Design of Nonlinear Observer by Using Augmented Linear System based on Formal Linearization of Polynomial Type
The objective of this study is to propose an observer design for nonlinear systems by using an augmented linear system derived by application of a formal linearization method. A given nonlinear differential equation is linearized by the formal linearization method which is based on Taylor expansion considering up to the higher order terms, and a measurement equation is transformed into an augmented linear one. To this augmented dimensional linear system, a linear estimation theory is applied and a nonlinear observer is derived. As an application of this method, an estimation problem of transient state of electric power systems is studied, and its numerical experiments indicate that this observer design shows remarkable performances for nonlinear systems.
Smith Predictor Design by CDM for Temperature Control System
Smith Predictor control is theoretically a good solution to the problem of controlling the time delay systems. However, it seldom gets use because it is almost impossible to find out a precise mathematical model of the practical system and very sensitive to uncertain system with variable time-delay. In this paper is concerned with a design method of smith predictor for temperature control system by Coefficient Diagram Method (CDM). The simulation results show that the control system with smith predictor design by CDM is stable and robust whilst giving the desired time domain system performance.
Natural Language Database Interface for Selection of Data Using Grammar and Parsing
Databases have become ubiquitous. Almost all IT applications are storing into and retrieving information from databases. Retrieving information from the database requires knowledge of technical languages such as Structured Query Language (SQL). However majority of the users who interact with the databases do not have a technical background and are intimidated by the idea of using languages such as SQL. This has led to the development of a few Natural Language Database Interfaces (NLDBIs). A NLDBI allows the user to query the database in a natural language. This paper highlights on architecture of new NLDBI system, its implementation and discusses on results obtained. In most of the typical NLDBI systems the natural language statement is converted into an internal representation based on the syntactic and semantic knowledge of the natural language. This representation is then converted into queries using a representation converter. A natural language query is translated to an equivalent SQL query after processing through various stages. The work has been experimented on primitive database queries with certain constraints.
Study of Efficiency and Capability LZW++ Technique in Data Compression
The purpose of this paper is to show efficiency and capability LZWµ in data compression. The LZWµ technique is enhancement from existing LZW technique. The modification the existing LZW is needed to produce LZWµ technique. LZW read one by one character at one time. Differ with LZWµ technique, where the LZWµ read three characters at one time. This paper focuses on data compression and tested efficiency and capability LZWµ by different data format such as doc type, pdf type and text type. Several experiments have been done by different types of data format. The results shows LZWµ technique is better compared to existing LZW technique in term of file size.
Efficient Block Matching Algorithm for Motion Estimation
Motion estimation is a key problem in video
processing and computer vision. Optical flow motion estimation can
achieve high estimation accuracy when motion vector is small.
Three-step search algorithm can handle large motion vector but not
very accurate. A joint algorithm was proposed in this paper to
achieve high estimation accuracy disregarding whether the motion
vector is small or large, and keep the computation cost much lower
than full search.
Scale-Space Volume Descriptors for Automatic 3D Facial Feature Extraction
An automatic method for the extraction of feature points for face based applications is proposed. The system is based upon volumetric feature descriptors, which in this paper has been extended to incorporate scale space. The method is robust to noise and has the ability to extract local and holistic features simultaneously from faces stored in a database. Extracted features are stable over a range of faces, with results indicating that in terms of intra-ID variability, the technique has the ability to outperform manual landmarking.
Robust Statistics Based Algorithm to Remove Salt and Pepper Noise in Images
In this paper, a robust statistics based filter to remove salt and pepper noise in digital images is presented. The function of the algorithm is to detect the corrupted pixels first since the impulse noise only affect certain pixels in the image and the remaining pixels are uncorrupted. The corrupted pixels are replaced by an estimated value using the proposed robust statistics based filter. The proposed method perform well in removing low to medium density impulse noise with detail preservation upto a noise density of 70% compared to standard median filter, weighted median filter, recursive weighted median filter, progressive switching median filter, signal dependent rank ordered mean filter, adaptive median filter and recently proposed decision based algorithm. The visual and quantitative results show the proposed algorithm outperforms in restoring the original image with superior preservation of edges and better suppression of impulse noise
Load Balancing in Genetic Zone Routing Protocol for MANETs
Genetic Zone Routing Protocol (GZRP) is a new
hybrid routing protocol for MANETs which is an extension of ZRP
by using Genetic Algorithm (GA). GZRP uses GA on IERP and BRP
parts of ZRP to provide a limited set of alternative routes to the
destination in order to load balance the network and robustness
during node/link failure during the route discovery process. GZRP is
studied for its performance compared to ZRP in many folds like
scalability for packet delivery and proved with improved results. This
paper presents the results of the effect of load balancing on GZRP.
