Lectures


The course schedule will be updated frequently, please keep visit this webpage for updates.

Schedule and Reading List

Date Slides Topic Presenter Administrivia Notes Papers
08/24 Lecture 1 Introduction to dependable AI systems course and logistics Ravi Iyer Assignment 0 released N/A
08/26 Lecture 2 and 3 Video Introduction: Safety, Reliability, and Security-sensitive systems Ravi Iyer Assignment 0 due Optional reading Bagchi, Saurabh, et al. “Vision Paper: Grand Challenges in Resilience: Autonomous System Resilience through Design and Runtime Measures.” IEEE Open Journal of the Computer Society (2020). link A. Avizienis, J. -. Laprie, B. Randell and C. Landwehr, “Basic concepts and taxonomy of dependable and secure computing,” in IEEE Transactions on Dependable and Secure Computing, vol. 1, no. 1, pp. 11-33, Jan.-March 2004, doi: 10.1109/TDSC.2004.2. link
08/31 Lecture 2 and 3 Video Real-world dependability hazards (in safety, reliability and security) Ravi Iyer
09/02 Survey results Video Recap on AI algorithms Ravi Iyer
09/07 Labor day holiday
09/09 Overview, Video Group discussion on Unmanned Aerial Vehicles Students and Ravi Iyer Assignment 1
09/14 Overview Paper 1, Paper 2, Video) Formal methods and Game theory Student presenter Paper 1: Game-Theoretic Methods for Robustness, Security, and Resilience of CPS Control Systems Paper 2: Conformance Testing as Falsification for Cyber-Physical Systems Optional reading Causality-Aided Falsification
09/16 Overview Paper 1, Video) Safety + Group Discussion Student presenter Assignment 2 Paper 1: Autonomous Vehicles Meet the Physical World: RSS, Variability, Uncertainty, and Proving Safety Optional reading Safety force field
09/21 PDF, Video Reinforcement learning Ravi Iyer
09/23 Overview, Video, Paper 1 Resilience + Group discussion Student presenter Paper 1: Measuring the Reliability of Reinforcement Learning Algorithms Discussion reading 1: Deep Reinforcement Learning Doesn’t Work Yet Discussion reading 2:Faulty reward functions in the wild
09/28 PDF, Video Generative adversarial networks Ravi Iyer Optional reading Goodfellow, Ian, et al. “Generative adversarial nets.” Advances in neural information processing systems. 2014. link
09/30 Overview Paper 1, Paper 2, Video Identifying robustness issues Student presenter Paper 1: Tramèr, Florian, et al. “Ensemble adversarial training: Attacks and defenses.” arXiv preprint arXiv:1705.07204 (2017). Paper 2: Adversarial Robustness Toolbox v1.0.0
10/05 PDF, Video Guest Lecture - Prof. Bo Li - password same as class password
10/07 Video Project idea presentation Signup sheet (via Piazza)
10/12 Paper 1, Paper 2, Video Trojan attacks and stealing models Student presenter Paper 1: Trojan attack on Neural Networks (NDSS 2018) link, Paper 2: Neural Cleanse: Identifying and Mitigating Backdoor Attacks in Neural Networks link
10/14 Paper 1, Video Fuzzing Student presenter Paper 1: MTFuzz: Fuzzing with a Multi-Task Neural Network link, Paper 2: TensorFuzz: Debugging Neural Networks with Coverage-Guided Fuzzing link
10/19 Paper 1, Paper 2, Video Integrity checks and monitoring Student presenter Paper 1: Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors (TCAV) link, Paper 2: Model Assertions for Monitoring and improving ML models link
10/21 Video Guest Lecture - Dr. Jonathan Petit Zoom link
10/26 Overview, Paper 1, Paper 2, Video Domain-driven verfication & vaildation Student presenter Paper 1: Formal Scenario-Based Testing of Autonomous Vehicles: From Simulation to the Real Worldlink, Paper 2: Fast and effective robustness certification link
10/28 Video Guest Lecture - Prof. Sanjit Seshia
11/02 Paper 1, Paper 2, Video Certification & provable Defense Student presenter Paper 1: Cnn-cert: An efficient framework for certifying robustness of convolutional neural networks link, Paper 2: DRYVR:Data-driven verification and compositional reasoning for automotive systemslink
11/04 Video Guest Lecture - Prof. Chuchu Fan
11/09 To be posted Misc. topics in ML/AI Ravi Iyer and Saurabh Jha
11/11 Video Guest Lecture - Prof. Philip Koopman
11/16 Midterm project presentation
11/18 Health & AI Student presenter Paper 1: Feature Robustness in Non-stationary Health Records: Caveats to Deployable Model Performance in Common Clinical Machine Learning Tasks link, Paper 2: Injured Brains and Adaptive Networks: The Benefits and Costs of Hyperconnectivity link
11/23 FALL BREAK
11/25 FALL BREAK
11/30 Final project presentation
12/02 Final project presentation
12/07 Summary and path forward