mloss.org GPUML GPUs for kernel machineshttp://mloss.orgUpdates and additions to GPUML GPUs for kernel machinesenFri, 26 Feb 2010 18:12:46 -0000GPUML GPUs for kernel machines 4http://mloss.org/software/view/236/<html><p>Kernel machines are integral part of many learning approaches. We have identified 3 key computations encountered in these algorithms.
1. Weighted summation of kernel function (ex. kernel density estimation)
2. Kernel matrix-vector product in an iterative algorithm (ex. kernel regression)
3. Operations on the kernel matrix (ex. kernel PCA)
</p>
<p>In this software, we accelerate each of these on a GPU. See the documentation for more details, and demo to see the speedups.
</p>
<p>System requirement: Cuda 2.2+, Visual studio 2008 (if used in Windows)
</p></html>Balaji Vasan Srinivasan, Qi Hu, Ramani DuraiswamiFri, 26 Feb 2010 18:12:46 -0000http://mloss.org/software/rss/comments/236http://mloss.org/software/view/236/kernel methods