The results show that GZRP outperforms ZRP while balancing the
Using Secure-Image Mechanism to Protect Mobile Agent Against Malicious Hosts
The usage of internet is rapidly increasing and the usage of mobile agent technology in internet environment has a great demand. The security issue one of main obstacles that restrict the mobile agent technology to spread. This paper proposes Secure-Image Mechanism (SIM) as a new mechanism to protect mobile agents against malicious hosts. . SIM aims to protect mobile agent by using the symmetric encryption and hash function in cryptography science. This mechanism can prevent the eavesdropping and alteration attacks. It assists the mobile agents to continue their journey normally incase attacks occurred.
Construct Pairwise Test Suites Based on the Bak-Sneppen Model of Biological Evolution
Pairwise testing, which requires that every
combination of valid values of each pair of system factors be covered
by at lease one test case, plays an important role in software testing
since many faults are caused by unexpected 2-way interactions among
system factors. Although meta-heuristic strategies like simulated
annealing can generally discover smaller pairwise test suite, they may
cost more time to perform search, compared with greedy algorithms.
We propose a new method, improved Extremal Optimization (EO)
based on the Bak-Sneppen (BS) model of biological evolution, for
constructing pairwise test suites and define fitness function according
to the requirement of improved EO. Experimental results show that
improved EO gives similar size of resulting pairwise test suite and
yields an 85% reduction in solution time over SA.
Free-Form Query for Cell Phones
It is a challenge to provide a wide range of queries to
database query systems for small mobile devices, such as the PDAs
and cell phones. Currently, due to the physical and resource
limitations of these devices, most reported database querying systems
developed for them are only offering a small set of pre-determined
queries for users to possibly pose. The above can be resolved by
allowing free-form queries to be entered on the devices. Hence, a
query language that does not restrict the combination of query terms
entered by users is proposed. This paper presents the free-form query
language and the method used in translating free-form queries to
their equivalent SQL statements.
Reversible Watermarking on Stereo Image Sequences
In this paper, a new reversible watermarking method is presented that reduces the size of a stereoscopic image sequence while keeping its content visible. The proposed technique embeds the residuals of the right frames to the corresponding frames of the left sequence, halving the total capacity. The residual frames may result in after a disparity compensated procedure between the two video streams or by a joint motion and disparity compensation. The residuals are usually lossy compressed before embedding because of the limited embedding capacity of the left frames. The watermarked frames are visible at a high quality and at any instant the stereoscopic video may be recovered by an inverse process. In fact, the left frames may be exactly recovered whereas the right ones are slightly distorted as the residuals are not embedded intact. The employed embedding method reorders the left frame into an array of consecutive pixel pairs and embeds a number of bits according to their intensity difference. In this way, it hides a number of bits in intensity smooth areas and most of the data in textured areas where resulting distortions are less visible. The experimental evaluation demonstrates that the proposed scheme is quite effective.
Fuzzy Fingerprint Vault using Multiple Polynomials
Fuzzy fingerprint vault is a recently developed cryptographic construct based on the polynomial reconstruction problem to secure critical data with the fingerprint data. However, the previous researches are not applicable to the fingerprint having a few minutiae since they use a fixed degree of the polynomial without considering the number of fingerprint minutiae. To solve this problem, we use an adaptive degree of the polynomial considering the number of minutiae extracted from each user. Also, we apply multiple polynomials to avoid the possible degradation of the security of a simple solution(i.e., using a low-degree polynomial). Based on the experimental results, our method can make the possible attack difficult 2192 times more than using a low-degree polynomial as well as verify the users having a few minutiae.
Business Rules for Data Warehouse
Business rules and data warehouse are concepts and
technologies that impact a wide variety of organizational tasks. In
general, each area has evolved independently, impacting application
development and decision-making. Generating knowledge from data
warehouse is a complex process. This paper outlines an approach to
ease import of information and knowledge from a data warehouse
star schema through an inference class of business rules. The paper
utilizes the Oracle database for illustrating the working of the
concepts. The star schema structure and the business rules are stored
within a relational database. The approach is explained through a
prototype in Oracle-s PL/SQL Server Pages.
