Award from National Science Foundation and Amazon to Improve AI Fairness

My T. Thai
My T. Thai, Associate Director of the Warren B. Nelms Institute and UF Foundation Research Professor, Department of Computer and Information Science and Engineering (CISE)  in collaboration with professors Hanghang Tong from the University of Illinois at Urbana-Champaign and Ross Maciejewski from Arizona State University, recently received a three-year award for over $1 million from the National Science Foundation and Amazon. The award is a part of their joint Fairness in Artificial Intelligence (FAI) program, which supports computational research focused on fairness in AI, with the goal of contributing to trustworthy AI systems that can help tackle society’s biggest challenges.

In this project, the team seeks a paradigm shift, from the answers to “what and who” types of questions in network learning to the answers of “how and why”, which improve the explanability, transparency, and fairness of network learning via AI. Together on this multi-disciplinary team where Thai is on the theory of explainable AI and fairness, Tong on data mining, and Maciejewski on visual system, the project will develop computational theories, algorithms, and prototype systems in the context of network learning, forming three key pillars – interpretation, auditing, and de-biasing – of fair machine learning for network data. Underpinning these pillars is a human-in-the-loop visual analytics framework to support users in identifying and mitigating bias in the learning process.

“I am excited about this opportunity to work on this research direction,” Thai said. “Machine learning models have been using as a blackbox without a clear explanation, which is hinder our understanding on the accuracy and fairness of the models’ decision, especially in network data which are no longer independent to each other. Thus it is time for us to take a deeper look to make network learning explainable, transparent, and fair.”

The broader impacts of this research include improvements of social network analysis, neural science, team science and management, intelligent transportation systems, critical infrastructures, and blockchain networks.

Partnerships between private companies and public entities like the NSF give researchers a better picture of the big questions being asked by industry that require basic research to answer and help accelerate the translation of research from the lab to commercialization, the NSF says.