From the source
Google Research introduces GIST, a novel algorithm for data subset selection that balances diversity and utility with provable guarantees, presented at NeurIPS 2025.
From the source
At NeurIPS 2025, we introduced Greedy Independent Set Thresholding (GIST), a novel algorithm that helps solve this issue by balancing data “diversity” (ensuring the selected data is not redundant) and data “utility“ (data that is relevant and useful for the task).
research.google