Paper 2

Divide-and-Conquer Parallelism for Learning Mixture Models

Authors: Takaya Kawakatsu, Akira Kinoshita, Atsuhiro Takasu, and Jun Adachi

Volume 28 (2016)

Abstract

From the viewpoint of load balancing among processors, the acceleration of machine-learning algorithms by using parallel loops is not realistic for some models involving hierarchical parameter estimation. There are also other serious issues such as memory access speed and race conditions. Some approaches to the race condition problem, such as mutual exclusion and atomic operations, degrade the memory access performance. Another issue is that the rst-in-rst-out (FIFO) scheduler supported by frameworks such as Hadoop can waste considerable time on queuing and this will also affect the learning speed. In this paper, we propose a recursive divide-and-conquer-based parallelization method for high-speed machine learning. Our approach exploits a tree structure for recursive tasks, which enables effective load balancing. Race conditions are also avoided, without slowing down the memory access, by separating the variables for summation. We have applied our approach to tasks that involve learning mixture models. Our experimental results show scalability superior to FIFO scheduling with an atomic-based solution to race conditions and robustness against load imbalance.