American Family Insurance Data Science Institute awards “mini grants” to advance data science

MADISON — Nine teams of University of Wisconsin–Madison faculty and collaborators have been awarded nearly $1 million through the American Family Funding Initiative, a research competition for data science projects.

“The American Family Insurance Data Science Institute is honored to announce the results of the first round of the American Family Funding Initiative,” says Brian Yandell, interim director of AFIDSI, which brings together faculty from across campus to launch new research in data science and apply findings to solve problems. “The nine successful proposals were chosen from a strong pool of 37 applicants.”

The emerging field of data science is the study, development or application of methods that reveal new insights from data. The successful projects will further research ranging from third-wave artificial intelligence to student entrepreneurship.

American Family Insurance has partnered with UW–Madison through the Institute to offer “mini grants” of $75,000-to-150,000 per year for data science research at UW–Madison.

“This is the initial installation of a $10 million research agreement, promising many exciting research collaborations over the coming decade,” adds Yandell.

Since opening in July 2019, AFIDSI has worked to build on UW–Madison data science initiatives. In collaboration with researchers across campus and beyond, AFIDSI focuses on the fundamentals of data science research and on translating that research into practice.

The goal of the American Family Funding Initiative is to stimulate and support highly innovative research. The nine successful projects, reviewed by faculty and staff from across campus, were evaluated based on their potential contributions to the field of data science, practical use and the novelty of their approaches.

“ We expect the American Family Funding Initiative will enable UW–Madison faculty to advance data science research and position our scientists to be more competitive when applying for extramural research funding,” Yandell says.

Applications for a second round of awards are due July 8. More information about this initiative is available at datascience.wisc.edu.

The first round of American Family Funding Initiative projects include:

Question Asking with Differing Knowledge and Goals

Principal investigator: Joe Austerweil, Psychology

Despite tremendous progress in machine learning, automated answers to questions are still inferior to answers from humans. This project investigates whether incorporating psycholinguistic factors that influence how people respond to language can improve automated question-answering methods.

Using Data to Foster Entrepreneurship and Innovation in the Madison Ecosystem

Principal investigator: Jon Eckhardt, Wisconsin School of Business. Collaborators: Brent Goldfarb, University of Maryland; Molly Carnes, UW–Madison Women in Science and Engineering Leadership Institute

Student entrepreneurship is an important path for upward mobility and wealth creation, but little is known about what drives students’ interest in activities such as starting a business. Insights from this research will support the creation of evidence-based support for student entrepreneurs.

Machine Learning for Usage-Based Insurance

Principal investigator: Robert Holz, Space Science Engineering Center. Co-PI: Willem Marais, Space Science Engineering Center. Collaborator: Rebecca Willett, University of Chicago

Usage-based insurance is a type of vehicle insurance that promotes safer driving behavior, reduces accidents and helps lower costs. This project investigates machine-learning methods that analyze very large UBI datasets in order to produce a measure of driver risk and safety.

Optimizing Question and Answer Systems via User Feedback

Principal investigator: Robert Nowak, Wisconsin Institute for Discovery, College of Engineering

Question-and-answer online software systems are increasingly common throughout business, industry and healthcare. This project aims to develop new theory and methods for optimizing Q&A systems, based on user feedback.

Improving Traffic Safety Outcomes Through Data Science

Principal investigator: David Noyce, College of Engineering

While advances during the last 40 years have led to significant improvements in transportation safety, recent trends have shown a leveling—and in some cases an increase—in the number of traffic crash fatalities. The vision for this research is to translate advances in data science into tools that will improve driver safety, bolster the safety performance of automotive technologies and move the trend towards zero fatalities.

Learning Causal Relationships from Data

Principal investigator: Irene Ong, School of Medicine and Public Health. Co-PI: Aubrey Barnard, Biostatistics and Medical Informatics

While humans naturally develop an understanding of cause and effect by exploring the world, causal understanding is often missing from artificially intelligent systems, as you may have noticed when your digital assistant goes awry. This research will develop an algorithm to improve the causal reasoning abilities of such systems, and apply the algorithm in healthcare settings to improve patient outcomes.

3D Capture and Scanning Technology for Insurance Documentation

Principal investigator: Kevin Ponto, School of Human Ecology

Insurance claims adjusters constantly face the challenge of inspecting and assessing a scene to understand potential risk, or what took place after an event. This project will design and implement a system that utilizes 3D scanning and capture technology to reduce disputes between insurance companies and their clients, saving money and time for both parties.

Lightweight Natural Language and Vision Algorithms for Data Analysis

Principal investigator: Vikas Singh, School of Medicine and Public Health. Collaborators: Zhanpeng Zeng, Computer Sciences; Shailesh Acharya and Glenn Fung, American Family Insurance

Natural language processing is a form of artificial intelligence that helps computers read and understand human language. The overarching goal of this project is to accelerate the time it takes to train and test efficient, accurate natural language processing models.

Ultra-Fast Training for the Third Wave of Artificial Intelligence: Novel Categories in Text Classification

Principal investigator: Jerry Zhu, Computer Sciences. Co-PI: Ara Vartanian, Computer Sciences

Second-wave artificial intelligence networks are trained on enormous data sets labeled to recognize patterns. This project aims to take a step toward the yet-to-come third wave of AI systems that will require far fewer data items for training, and will apply rules in ways that are more similar to human cognition.