A novel algorithm for Bayesian document segmentation is proposed based on the wavelet domain hidden Markov tree (HMT) model. Once the parameters of model are known, according to the sequential maximum a posterior prob...A novel algorithm for Bayesian document segmentation is proposed based on the wavelet domain hidden Markov tree (HMT) model. Once the parameters of model are known, according to the sequential maximum a posterior probability (SMAP) rule, firstly, the likelihood probability of HMT model for each pattern is computed from fine to coarse procedure. Then, the interscale state transition probability is solved using Expectation Maximum (EM) algorithm based on hybrid-quadtree and multiscale context information is fused from coarse to fine procedure. In order to get pixel-level segmentation, the redundant wavelet domain Gaussian mixture model (GMM) is employed to formulate pixel-level statistical property. The experiment results show that the proposed scheme is feasible and robust.展开更多
Many kinds of channel currents are especially weak and the background noise dominates in the patch clamp recordings. This makes the threshold detection fail during estimating of the transition probabilities. So direct...Many kinds of channel currents are especially weak and the background noise dominates in the patch clamp recordings. This makes the threshold detection fail during estimating of the transition probabilities. So direct fitting of the patch clamp recording, not of the histogram coming from the recordings, is a desirable way to estimate the transition probabilities. Iterative batch EM algorithm based on hidden markov model has been used in this field but which has the "curse of dimensionality" and besides cant keep tracking the varying of the parameters. A new on line sequential iterative one is proposed here, which needs fewer computational efforts and can adaptively keep tracking the varying of parameters. Simulations suggest its robust, effective and convenient.展开更多
论文在传统一阶隐马尔可夫模型的基础上,针对隐马尔可夫模型结构信息挖掘不全面的问题,提出了一种双层隐马尔可夫模型。双层隐马尔可夫模型在使用Baum-Welch算法的过程中将词性序列视为观测序列,通过Baum-Welch算法提取更多信息并最大...论文在传统一阶隐马尔可夫模型的基础上,针对隐马尔可夫模型结构信息挖掘不全面的问题,提出了一种双层隐马尔可夫模型。双层隐马尔可夫模型在使用Baum-Welch算法的过程中将词性序列视为观测序列,通过Baum-Welch算法提取更多信息并最大化词性序列概率从而更加贴合实际情况,同时对Viterbi算法做了相应的改动。模型在Penn Treebank语料库和Groningen Meaning Bank语料库上进行10折交叉验证,并与传统一阶、二阶隐马尔可夫模型进行对比。结果表明双层隐马尔可夫模型相较传统一阶、二阶隐马尔可夫模型词性标注正确率更高。展开更多
基金国家自然科学基金(the National Natural Science Foundation of China under Grant No.50279041)陕西省自然科学基金(the Natural Science Foundation of Shaanxi Province of China under Grant No.2005F07)。
文摘A novel algorithm for Bayesian document segmentation is proposed based on the wavelet domain hidden Markov tree (HMT) model. Once the parameters of model are known, according to the sequential maximum a posterior probability (SMAP) rule, firstly, the likelihood probability of HMT model for each pattern is computed from fine to coarse procedure. Then, the interscale state transition probability is solved using Expectation Maximum (EM) algorithm based on hybrid-quadtree and multiscale context information is fused from coarse to fine procedure. In order to get pixel-level segmentation, the redundant wavelet domain Gaussian mixture model (GMM) is employed to formulate pixel-level statistical property. The experiment results show that the proposed scheme is feasible and robust.
文摘Many kinds of channel currents are especially weak and the background noise dominates in the patch clamp recordings. This makes the threshold detection fail during estimating of the transition probabilities. So direct fitting of the patch clamp recording, not of the histogram coming from the recordings, is a desirable way to estimate the transition probabilities. Iterative batch EM algorithm based on hidden markov model has been used in this field but which has the "curse of dimensionality" and besides cant keep tracking the varying of the parameters. A new on line sequential iterative one is proposed here, which needs fewer computational efforts and can adaptively keep tracking the varying of parameters. Simulations suggest its robust, effective and convenient.
文摘论文在传统一阶隐马尔可夫模型的基础上,针对隐马尔可夫模型结构信息挖掘不全面的问题,提出了一种双层隐马尔可夫模型。双层隐马尔可夫模型在使用Baum-Welch算法的过程中将词性序列视为观测序列,通过Baum-Welch算法提取更多信息并最大化词性序列概率从而更加贴合实际情况,同时对Viterbi算法做了相应的改动。模型在Penn Treebank语料库和Groningen Meaning Bank语料库上进行10折交叉验证,并与传统一阶、二阶隐马尔可夫模型进行对比。结果表明双层隐马尔可夫模型相较传统一阶、二阶隐马尔可夫模型词性标注正确率更高。