A Complexity-Based Approach in Image Compression using Neural Networks
In this paper we present an adaptive method for image
compression that is based on complexity level of the image. The
basic compressor/de-compressor structure of this method is a multilayer
perceptron artificial neural network. In adaptive approach
different Back-Propagation artificial neural networks are used as
compressor and de-compressor and this is done by dividing the
image into blocks, computing the complexity of each block and then
selecting one network for each block according to its complexity
value. Three complexity measure methods, called Entropy, Activity
and Pattern-based are used to determine the level of complexity in
image blocks and their ability in complexity estimation are evaluated
and compared. In training and evaluation, each image block is
assigned to a network based on its complexity value. Best-SNR is
another alternative in selecting compressor network for image blocks
in evolution phase which chooses one of the trained networks such
that results best SNR in compressing the input image block. In our
evaluations, best results are obtained when overlapping the blocks is
allowed and choosing the networks in compressor is based on the
Best-SNR. In this case, the results demonstrate superiority of this
method comparing with previous similar works and JPEG standard
Building the Reliability Prediction Model of Component-Based Software Architectures
Reliability is one of the most important quality attributes of software. Based on the approach of Reussner and the approach of Cheung, we proposed the reliability prediction model of component-based software architectures. Also, the value of the model is shown through the experimental evaluation on a web server system.
Adaptive Kernel Principal Analysis for Online Feature Extraction
The batch nature limits the standard kernel principal component analysis (KPCA) methods in numerous applications, especially for dynamic or large-scale data. In this paper, an efficient adaptive approach is presented for online extraction of the kernel principal components (KPC). The contribution of this paper may be divided into two parts. First, kernel covariance matrix is correctly updated to adapt to the changing characteristics of data. Second, KPC are recursively formulated to overcome the batch nature of standard KPCA.This formulation is derived from the recursive eigen-decomposition of kernel covariance matrix and indicates the KPC variation caused by the new data. The proposed method not only alleviates sub-optimality of the KPCA method for non-stationary data, but also maintains constant update speed and memory usage as the data-size increases. Experiments for simulation data and real applications demonstrate that our approach yields improvements in terms of both computational speed and approximation accuracy.
A Scheme of Model Verification of the Concurrent Discrete Wavelet Transform (DWT) for Image Compression
The scientific community has invested a great deal of effort in the fields of discrete wavelet transform in the last few decades. Discrete wavelet transform (DWT) associated with the vector quantization has been proved to be a very useful tool for the compression of image. However, the DWT is very computationally intensive process requiring innovative and computationally efficient method to obtain the image compression. The concurrent transformation of the image can be an important solution to this problem. This paper proposes a model of concurrent DWT for image compression. Additionally, the formal verification of the model has also been performed. Here the Symbolic Model Verifier (SMV) has been used as the formal verification tool. The system has been modeled in SMV and some properties have been verified formally.
Fully Parameterizable FPGA based Crypto-Accelerator
In this paper, RSA encryption algorithm and its hardware
implementation in Xilinx-s Virtex Field Programmable Gate
Arrays (FPGA) is analyzed. The issues of scalability, flexible performance,
and silicon efficiency for the hardware acceleration of
public key crypto systems are being explored in the present work.
Using techniques based on the interleaved math for exponentiation,
the proposed RSA calculation architecture is compared to existing
FPGA-based solutions for speed, FPGA utilization, and scalability.
The paper covers the RSA encryption algorithm, interleaved multiplication,
Miller Rabin algorithm for primality test, extended Euclidean
math, basic FPGA technology, and the implementation details of
the proposed RSA calculation architecture. Performance of several
alternative hardware architectures is discussed and compared. Finally,
conclusion is drawn, highlighting the advantages of a fully flexible
& parameterized design.
