首页> 外文会议>Principles and practice of constraint programming >Exact Sampling for Regular and Markov Constraints with Belief Propagation
【24h】

Exact Sampling for Regular and Markov Constraints with Belief Propagation

机译:具有信念传播的正则和马尔可夫约束的精确采样

获取原文
获取原文并翻译 | 示例

摘要

Sampling random sequences from a statistical model, subject to hard constraints, is generally a difficult task. In this paper, we show that for Markov models and a set of Regular global constraints and unary constraints, we can perform perfect sampling. This is achieved by defining a factor graph, composed of binary factors that combine a Markov chain and an automaton. We apply a simplified version of belief propagation to sample random sequences satisfying the global constraints, with their correct probability. Since the factor graph is linear, this procedure is efficient and exact. We illustrate this approach to the generation of sequences of text or music, imitating the style of a corpus, and verifying validity constraints, such as syntax or meter.
机译:在严格的约束下,从统计模型中采样随机序列通常是一项艰巨的任务。在本文中,我们表明对于Markov模型以及一组正则全局约束和一元约束,我们可以执行完美的采样。这是通过定义一个因子图来实现的,该因子图由结合了马尔可夫链和自动机的二元因子组成。我们将信念传播的简化版本应用于满足全局约束且具有正确概率的样本随机序列。由于因子图是线性的,因此此过程高效且准确。我们说明了这种方法来生成文本或音乐序列,模仿语料库的样式以及验证有效性约束(例如语法或计量器)。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号