Keyword spotting system based on biomimetic pattern recognition

The mandarin keyword spotting system was investigated, and a new approach was proposed based on the principle of homology continuity and point location analysis in high-dimensional space geometry theory which are both parts of biomimetic pattern recognition theory. This approach constructed a hyper-polyhedron with sample points in the training set and calculated the distance between each test point and the hyper-polyhedron. The classification resulted from the value of those distances. The approach was tested by a speech database which was created by ourselves. The performance was compared with the classic HMM approach and the results show that the new approach is much better than HMM approach when the training data is not sufficient.

Information geometry theory of high-dimension space and application for speaker independent continuous digit speech recognition

In the light of descriptive geometry and notions in set theory, this paper re-defines the basic elements in space such as curve and surface and so on, presents some fundamental notions with respect to the point cover based on the High-dimension space (HDS) point covering theory, finally takes points from mapping part of speech signals to HDS, so as to analyze distribution information of these speech points in HDS, and various geometric covering objects for speech points and their relationship. Besides, this paper also proposes a new algorithm for speaker independent continuous digit speech recognition based on the HDS point dynamic searching theory without end-points detection and segmentation. First from the different digit syllables in real continuous digit speech, we establish the covering area in feature space for continuous speech. During recognition, we make use of the point covering dynamic searching theory in HDS to do recognition, and then get the satisfying recognized results. At last, compared to HMM (Hidden Markov models)-based method, from the development trend of the comparing results, as sample amount increasing, the difference of recognition rate between two methods will decrease slowly, while sample amount approaching to be very large, two recognition rates all close to 100% little by little. As seen from the results, the recognition rate of HDS point covering method is higher than that of in HMM (Hidden Markov models) based method, because, the point covering describes the morphological distribution for speech in HDS, whereas HMM-based method is only a probability distribution, whose accuracy is certainly inferior to point covering.

A speaker-independent continuous speech recognition system using biomimetic pattern recognition

In speaker-independent speech recognition, the disadvantage of the most diffused technology (HMMs, or Hidden Markov models) is not only the need of many more training samples, but also long train time requirement. This paper describes the use of Biomimetic pattern recognition (BPR) in recognizing some mandarin continuous speech in a speaker-independent manner. A speech database was developed for the course of study. The vocabulary of the database consists of 15 Chinese dish's names, the length of each name is 4 Chinese words. Neural networks (NNs) based on Multi-weight neuron (MWN) model are used to train and recognize the speech sounds. The number of MWN was investigated to achieve the optimal performance of the NNs-based BPR. This system, which is based on BPR and can carry out real time recognition reaches a recognition rate of 98.14% for the first option and 99.81% for the first two options to the persons from different provinces of China speaking common Chinese speech. Experiments were also carried on to evaluate Continuous density hidden Markov models (CDHMM), Dynamic time warping (DTW) and BPR for speech recognition. The Experiment results show that BPR outperforms CDHMM and DTW especially in the cases of samples of a finite size.

Biomimetic pattern recognition for speaker-independent speech recognition

In speaker-independent speech recognition, the disadvantage of the most diffused technology ( Hidden Markov Models) is not only the need of many more training samples, but also long train time requirement. This paper describes the use of Biomimetic Pattern Recognition (BPR) in recognizing some Mandarin Speech in a speaker-independent manner. The vocabulary of the system consists of 15 Chinese dish's names. Neural networks based on Multi-Weight Neuron (MWN) model are used to train and recognize the speech sounds. Experimental results are presented to show that the system, which can carry out real time recognition of the persons from different provinces speaking common Chinese speech, outperforms HMMs especially in the cases of samples of a finite size.

