Year: 2016
Type: preprint
Abstract: We demonstrate that distributed block coordinate descent can quickly solve kernel regression and classification problems with millions of data points. Armed with this capability, we conduct a thorough comparison between the full kernel, the Nyström method, and random features on three large classification tasks from various domains. Our results suggest that the Nyström method generally achieves better statistical accuracy than random features, but can require significantly more iterations of optimization. Lastly, we derive new rates for block coordinate descent which support our experimental findings when specialized to kernel methods.
Source: arXiv (Cornell University)
Authors: Tu, Stephen, Roelofs, Rebecca, Venkataraman, Shivaram, Recht, Benjamin
Institution: University of California, Berkeley
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funders: National Science Foundation, U.S. Department of Energy, International Business Machines Corporation, Cisco Systems, Amazon Web Services, Thomas and Stacey Siebel Foundation, NetApp, VMware, Directorate for Computer and Information Science and Engineering, Defense Advanced Research Projects Agency, Samsung, Office of Naval Research, Lawrence Berkeley National Laboratory
Awards: G1079818582, G1286236842, G1519837837, G3026391708, G3083819904, G4654920669, G4713059963, G5222170445, Making Sense at Scale with Algorithms, Machines, and People, G674549868, CIF: Small: Decoding Error-Correcting Codes using Large-Scale Decomposition Methods, G7103680314, G7354628648, CAREER: Efficient Atomic Decompositions of Massive Data Sets, G8560804567, G8876996369
OpenAlex ID: W2278578264
