Query Languages for Neural Networks. (R-12216)

Crucial elements in this methodology are logical data models and declarative query languages. In the new field of Data Science, large volumes of data are analyzed by machine learning algorithms that produce predictive models such as neural networks. Over the course of time, large amounts of neural network and training data are produced in a data science enterprise. For reasons of transparency, accountability, and plain efficiency, these data need to be managed in a structured manner, just like the data in a classical database system. Thus, the project proposes to develop new logical data models and declarative query languages for Data Science, with a focus on neural networks. We will base our research on past experience in graph data management, type systems, reflection, and automated verification.

Keywords:Data management, Neural networks, Query languages

Disciplines:Neural, evolutionary and fuzzy computation, Data models, Database theory, Language design, constructs and features, Computational logic and formal languages