Roni Stern, PhD
Developing a software program without bugs is probably impossible. In this talk I will show how Artificial Intelligence (AI) techniques can be integrated in software development and maintenance processes to reduce the number of observed bugs and the costs of handling them. The AI we developed is able to predict where bugs will occur before hand by mining data from industry-standard software engineering tools. If a bug does occur, then our AI is able to consider the observed system behavior and automatically identify its root cause of bugs, i.e., the faulty software component that cause it. Lastly, our AI is also able to automatically plan additional tests for software testers to perform when they find a bug to gather more diagnostic information. All this is done with a combination of various techniques from the AI literature, including, Machine Learning, Automated Planning, and Model-Based Diagnosis. If time permits, I will also talk about the applications of this research to detecting vulnerabilities.
This talk will cover topics from papers published in AAAI and DX conferences, and contains joint work with Dr. Meir Kalech, Tom Zamir, and Amir Elmishali.
UC6YMneXOTYhaLOukqqQ8CqQ
source
