GPU-Accelerated Clustering for Financial Instruments at Scale
NVIDIA, Friday, August 21st, 2026
NVIDIA presents AdaptGrow, a GPU-accelerated matrix factorization method for clustering financial instruments.
NVIDIA describes AdaptGrow, a GPU-accelerated matrix factorization algorithm for clustering financial instruments at scale. It turns rolling correlation and tail-dependence matrices into both hard clusters and soft factor loadings.
That dual output means practitioners get discrete groupings and continuous exposures from the same computation. The method also supports efficient detection of structural breaks in the underlying relationships. Authors are Dhruv Desai, Lavinia Ghita and Roman Enzmann.