Ismini Lourentzou awarded NSF grant to develop infrastructure for more effective AI in U.S. manufacturing industry

Because artificial intelligence benefits from training on large datasets, trying to implement AI within the U.S. manufacturing industry poses a critical problem, according to Ismini Lourentzou, assistant professor in the Department of Computer Science and faculty at the Sanghani Center for Artificial Intelligence and Data Analytics. “Manufacturers not only tend to be slow and repetitive with data collection efforts, but they typically keep their data secret and partnerships are rare,” she said.

Lourentzou was recently awarded an EArly-concept Grant for Exploratory Research (EAGER) from the National Science Foundation for a project, Cost-sensitive Federated AI for Smart Manufacturing Data-Sharing, to develop a manufacturing service infrastructure that would encourage U.S. manufacturers to accelerate the use of AI in smart manufacturing and exchange data with trusted partners.  

Ran Jin, associate professor in the Grado Department of Industrial and Systems Engineering is serving as co-principal investigator for the project.

The proposed cost-sensitive data-sharing framework can assess and differentiate the contributions from multiple manufacturing data owners via a learned hierarchical task-driven similarity that decomposes the underlying retrieval scoring mechanism into two interconnected elements, manufacturer similarity and data similarity. 

“It can be used by manufacturers who wish to improve AI model training, testing, and deployment within their organizations or find potential collaborators and partners for new product development,” Lourentzou said. “Long-term, we hope that establishing a data-sharing market will enhance the United States’ international market share.”