Sumanth Dathathri
Research Scientist, DeepMind
Email: sdathath at caltech dot edu
Research
- I work on designing reliable efficient autonomous systems.
- If any of my works are behind a paywall, and you would like access please email me for a private copy.
Note
-
This website is underconstruction and incomplete. If you notice any errors, please feel free to email me about the same.
Preprints
- Scalable Neural Learning for Verifiable Consistency with Temporal Specifications
Sumanth Dathathri, Johannes Welbl, Krishnamurthy (Dj) Dvijotham, Ramana Kumar, Aditya Kanade, Jonathan Uesato, Sven Gowal, Po-Sen Huang and Pushmeet Kohli [Preprint]
Recent Papers in Peer-Reviewed Conferences (Since 2017)
- Enabling certification of verification-agnostic networks via memory-efficient semidefinite programming
Sumanth Dathathri**, Krishnamurthy Dvijotham**,
Alexey Kurakin**, Aditi Raghunathan**,
Jonathan Uesato**,
Rudy Bunel, Shreya Shankar, Jacob Steinhardt, Ian Goodfellow, Percy Liang and Pushmeet Kohli
In Neural Information Processing Systems (NeurIPS) 2020
- Plug-and-Play Conversational Models
Andrea Madotto*, Etsuko Ishii*, Zhaojiang Lin*, Sumanth Dathathri* and Pascale Fung
In Conference on Empirical Methods in Natural Language Processing (EMNLP) - Findings Track 2020
- Plug and Play Language Models: A Simple Baseline for Controlled Language Generation[Paper][Code][Press MIT Techreview]
Sumanth Dathathri, Andrea Madotto, Janice Lan, Jane Hung, Eric Frank, Piero Molino, Jason Yosinski and Rosanne Liu
In International Conference on Learning Representations (ICLR) 2020
- Counter-example guided learning of bounds on environment behavior [Paper][Code]
Yuxiao Chen, Sumanth Dathathri, Tung M. Phan and Richard M. Murray
In Conference on Robot Learning (CoRL) 2019
- Inverse Abstraction of Neural Networks Using Symbolic Interpolation
Sumanth Dathathri, Sicun Gao and Richard M. Murray
In Thirty-Third AAAI Conference on Aritificial Intelligence (AAAI) 2019
- Parallelizing Synthesis from Temporal Logic Specifications by
Identifying Equicontrollable States[Paper]
Sumanth Dathathri, Ioannis Filippidis and Richard M. Murray
In International Symposium on Robotics Research (ISRR) 2017
- Decomposing GR(1) Games with Singleton Liveness Guarantees for Efficient Synthesis
Sumanth Dathathri and Richard M. Murray
In Conference on Decision and Control (CDC) 2017
- Learning-Based Abstractions for Nonlinear Constraint Solving
Sumanth Dathathri, Nikos Arechiga, Sicun Gao and Richard M. Murray
In Int'l Joint Conference on Artificial Intelligence (IJCAI) 2017
- Enhancing tolerance to unexpected jumps in GR(1) games
Sumanth Dathathri, Scott C. Livingston and Richard M. Murray
In Int'l Conference on Cyberphysical Systems (ICCPS) 2017
- Control design for hybrid systems with TuLiP: The temporal logic planning toolbox
Ioannis Filippidis, Sumanth Dathathri, Scott C. Livingston, Necmiye Ozay and Richard M. Murray
In IEEE Conference on Control Applications (CCA) 2016
- Interfacing TuLiP with the JPL Statechart Autocoder:
Initial progress toward synthesis of flight software from formal specifications
Sumanth Dathathri, Scott C. Livingston, Leonard J. Reder, and Richard M. Murray
In IEEE Aerospace Conference 2016
**Equal Contributon, Alphabetical Ordering
*Equal Contributon
Background
- Ph.D. - California Institute of Technology
- B.Tech, M.Tech - Indian Institute of Technology Madras
Break the cycle Morty, rise above, focus on science. -- Rick Sanchez