reactive Power Control of Smart Inverters under Communication Constraints"
I will teach ECE4424/CS4824: Machine learning (fall, 2020)
Invited presentation at the IEEE Conference on Control Technology and Applications (CCTA) on "Adversarial machine learning for energy systems"
Paper on reinforcement learning for thermodynamic systems to appear in Applied Energy "Control of superheat of organic Rankine cycle under transient heat source based on deep reinforcement learning"
Thanks NSF for supporting our proposal Machine Learning for Communication-Cognizant Smart Inverter Control with Prof. Vassilis Kekatos
Conversation with Berkeley IEOR (interview article)
The paper "Conic Relaxations of Power System Optimization: Theory and Algorithms" to appear in European Journal of Operational Research
New paper on cybersecurity for power grid "Boundary Defense against Cyber Threat for Power System Operation" (Supplementary material)
Paper to appear in Proc. 58th IEEE Conference on Decision and Control "Towards Robust and Scalable Power System State Estimation"
New paper on robust graph convolutional network "Power up! Robust Graph Convolutional Network against Evasion Attacks based on Graph Powering"
New paper on power system state estimation "Scalable and robust state estimation from abundant but untrusted data"
New paper on stability certfied reinforcement learning "Stability-certified reinforcement learning: a control-theoretic perspective"
Presentations at the International Conference on Applied Energy: "BISCUIT: Building Intelligent System Customer Investment Tools" (paper | slides) and "Advanced Building Control via Deep Reinforcement Learning" (paper)
Presentation at the 23rd International Symposium on Mathematical Programming, Bordeaux, France: "Vulnerability analysis and robustification of power grid state estimation" (slides)
New paper on power system vulnerability analysis "Power Grid AC-based State Estimation: Vulnerability Analysis Against Cyber Attacks" accepted in IEEE Transactions on Automatic Control
New paper to guarantee safety of smooth RL in real-world systems like power grids: "Control-theoretic analysis of smoothness for stability-certified reinforcement learning"
New paper on observer design for large-scale nonlinear systems: "Multiplier-based observer design for large-scale Lipschitz systems"
We are organizing an international workshop on "AI for energy-cyber-physical systems" as part of the International Conference on Applied Energy (ICAE 2018). See here the CFP
I have finished my PhD thesis: Data-efficient Analytics for Optimal Human-Cyber-Physical Systems
Centerline has reported our IEQ Bot in its November issue: "CBE Panel Session Explores Innovative Methods for Monitoring Indoor Environments"
Paper on our new IEQ Bot "Automated mobile sensing: Towards high-granularity agile indoor environmental quality monitoring" accepted by Building and Environment (paper | slide | code)
INFORMS invited talk on "Robustness Analysis of Power Grid under False Data Attacks Against AC State Estimation" at Houston, Texas. Panelist in the Industrial Advisory Board meeting at the Center of the Built Environment, "IEQ Bot", Berkeley, California
New paper on power system vulnerability analysis "Power Grid AC-based State Estimation: Vulnerability Analysis Against Cyber Attacks"
Berkeley Engineer magzine featured an article "Brains for buildings, packaged in a smart briefcase" on our Building-in-Briefcase sensor platform
Three papers "Distributed energy resource integration by dispatch and retail optimization", "WinIPS: An WiFi-based non-intrusive indoor positioning system enabling online radio map construction" and "Review of microgrid development in the United States and China and lessons learned for China" accepted in IEEE PES ISGT-Asia (2017), IEEE Transactions on Wireless Communications and Applied Energy Symposium and Forum (REM2017)
Two papers "A semidefinite programming relaxation under false data injection attacks against power grid AC state estimation" and "Indoor environmental quality monitoring by autonomous mobile sensing" accepted in Allerton (2017) and ACM BuildSys (2017)
Oral presentation of paper Inverse reinforcement learning via deep Gaussian Process at UAI 2017 (slides and a blog from Daniel Seita), Sydney, Australia
IEEE Spectrum featured an article What Does Your Smart Meter Know About You? on my recent work Virtual occupancy sensing: Using smart meters to indicate your presence
CO2Meter.com featured an article CO2 Sensor Occupancy Detection on my work Sensing by proxy: Occupancy detection based on indoor CO2 concentration (best paper award)
Our proposal for "Social Game for Smart Building Energy Efficiency" accepted (Total: $14,000, duration 1 year, with Ioannis Konstantakopoulos)
Three papers "Microgrid to enable optimal distributed energy retail and end-user demand response", "A Robust Utility Learning Framework via Inverse Optimization", and "Measuring fine-grained Metro interchange time via smartphones" accepted in Applied Energy, TCST, and Transportation Research Part C
Paper Virtual occupancy sensing: Using smart meters to indicate your presence accepted in TMC
My proposal for Energy Utopia is awarded with the Student Technology Fund of $3000
We have organized the first international workshop on smart building collocated with IEEE SmartGridComm in November, Sydney! (slides for keynotes are now available HERE).
I'm teaching the innovation course at Berkeley, EE 16B: Designing Information Devices and Systems II as a Graduate Student Instructor (GSI), together with Prof. Anant Sahai and Prof. Michel Maharbiz
Our paper "SoundLoc: Accurate room-level indoor localization using acoustic signatures"(joint work with Ruoxi Jia and Prof. Costas Spanos) has been featured in MIT Technology Review: An Indoor Positioning System Based On Echolocation