Anqi “Angie” Liu, an assistant professor of computer science, has been named a recipient of an Amazon Research Award in the Amazon Web Services AI category for her proposal, “(Multi-)Calibrated Active Learning Under Subpopulation Shift.”
Amazon Research Awards provide unrestricted funds to academic researchers investigating various research topics in multiple disciplines. Awardees additionally have access to more than 300 Amazon public datasets and AWS AI/ML services and tools. AWS AI specifically aims to advance machine learning research by funding the development of open-source tools and research that benefit the ML community at large.
This award will support Liu’s work on the responsible deployment of AI technologies. Using AWS cloud computing resources, Liu plans to develop a set of methods for active data collection and calibrated learning under subpopulation shift, or when the demographic information in an AI model’s training data is different from its target. This is especially important in socially critical applications, as a model can be overconfident in incorrect predictions for underrepresented subpopulations due to service fragmentation and lack of access.