Download Ravi Mangal and more Lecture notes Programming Languages in PDF only on Docsity!
Ravi Mangal
Research Interests
Trustworthy Machine Learning, Program Verification, Formal Methods
Education Georgia Institute of Technology, Atlanta, Georgia USA Ph.D., Computer Science, Dec, 2020 Advisor: Dr. Alessandro Orso
Georgia Institute of Technology, Atlanta, Georgia USA M.S., Computer Science, May, 2012
Veermata Jijabai Technological Institute, Mumbai, India B.Tech., Information Technology, May, 2010
Work Experience
Cylab, Carnegie Mellon University, Silicon Valley, California USA Postdoctoral Researcher with Dr. Corina P˘as˘areanu Jan, 2021 - Present
- Drove research on techniques for improving the local robustness of neural networks via training and run-time certification as part of the DARPA Guaranteeing AI Robustness Against Deception (GARD) program.
- Designed algorithms for repairing neural networks at run-time in order to ensure compliance with user-provided safety specifications.
- Developed techniques for quantifying the uncertainty of ML models for use in safety analysis of discrete-event controllers that use neural networks for perception in collaboration with researchers at University of York as part of the Assured Autonomy International Program.
- Initiated collaboration with researchers at VMware Research to develop techniques for verifiable personalization of ML models in the context of federated learning.
- Authored 6 research papers, raised $60K grant money, and mentored PhD and Masters students.
Georgia Institute of Technology, Atlanta, Georgia USA Graduate Research Assistant Jan, 2012 - Dec, 2020
- Developed new theoretical frameworks for constructing scalable, precise static program analyses.
- Designed a new approach for interactive, user-guided static program analyses by combining formal methods with probabilistic techniques.
- Developed algorithms for analyzing robustness properties of neural networks.
- Published 10 research papers in top academic conferences including AAAI, ESEC/FSE, ESOP, ICSE, OOPSLA, PLDI, POPL, and SAT.
Microsoft Research, Redmond, Washington USA Research Intern May, 2016 - Aug, 2016 Developed a tool to help pen-testers perform security analysis of Android apps using a new algorithm for probabilistic and interactive information-flow analysis of programs with Dr. Patrice Godefroid and Marina Polishchuk.
Google, Mountain View, California USA Research Intern May, 2014 - Aug, 2014 Contributed in the design and development of an industry-strength static program analysis framework for analyzing security properties of Android apps with Dr. Jayanthkumar Kannan and Dr. Domagoj Babic.
Nvidia, Santa Clara, California USA Software Intern May, 2011 - Aug, 2011 Built a software simulator of DisplayPort devices for stress testing GPU device drivers as a member of the GPU Resource Manager team.
Microsoft, Hyderabad, India Software Development Engineer in Test Intern May, 2009 - Jul, 2009 Built a test status dashboard that featured real-time updates from multiple sources of software testing data for the team developing the Data Protection Manager product.
Indian Institute of Technology-Bombay, Mumbai, India Undergraduate Researcher May, 2008 - Jul, 2008 Worked on automated speech recognition algorithms in the Digital Audio Processing lab with Dr. Preeti Rao.
Research Articles
(* indicates equal contribution, (α) indicates alphabetical ordering) Preprints (α) Radu Calinescu, Calum Imrie, Ravi Mangal, Corina P˘as˘areanu, Misael Alpizar Santana, and Gricel V´azquez. Discrete-event controller synthesis for autonomous systems with deep-learning per- ception components. arXiv preprint arXiv:2202.03360, 2022
Conference Publications Ravi Mangal, Zifan Wang, Chi Zhang*, Klas Leino, Corina P˘as˘areanu, and Matt Fredrikson. On the perils of cascading robust classifiers. In International Conference on Learning Representations, ICLR ’23, 2023
(α) Divya Gopinath, Luca Lungeanu, Ravi Mangal, Corina P˘as˘areanu, Siqi Xie, and Huafeng Yu. A cascade of checkers for run-time certification of local robustness. In Fundamental Approaches to Software Engineering, FASE’23. Springer, 2023
Klas Leino, Chi Zhang, Ravi Mangal*, Matt Fredrikson, Bryan Parno, and Corina P˘as˘areanu. Degradation attacks on certifiably robust neural networks. Transactions on Machine Learning Re- search, 2022
Ravi Mangal, Kartik Sarangmath, Aditya V. Nori, and Alessandro Orso. Probabilistic lipschitz analysis of neural networks. In International Static Analysis Symposium, SAS ’20. Springer, 2020
Ravi Mangal, Aditya V. Nori, and Alessandro Orso. Robustness of neural networks: A probabilistic and practical approach. In Proceedings of the 41st International Conference on Software Engineering: New Ideas and Emerging Results, ICSE-NIER ’19, 2019
Sulekha Kulkarni, Ravi Mangal, Xin Zhang, and Mayur Naik. Accelerating program analyses by cross-program training. In Proceedings of the 2016 ACM SIGPLAN International Conference on Object-Oriented Programming, Systems, Languages, and Applications, OOPSLA ’16, 2016
Ravi Mangal, Xin Zhang, Aditya Kamath, Aditya V. Nori, and Mayur Naik. Scaling relational inference using proofs and refutations. In Thirtieth AAAI Conference on Artificial Intelligence, AAAI ’16, 2016
Xin Zhang, Ravi Mangal, Aditya V. Nori, and Mayur Naik. Query-guided maximum satisfiability. In Proceedings of the 43rd Annual ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages, POPL ’16, 2016
Ravi Mangal, Xin Zhang, Aditya V. Nori, and Mayur Naik. Volt: A lazy grounding framework for solving very large maxsat instances. In International Conference on Theory and Applications of Satisfiability Testing, SAT ’15, 2015