Image Adaptive Watermarking with Visual Model in Orthogonal Polynomials based Transformation Domain
In this paper, an image adaptive, invisible digital
watermarking algorithm with Orthogonal Polynomials based
Transformation (OPT) is proposed, for copyright protection of digital
images. The proposed algorithm utilizes a visual model to determine
the watermarking strength necessary to invisibly embed the
watermark in the mid frequency AC coefficients of the cover image,
chosen with a secret key. The visual model is designed to generate a
Just Noticeable Distortion mask (JND) by analyzing the low level
image characteristics such as textures, edges and luminance of the
cover image in the orthogonal polynomials based transformation
domain. Since the secret key is required for both embedding and
extraction of watermark, it is not possible for an unauthorized user to
extract the embedded watermark. The proposed scheme is robust to
common image processing distortions like filtering, JPEG
compression and additive noise. Experimental results show that the
quality of OPT domain watermarked images is better than its DCT
Visualisation and Navigation in Large Scale P2P Service Networks
In Peer-to-Peer service networks, where peers offer any kind of publicly available services or applications, intuitive navigation through all services in the network becomes more difficult as the number of services increases. In this article, a concept is discussed that enables users to intuitively browse and use large scale P2P service networks. The concept extends the idea of creating virtual 3D-environments solely based on Peer-to-Peer technologies. Aside from browsing, users shall have the possibility to emphasize services of interest using their own semantic criteria. The appearance of the virtual world shall intuitively reflect network properties that may be of interest for the user. Additionally, the concept comprises options for load- and traffic-balancing. In this article, the requirements concerning the underlying infrastructure and the graphical user interface are defined. First impressions of the appearance of future systems are presented and the next steps towards a prototypical implementation are discussed.
Security Analysis on the Online Office and Proposal of the Evaluation Criteria
The online office is one of web application. We can
easily use the online office through a web browser with internet
connected PC. The online office has the advantage of using
environment regardless of location or time. When users want to use the
online office, they access the online office server and use their content.
However, recently developed and launched online office has the
weakness of insufficient consideration. In this paper, we analyze the
security vulnerabilities of the online office. In addition, we propose
the evaluation criteria to make secure online office using Common
Criteria. This evaluation criteria can be used to establish trust between
the online office server and the user. The online office market will be
more active than before.
Security Analysis on Anonymous Mutual Authentication Protocol for RFID Tag without Back-End Database and its Improvement
RFID (Radio Frequency IDentification) system has
been widely used in our life, such as transport systems, passports,
automotive, animal tracking, human implants, library, and so on.
However, the RFID authentication protocols between RF (Radio
Frequency) tags and the RF readers have been bring about various
privacy problems that anonymity of the tags, tracking, eavesdropping,
and so on. Many researchers have proposed the solution of the
problems. However, they still have the problem, such as location
privacy, mutual authentication. In this paper, we show the problems of
the previous protocols, and then we propose a more secure and
efficient RFID authentication protocol.
Dempster-Shafer Evidence Theory for Image Segmentation: Application in Cells Images
In this paper we propose a new knowledge model using
the Dempster-Shafer-s evidence theory for image segmentation and
fusion. The proposed method is composed essentially of two steps.
First, mass distributions in Dempster-Shafer theory are obtained from
the membership degrees of each pixel covering the three image
components (R, G and B). Each membership-s degree is determined by
applying Fuzzy C-Means (FCM) clustering to the gray levels of the
three images. Second, the fusion process consists in defining three
discernment frames which are associated with the three images to be
fused, and then combining them to form a new frame of discernment.
The strategy used to define mass distributions in the combined
framework is discussed in detail. The proposed fusion method is
illustrated in the context of image segmentation. Experimental
investigations and comparative studies with the other previous methods
are carried out showing thus the robustness and superiority of the
proposed method in terms of image segmentation.
Improved Wavelet Neural Networks for Early Cancer Diagnosis Using Clustering Algorithms
Wavelet neural networks (WNNs) have emerged as a vital alternative to the vastly studied multilayer perceptrons (MLPs) since its first implementation. In this paper, we applied various clustering algorithms, namely, K-means (KM), Fuzzy C-means (FCM), symmetry-based K-means (SBKM), symmetry-based Fuzzy C-means (SBFCM) and modified point symmetry-based K-means (MPKM) clustering algorithms in choosing the translation parameter of a WNN. These modified WNNs are further applied to the heterogeneous cancer classification using benchmark microarray data and were compared against the conventional WNN with random initialization method. Experimental results showed that a WNN classifier with the MPKM algorithm is more precise than the conventional WNN as well as the WNNs with other clustering algorithms.
