EAGER: Mapping Future Synergies between Deep Learning and Software Engineering

点击翻译按钮获取中文摘要

The power of AI, more specifically, machine learning (ML), is enabling new classes of software applications. Software Engineering (SE) goals such as construction of correct software, and devising effective software development processes for testing and debugging, require new techniques when ML is embedded in a system, in order to test learning models, validate training data, and test for valid results. Researchers have begun working toward improving the quality and productivity of software for ML-learning-based applications. Conversely, SE researchers are using ML to accelerate advances in the SE research in order to achieve better results. This EAGER grant is for the purpose of assembling a research road map in the intersection of ML and SE, and to make recommendations for those doing research in these areas.The project will focus mainly on a specific mode of learning called Deep Learning (DL). The agenda is bidirectional: use of DL to accelerate research in SE, and SE to address correctness and quality of DL-based applications. The project will survey types of SE tasks where DL has achieved or promised significant advances, and it will identify areas where DL-based applications have been studied and improved by SE. The project will make available the current state of knowledge and best practices for data gathering, representation and processing for this research area and explore common infrastructure, architectures, data resources and benchmarks. The project will construct a systematic literature review that will become a living/growing repository of work in the SE/DL intersection for future research and education purposes.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.