Congratulations to Profesor Chelsea Finn on being selected as AI2050 Early Career Fellow. The AI2050 Early Career Fellows are a part of the Artificial Intelligence and Advanced Computing Institute at Schmidt Sciences, which supports fundamental research in AI, the application of AI and data science across a wide variety of disciplines, and the creation of high-impact research platforms that can speed discovery.

The AI2050 Early Career Fellowship is an enabling fellowship to encourage postdoctoral and pre-tenure researchers to pursue bold and ambitious work on hard problems in AI, which often is multidisciplinary and typically hard to fund. Each year, approximately 10-20 early career researchers around the world are selected through a rigorous process. Early Career Fellows receive an award to support a two year project that advances their research and receive non-monetary support, such as connections to stakeholders to help them advance their research and amplify its results.

Chelsea’s AI2050 Project
Machine learning models are known to be brittle when deployed in the real world, when correlations that exist in the training data do not hold in new scenarios. In contrast, humans can learn concepts efficiently and robustly by having a teacher correct their mistakes. Chelsea Finn hypothesizes that machine learning models can similarly learn robust functions by combining the training data with corrective information that targets the model’s underlying “misconceptions.” This project’s goal is to develop a new framework that seeks out small amounts of targeted information about a model’s misconceptions and uses that interaction to robustify the model.

Please join us in congratulating Chelsea on her well-deserved recognition!

Schmidt Sciences Press Release, 'Second Cohort of AI2050 Early Career Fellows Named By Schmidt Sciences,’ March 4, 2024.