一句话：指出仅以端到端精度评估 MCMC 加速器不足，需关注硬件优化对统计性质的影响。
概率机器学习常使用 MCMC 等概率模型与算法，这类计算在通用处理器上较慢，可通过专用硬件并行化与近似优化来加速。然而，现有概率加速器正确性评估多只关注最终结果质量（“精度”），忽视了硬件优化对统计性质的影响。论文主张硬件设计者与领域专家应超越端到端精度，审视并行和近似等设计选择如何改变推断算法的统计行为，从而更全面地判断加速器的统计稳健性。摘要未展开具体实验与结果。
Statistical machine learning often uses probabilistic models and algorithms, such as Markov Chain Monte Carlo (MCMC), to solve a wide range of problems. Probabilistic computations, often considered too slow on conventional processors, can be accelerated with specialized hardware by exploiting parallelism and optimizing the design using various approximation techniques. Current methodologies for evaluating correctness of probabilistic accelerators are often incomplete, mostly focusing only on end-point result quality ("accuracy"). It is important for hardware designers and domain experts to look beyond end-point "accuracy" and be aware of how hardware optimizations impact statistical properties.
DOI 原文 ·
@inproceedings{paperbot6125,
title = {Statistical robustness of Markov chain Monte Carlo accelerators},
author = {Xiangyu Zhang and Ramin Bashizade and Yicheng Wang and Sayan Mukherjee and Alvin R. Lebeck},
booktitle = {ASPLOS},
year = {2021},
doi = {10.1145/3445814.3446697}
}