An Efficient Graph Query Algorithm Based on Important Vertices and Decision Features
Graph has become increasingly important in modeling
complicated structures and schemaless data such as proteins, chemical
compounds, and XML documents. Given a graph query, it is desirable
to retrieve graphs quickly from a large database via graph-based
indices. Different from the existing methods, our approach, called
VFM (Vertex to Frequent Feature Mapping), makes use of vertices
and decision features as the basic indexing feature. VFM constructs
two mappings between vertices and frequent features to answer graph
queries. The VFM approach not only provides an elegant solution to
the graph indexing problem, but also demonstrates how database
indexing and query processing can benefit from data mining,
especially frequent pattern mining. The results show that the proposed
method not only avoids the enumeration method of getting subgraphs
of query graph, but also effectively reduces the subgraph isomorphism
tests between the query graph and graphs in candidate answer set in
Optimization Parameters of Rotary Positioner Controller using CDM
The authors present optimization parameters of rotary
positioner controller in hard disk drive servo track writing process
using coefficient diagram method; CDM. Due to estimation
parameters in PI Positioning Control System by expected ratio
method cannot meet the required specification of response
effectively, we suggest coefficient diagram method for defining
controller parameters under the requirement of the system. Finally,
the simulation results show that our proposed method can improve
the problem in tuning parameter of rotary positioner controller. It is
satisfied specification of performance of control system. Furthermore,
it is very convenient as a fast adjustment damping ratio as well as a
high speed response.
Non-contact Gaze Tracking with Head Movement Adaptation based on Single Camera
With advances in computer vision, non-contact gaze tracking systems are heading towards being much easier to operate and more comfortable for use, the technique proposed in this paper is specially designed for achieving these goals. For the convenience in operation, the proposal aims at the system with simple configuration which is composed of a fixed wide angle camera and dual infrared illuminators. Then in order to enhance the usability of the system based on single camera, a self-adjusting method which is called Real-time gaze Tracking Algorithm with head movement Compensation (RTAC) is developed for estimating the gaze direction under natural head movement and simplifying the calibration procedure at the same time. According to the actual evaluations, the average accuracy of about 1° is achieved over a field of 20×15×15 cm3.
An Improved Greedy Routing Algorithm for Grid using Pheromone-Based Landmarks
This paper objects to extend Jon Kleinberg-s research. He introduced the structure of small-world in a grid and shows with a greedy algorithm using only local information able to find route between source and target in delivery time O(log2n). His fundamental model for distributed system uses a two-dimensional grid with longrange random links added between any two node u and v with a probability proportional to distance d(u,v)-2. We propose with an additional information of the long link nearby, we can find the shorter path. We apply the ant colony system as a messenger distributed their pheromone, the long-link details, in surrounding area. The subsequence forwarding decision has more option to move to, select among local neighbors or send to node has long link closer to its target. Our experiment results sustain our approach, the average routing time by Color Pheromone faster than greedy method.
Using HMM-based Classifier Adapted to Background Noises with Improved Sounds Features for Audio Surveillance Application
Discrimination between different classes of environmental
sounds is the goal of our work. The use of a sound recognition
system can offer concrete potentialities for surveillance and
security applications. The first paper contribution to this research
field is represented by a thorough investigation of the applicability
of state-of-the-art audio features in the domain of environmental
sound recognition. Additionally, a set of novel features obtained by
combining the basic parameters is introduced. The quality of the
features investigated is evaluated by a HMM-based classifier to which
a great interest was done. In fact, we propose to use a Multi-Style
training system based on HMMs: one recognizer is trained on a
database including different levels of background noises and is used
as a universal recognizer for every environment. In order to enhance
the system robustness by reducing the environmental variability, we
explore different adaptation algorithms including Maximum Likelihood
Linear Regression (MLLR), Maximum A Posteriori (MAP)
and the MAP/MLLR algorithm that combines MAP and MLLR.
Experimental evaluation shows that a rather good recognition rate
can be reached, even under important noise degradation conditions
when the system is fed by the convenient set of features.
Design as Contract and Blueprint – Tackling Maturity Level 1 Software Vendors in an e-School Project
Process improvements have drawn much attention in
practical software engineering. The capability maturity levels from
CMMI have become an important index to assess a software company-s
software engineering capability. However, in countries like
Taiwan, customers often have no choices but to deal with vendors that
are not CMMI prepared or qualified. We call these vendors maturitylevel-
1 (ML1) vendors. In this paper, we describe our experience
from consulting an e-school project. We propose an approach to help
our client tackle the ML1 vendors. Through our system analysis, we
produce a design. This design is suggested to be used as part of
contract and a blueprint to guide the implementation.
