TorchKM: A GPU-Oriented Library for Kernel Learning and Model Selection
Researchers have developed TorchKM, an open-source library designed to accelerate kernel machine learning tasks on GPUs. The library offers a scikit-learn-compatible API and leverages GPU-friendly linear algebra to speed up training and model selection. Benchmarks indicate that TorchKM achieves competitive predictive performance while offering significant speedups compared to existing methods. AI
IMPACT Accelerates kernel machine learning workflows, potentially enabling faster experimentation and deployment of models like SVMs.