The National Science Foundation awarded a $666,000 Project Grant to The Trustees of Columbia University in the City of New York (Columbia University) through the Computer and Information Science and Engineering Program (CFDA #47.070). The objective is to improve the performance, robustness, generalizability, and efficiency of deep learning models for software assurance tasks such as bug detection, debugging, test input generation, and test suite prioritization. The research focuses on encoding program semantics into the program representation by combining program analysis, software engineering, and deep learning expertise. Key research thrusts include learning with abstract semantics by combining static analysis with deep learning, learning with concrete semantics using program execution traces, and identifying and discouraging the use of spurious features in the models. The research results, datasets, and tools will be disseminated to the research community, and workshops will be organized to strengthen the community of deep learning for code.
Mod # | Description | Reason For Modification | Federal Obligation (Click to sort descending) | Date (Click to sort ascending) |
|---|---|---|---|---|
| Not listed | $666.0k | 6/23/23 |