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.

Study on text-dependent speaker recognition based on Biomimetic Pattern Recognition

We studied the application of Biomimetic Pattern Recognition to speaker recognition. A speaker recognition neural network using network matching degree as criterion is proposed. It has been used in the system of text-dependent speaker recognition. Experimental results show that good effect could be obtained even with lesser samples. Furthermore, the misrecognition caused by untrained speakers occurring in testing could be controlled effectively. In addition, the basic idea "cognition" of Biomimetic Pattern Recognition results in no requirement of retraining the old system for enrolling new speakers.

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.

High-performance EML grown on taper-masked pattern substrates by ultra-low-pressure MOCVD

A novel in-plane bandgap energy controlling technique by ultra-low pressure (22 mbar) selective area growth (SAG) has been developed. To our knowledge, this is the lowest pressure condition during SAG process ever reported. In this work, high crystalline quality InGaAsP-InP MQWs with a photoluminescence (PL) full-width at half-maximum (FWHM) of less than 35meV are selectively grown on mask-patterned planar InP substrates by ultra-low pressure (22 mbar) metal-organic chemical vapor deposition (MOCVD). In order to study the uniformity of the MQWs grown in the selective area, novel tapered masks are designed and used. Through optimizing growth conditions, a wide wavelength shift of over 80 nm with a rather small mask width variation (0-30 mu m) is obtained. The mechanism of ultra-low pressure SAG is detailed by analyzing the effect of various mask designs and quantum well widths. This powerful technique is then applied to fabricate an electroabsorption-modulated laser (EML). Superior device characteristics are achieved, such as a low threshold current of 19mA and an output power of 7mW. (c) 2005 Elsevier B.V. All rights reserved.

A new development on ANN in China - Biomimetic pattern recognition and multi weight vector neurons

A new model of pattern recognition principles-Biomimetic Pattern Recognition, which is based on "matter cognition" instead of "matter classification", has been proposed. As a important means realizing Biomimetic Pattern Recognition, the mathematical model and analyzing method of ANN get breakthrough: a novel all-purpose mathematical model has been advanced, which can simulate all kinds of neuron architecture, including RBF and BP models. As the same time this model has been realized using hardware; the high-dimension space geometry method, a new means to analyzing ANN, has been researched.

Anisotropic radiation pattern from InGaAlP quantum well mesa-like microdisks

To overcome the isotropic directional emission of an ideal circular microdisk, two kinds of cylindrical mesa-like InGaAlP single quantum well (SQW) microdisks emitting at a visible red wavelength of 0.66 mu m have been fabricated. An anisotropic luminescence pattern was revealed by the microscopic fluorescence (FL) image. FL intensity, preferentially enhanced with twofold symmetry, appeared at the circumference of the InGaAlP SQW microdisks. Our results demonstrated that anisotropic radiation can be achieved by geometry shaping of the disks on the top view two-dimensional boundary slightly deformed from circular shape and/or on the side-view cross-section of the circular mesa by wet etching anisotropic undercut. (C) 2000 Elsevier Science Ltd. All rights reserved.

Object-recognition with oblique observation directions based on biomimetic pattern recognition

In this paper, we propose a new scheme for omnidirectional object-recognition in free space. The proposed scheme divides above problem into several onmidirectional object-recognition with different depression angles. An onmidirectional object-recognition system with oblique observation directions based on a new recognition theory-Biomimetic Pattern Recognition (BPR) is discussed in detail. Based on it, we can get the size of training samples in the onmidirectional object-recognition system in free space. Omnidirection ally cognitive tests were done on various kinds of animal models of rather similar shapes. For the total 8400 tests, the correct recognition rate is 99.89%. The rejection rate is 0.11% and on the condition of zero error rates. Experimental results are presented to show that the proposed approach outperforms three types of SVMs with either a three degree polynomial kernel or a radial basis function kernel.

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.

A more complex neuron in biomimetic pattern recognition

Biomimetic pattern recognition has been proposed for several years, but the discussion of its neuron was not very wide and deep. In this paper, we propose a new more complex neuron named M-neuron and give the application in the last part of the paper.

Study on text-dependent speaker recognition based on Biomimetic Pattern Recognition

We studied the application of Biomimetic Pattern Recognition to speaker recognition. A speaker recognition neural network using network matching degree as criterion is proposed. It has been used in the system of text-dependent speaker recognition. Experimental results show that good effect could be obtained even with lesser samples. Furthermore, the misrecognition caused by untrained speakers occurring in testing could be controlled effectively. In addition, the basic idea "cognition" of Biomimetic Pattern Recognition results in no requirement of retraining the old system for enrolling new speakers.

High-performance EML grown on taper-masked pattern substrates by ultra-low-pressure MOCVD

A novel in-plane bandgap energy controlling technique by ultra-low pressure (22 mbar) selective area growth (SAG) has been developed. To our knowledge, this is the lowest pressure condition during SAG process ever reported. In this work, high crystalline quality InGaAsP-InP MQWs with a photoluminescence (PL) full-width at half-maximum (FWHM) of less than 35meV are selectively grown on mask-patterned planar InP substrates by ultra-low pressure (22 mbar) metal-organic chemical vapor deposition (MOCVD). In order to study the uniformity of the MQWs grown in the selective area, novel tapered masks are designed and used. Through optimizing growth conditions, a wide wavelength shift of over 80 nm with a rather small mask width variation (0-30 mu m) is obtained. The mechanism of ultra-low pressure SAG is detailed by analyzing the effect of various mask designs and quantum well widths. This powerful technique is then applied to fabricate an electroabsorption-modulated laser (EML). Superior device characteristics are achieved, such as a low threshold current of 19mA and an output power of 7mW. (c) 2005 Elsevier B.V. All rights reserved.

Comparing the Performance of Biomimetic Pattern Recognition with W8 and SVM on Prediction of Horizontal Gene Transfers in Bacteria Genomes

National Natural Science Foundation of China 60753001

Design of surface emitting distributed feedback quantum cascade laser with single-lobe far-field pattern and high outcoupling efficiency

A 7.8-mu m surface emitting second-order distributed feedback quantum cascade laser (DFB QCL) structure with metallized surface grating is studied. The modal property of this structure is described by utilizing coupled-mode theory where the coupling coefficients are derived from exact Floquet-Bloch solutions of infinite periodic structure. Based on this theory, the influence of waveguide structure and grating topography as well as device length on the laser performance is numerically investigated. The optimized surface emitting second-order DFB QCL structure design exhibits a high surface outcoupling efficiency of 22% and a low threshold gain of 10 cm(-1). Using a pi phase-shift in the centre of the grating, a high-quality single-lobe far-field radiation pattern is obtained.

pattern matching compilation of functions defined in context-free languages

LFC is a functional language based on recursive functions defined in context-free languages. In this paper, a new pattern matching algorithm for LFC is presented, which can represent a sequence of patterns as an integer by an encoding method. It is a rather simple method and produces efficient case-expressions for pattern matching definitions of LFC. The algorithm can also be used for other functional languages, but for nested patterns it may become complicated and further studies are needed.