arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1502.01510stat.ME
收藏

哈密顿蒙特卡罗的基本不相容性与数据欠采样

The Fundamental Incompatibility of Hamiltonian Monte Carlo and Data Subsampling

M. J. Betancourt

摘要:

利用哈密顿流的相干探索,哈密顿蒙特卡罗产生计算效率高的蒙特卡罗估计,即使对于复杂和高维的目标分布。然而,当遇到数据密集型应用程序时,该算法的实现可能过于昂贵,这让我们需要考虑数据二次采样等近似方法的实用性。在本文中,我演示了如何从根本上妥协的有效探索汉密尔顿流的数据二次采样,因此可扩展性能的哈密顿蒙特卡罗本身。

英文摘要:

Leveraging the coherent exploration of Hamiltonian flow, Hamiltonian Monte Carlo produces computationally efficient Monte Carlo estimators, even with respect to complex and high-dimensional target distributions. When confronted with data-intensive applications, however, the algorithm may be too expensive to implement, leaving us to consider the utility of approximations such as data subsampling. In this paper I demonstrate how data subsampling fundamentally compromises the efficient exploration of Hamiltonian flow and hence the scalable performance of Hamiltonian Monte Carlo itself.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