AcMC 2: Accelerating Markov Chain Monte Carlo Algorithms for Probabilistic Models
AcMC 2: Accelerating Markov Chain Monte Carlo Algorithms for Probabilistic Models
复制标题
AcMC 2：加速概率模型的马尔可夫链蒙特卡罗算法
DOI：
--
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
R. Iyer
中科院分区：
文献类型：
--
作者：
Subho Sankar Banerjee;Z. Kalbarczyk;R. Iyer
概率模型（ProbabilityModel，PM）广泛应用于各种机器学习应用中。它们已被证明可以成功地整合有关数据的结构先验信息，并有效地量化不确定性，从而能够开发出更强大、更可解释、更高效的学习算法。本文介绍了AcMC 2，编译器，PM转换为优化的硬件加速器（用于FPGA或ASIC），利用马尔可夫链蒙特卡罗方法来推断和查询模型的后验样本的分布。编译器分析PM中的统计依赖关系，以驱动若干优化，从而最大限度地利用问题中可用的并行性和数据局部性。我们演示了使用AcMC 2在Xilinx Virtex-7 FPGA上实现几个学习和推理任务。AcMC 2生成的加速器在运行时性能上比6核IBM Power8 CPU提高了47-100倍，比NVIDIA K80 GPU提高了8-18倍。这相当于比CPU提高了753-1600倍，比GPU提高了248-463倍。
Probabilistic models (PMs) are ubiquitously used across a variety of machine learning applications. They have been shown to successfully integrate structural prior information about data and effectively quantify uncertainty to enable the development of more powerful, interpretable, and efficient learning algorithms. This paper presents AcMC2, a compiler that transforms PMs into optimized hardware accelerators (for use in FPGAs or ASICs) that utilize Markov chain Monte Carlo methods to infer and query a distribution of posterior samples from the model. The compiler analyzes statistical dependencies in the PM to drive several optimizations to maximally exploit the parallelism and data locality available in the problem. We demonstrate the use of AcMC2 to implement several learning and inference tasks on a Xilinx Virtex-7 FPGA. AcMC2-generated accelerators provide a 47-100× improvement in runtime performance over a 6-core IBM Power8 CPU and a 8-18× improvement over an NVIDIA K80 GPU. This corresponds to a 753-1600× improvement over the CPU and 248-463× over the GPU in performance-per-watt terms.
影响因子：
5.8
作者：
Carpenter, Bob;Gelman, Andrew;Riddell, Allen
通讯作者：
Riddell, Allen