Researcher interests: Creating complex software is difficult, and so people have started using machine learning (ML) to generate software instead of writing it manually. The problem: ML-produced software is opaque to humans, and it is difficult to certify that it always behave
FACULTY / SCHOOL: School of Computer Science and Engineering
DEPARTMENT: Computer Science
Selected Publications
on the fly construction of composite events in scenario based modeling using constraint solvers (2019)|
arXiv preprint arXiv:1909.00408|
Read more
the marabou framework for verification and analysis of deep neural networks (2019)|
International Conference on Computer Aided Verification|
Read more
an abstraction based framework for neural network verification (2019)|
arXiv preprint arXiv:1910.14574|
Read more
simplifying neural networks with the marabou verification engine (2019)|
arXiv preprint arXiv:1910.12396|
Read more
on the fly construction of composite events in scenario based modeling using constraint solvers (2019)|
Proceedings of the 7th International Conference on Model-Driven Engineering and Software Development|
Read more
verifying deep rl driven systems (2019)|
Proceedings of the 2019 Workshop on Network Meets AI & ML - NetAI'19|
Read more
deepsafe a data driven approach for assessing robustness of neural networks (2018)|
International Symposium on Automated Technology for Verification and Analysis|
Read more
Efficient Distributed Execution of Multi-component Scenario-Based Models (2018)|
Communications in Computer and Information Science|
Read more
ground truth adversarial examples (2018)|
arXiv: Learning|
Read more
wise computing toward endowing system development with proactive wisdom (2018)|
IEEE Computer|
Read more
toward scalable verification for safety critical deep networks (2018)|
arXiv preprint arXiv:1801.05950|
Read more
Finally, a peek inside the ‘black box’ of machine learning systems (2017)|
Computation & Data, Technology & Society|
Read more