基于MRF模型的油气储层属性随机建模方法研究

Stochastic reservoir modeling is a technique used in reservoir describing. Through this technique, multiple data sources with different scales can be integrated into the reservoir model and its uncertainty can be conveyed to researchers and supervisors. Stochastic reservoir modeling, for its digital models, its changeable scales, its honoring known information and data and its conveying uncertainty in models, provides a mathematical framework or platform for researchers to integrate multiple data sources and information with different scales into their prediction models. As a fresher method, stochastic reservoir modeling is on the upswing. Based on related works, this paper, starting with Markov property in reservoir, illustrates how to constitute spatial models for catalogued variables and continuum variables by use of Markov random fields. In order to explore reservoir properties, researchers should study the properties of rocks embedded in reservoirs. Apart from methods used in laboratories, geophysical means and subsequent interpretations may be the main sources for information and data used in petroleum exploration and exploitation. How to build a model for flow simulations based on incomplete information is to predict the spatial distributions of different reservoir variables. Considering data source, digital extent and methods, reservoir modeling can be catalogued into four sorts: reservoir sedimentology based method, reservoir seismic prediction, kriging and stochastic reservoir modeling. The application of Markov chain models in the analogue of sedimentary strata is introduced in the third of the paper. The concept of Markov chain model, N-step transition probability matrix, stationary distribution, the estimation of transition probability matrix, the testing of Markov property, 2 means for organizing sections-method based on equal intervals and based on rock facies, embedded Markov matrix, semi-Markov chain model, hidden Markov chain model, etc, are presented in this part. Based on 1-D Markov chain model, conditional 1-D Markov chain model is discussed in the fourth part. By extending 1-D Markov chain model to 2-D, 3-D situations, conditional 2-D, 3-D Markov chain models are presented. This part also discusses the estimation of vertical transition probability, lateral transition probability and the initialization of the top boundary. Corresponding digital models are used to specify, or testify related discussions. The fifth part, based on the fourth part and the application of MRF in image analysis, discusses MRF based method to simulate the spatial distribution of catalogued reservoir variables. In the part, the probability of a special catalogued variable mass, the definition of energy function for catalogued variable mass as a Markov random field, Strauss model, estimation of components in energy function are presented. Corresponding digital models are used to specify, or testify, related discussions. As for the simulation of the spatial distribution of continuum reservoir variables, the sixth part mainly explores 2 methods. The first is pure GMRF based method. Related contents include GMRF model and its neighborhood, parameters estimation, and MCMC iteration method. A digital example illustrates the corresponding method. The second is two-stage models method. Based on the results of catalogued variables distribution simulation, this method, taking GMRF as the prior distribution for continuum variables, taking the relationship between catalogued variables such as rock facies, continuum variables such as porosity, permeability, fluid saturation, can bring a series of stochastic images for the spatial distribution of continuum variables. Integrating multiple data sources into the reservoir model is one of the merits of stochastic reservoir modeling. After discussing how to model spatial distributions of catalogued reservoir variables, continuum reservoir variables, the paper explores how to combine conceptual depositional models, well logs, cores, seismic attributes production history.

基于随机网络编码的无线广播重传方案及性能分析

针对自动最复重传(ARQ)机制在无线广播系统中吞吐量性能不佳的缺陷,提出一种基于随机网络编码的广播重传方案RNC-ARQ。对于广播节点,采用随机线性码对所有丢失包进行编码组合重传。对于接收节点,当接收的编码包累积到一定数量后可通过解码操作恢复出原始数据。该方案可有效减少重传次数,改善无线广播的吞吐量性能。基于Gilbert-Elliott模型描述的突发错误信道,建立了信道状态和节点接收处理流程合并的多状态马尔可夫模型,并以此为基础推导了RNC-ARQ方案的TQ吐量闭合解。最后,使用NS-2模拟器评估RNC-ARQ方案的性能,结果表明在突发差错信道下,基于随机网络编码重传方案的吞吐量优于传统的选择重传ARQ方案和基于异或编码的重传方案。

基于HMM-FNN模型的复杂动态手势识别

复杂动态手势识别是利用视频手势进行人机交互的关键问题.提出一种HMM-FNN模型结构.它整合了隐马尔可夫模型对时序数据的建模能力与模糊神经网络的模糊规则构建与推理能力,并将其应用到复杂动态手势的识别中.复杂动态手势具备两大特点:运动特征的可分解性与定义描述的模糊性.针对这两种特性,复杂手势被分解为手形变化、2D平面运动与Z轴方向运动3个子部分,分别利用HMM进行建模,HMM模型对观察子序列的似然概率被作为FNN的模糊隶属度,通过模糊规则推理,最终得到手势的分类类别.HMM-FNN方法将高维手势特征分解为低维子特征序列,降低了模型的复杂度.此外,它还可以充分利用人的经验辅助模型结构的创建与优化.实验表明,该方法是一种有效的复杂动态手势识别方法,并且优于传统的HMM模型方法.