UB-Tree Indexing for Semantic Query Optimization of Range Queries
Semantic query optimization consists in restricting the
search space in order to reduce the set of objects of interest for a
query. This paper presents an indexing method based on UB-trees
and a static analysis of the constraints associated to the views of the
database and to any constraint expressed on attributes. The result of
the static analysis is a partitioning of the object space into disjoint
blocks. Through Space Filling Curve (SFC) techniques, each
fragment (block) of the partition is assigned a unique identifier,
enabling the efficient indexing of fragments by UB-trees. The search
space corresponding to a range query is restricted to a subset of the
blocks of the partition. This approach has been developed in the
context of a KB-DBMS but it can be applied to any relational
Union is Strength in Lossy Image Compression
In this work, we present a comparison between
different techniques of image compression. First, the image is
divided in blocks which are organized according to a certain scan.
Later, several compression techniques are applied, combined or
alone. Such techniques are: wavelets (Haar's basis), Karhunen-Loève
Transform, etc. Simulations show that the combined versions are the
best, with minor Mean Squared Error (MSE), and higher Peak Signal
to Noise Ratio (PSNR) and better image quality, even in the presence
Humanoid Personalized Avatar Through Multiple Natural Language Processing
There has been a growing interest in implementing humanoid avatars in networked virtual environment. However, most existing avatar communication systems do not take avatars- social backgrounds into consideration. This paper proposes a novel humanoid avatar animation system to represent personalities and facial emotions of avatars based on culture, profession, mood, age, taste, and so forth. We extract semantic keywords from the input text through natural language processing, and then the animations of personalized avatars are retrieved and displayed according to the order of the keywords. Our primary work is focused on giving avatars runtime instruction from multiple natural languages. Experiments with Chinese, Japanese and English input based on the prototype show that interactive avatar animations can be displayed in real time and be made available online. This system provides a more natural and interesting means of human communication, and therefore is expected to be used for cross-cultural communication, multiuser online games, and other entertainment applications.
Dynamic Decompression for Text Files
Compression algorithms reduce the redundancy in
data representation to decrease the storage required for that data.
Lossless compression researchers have developed highly
sophisticated approaches, such as Huffman encoding, arithmetic
encoding, the Lempel-Ziv (LZ) family, Dynamic Markov
Compression (DMC), Prediction by Partial Matching (PPM), and
Burrows-Wheeler Transform (BWT) based algorithms.
Decompression is also required to retrieve the original data by
lossless means. A compression scheme for text files coupled with
the principle of dynamic decompression, which decompresses only
the section of the compressed text file required by the user instead of
decompressing the entire text file. Dynamic decompressed files offer
better disk space utilization due to higher compression ratios
compared to most of the currently available text file formats.
A High Bitrate Information Hiding Algorithm for Video in Video
In high bitrate information hiding techniques, 1 bit is
embedded within each 4 x 4 Discrete Cosine Transform (DCT)
coefficient block by means of vector quantization, then the hidden bit
can be effectively extracted in terminal end. In this paper high bitrate
information hiding algorithms are summarized, and the scheme of
video in video is implemented. Experimental result shows that the host
video which is embedded numerous auxiliary information have little
visually quality decline. Peak Signal to Noise Ratio (PSNR)Y of host
video only degrades 0.22dB in average, while the hidden information
has a high percentage of survives and keeps a high robustness in
H.264/AVC compression, the average Bit Error Rate(BER) of hiding
information is 0.015%.
An Adaptive Hand-Talking System for the Hearing Impaired
An adaptive Chinese hand-talking system is presented
in this paper. By analyzing the 3 data collecting strategies for new
users, the adaptation framework including supervised and unsupervised
adaptation methods is proposed. For supervised adaptation,
affinity propagation (AP) is used to extract exemplar subsets, and enhanced
maximum a posteriori / vector field smoothing (eMAP/VFS)
is proposed to pool the adaptation data among different models. For
unsupervised adaptation, polynomial segment models (PSMs) are
used to help hidden Markov models (HMMs) to accurately label
the unlabeled data, then the "labeled" data together with signerindependent
models are inputted to MAP algorithm to generate
signer-adapted models. Experimental results show that the proposed
framework can execute both supervised adaptation with small amount
of labeled data and unsupervised adaptation with large amount
of unlabeled data to tailor the original models, and both achieve
improvements on the performance of recognition rate.