摘要: In EM and related algorithms, E-step computations distribute easily, because data items are … We present a framework that fully distributes the entire EM procedure. Each node interacts …
Cheng-tao Chu, Sang Kim, Yi-An Lin, YuanYuan Yu, Gary Bradski, Kunle Olukotun, Andrew Ng, None, Map-Reduce for Machine Learning on Multicoreneural information processing systems. ,vol. 19, pp. 281- 288 ,(2006)