• The KIPS Transactions:PartB
• /
• v.10B no.6
• /
• pp.673-680
• /
• 2003

In this paper, we study the enhancement of VQ (Vector Quantization) design for text independent speaker recognition. In a concrete way, we present a non-iterative method which makes a vector quantization codebook and this method performs non-iterative learning so that the computational complexity is epochally reduced The proposed Classified Space VQ (CSVQ) design method for text Independent speaker recognition is generalized from Semi-noniterative VQ design method for text dependent speaker recognition. CSVQ contrasts with the existing desiEn method which uses the iterative learninE algorithm for every traininE speaker. The characteristics of a CSVQ design is as follows. First, the proposed method performs the non-iterative learning by using a Classified Space Codebook. Second, a quantization region of each speaker is equivalent for the quantization region of a Classified Space Codebook. And the quantization point of each speaker is the optimal point for the statistical distribution of each speaker in a quantization region of a Classified Space Codebook. Third, Classified Space Codebook (CSC) is constructed through Sample Vector Formation Method (CSVQ1, 2) and Hyper-Lattice Formation Method (CSVQ 3). In the numerical experiment, we use the 12th met-cepstrum feature vectors of 10 speakers and compare it with the existing method, changing the codebook size from 16 to 128 for each Classified Space Codebook. The recognition rate of the proposed method is 100% for CSVQ1, 2. It is equal to the recognition rate of the existing method. Therefore the proposed CSVQ design method is, reducing computational complexity and maintaining the recognition rate, new alternative proposal and CSVQ with CSC can be applied to a general purpose recognition.

• The KIPS Transactions:PartB
• /
• v.9B no.3
• /
• pp.311-318
• /
• 2002

Document image layout analysis contains the process to segment document image into detailed regions and the process to classify the segmented regions into text, picture, table or etc. In the region classification process, the size of a region, the density of black pixels, and the complexity of pixel distribution are the bases of region classification. But in case of picture, the ranges of these bases are so wide that it's difficult to decide the classification threshold between picture and others. As a result, the picture has a higher region classification error than others. In this paper, we propose document image layout analysis method which has a better performance for the picture and text region classification than that of previous methods including commercial softwares. In the picture and text region classification, median filter is used in order to reduce the influence of the size of a region, the density of black pixels, and the complexity of pixel distribution. Futhermore the classification error is corrected by the use of region expanding filter and constrained conditions.

• The Journal of the Acoustical Society of Korea
• /
• v.18 no.1
• /
• pp.86-91
• /
• 1999

This paper presents the text-prompt method to overcome the weakness of text-dependent and text-independent speaker recognition. Enhanced dynamic time warping for speaker recognition algorithm is applied. For the real-time processing, we use a simple algorithm for end-point detection without increasing computational complexity. The test shows that the weighted-cepstrum is most proper for speaker recognition among various speech parameters. As the experimental results of the proposed algorithm for three prompt words, the speaker identification error rate is 0.02%, and when the threshold is set properly, false rejection rate is 1.89%, false acceptance rate is 0.77% and verification total error rate is 0.97% for speaker verification.

• Journal of KIISE:Software and Applications
• /
• v.28 no.10
• /
• pp.720-731
• /
• 2001

Complexity metrics have been developed for the structured paradigm of software development are not suitable for use with the object-oriented(OO) paradigm, because they do not support key object-oriented concepts such as inheritance, polymorphism. message passing and encapsulation. There are many researches on OO software metrics such as program complexity or design metrics. But metrics measuring the complexity of classes at the OO analysis phase are needed because they provide earlier feedback to the development project. and earlier feedback means more effective developing and less costly maintenance. In this paper, we propose the new metrics to measure the complexity of analysis classes which draw out in the analysis based on RUP(Rational Unified Process). By the collaboration complexity, is denoted by CC, we mean the maximum number of the collaborations can be achieved with each of the collaborator and determine the potential complexity. And the interface complexity, is denoted by IC, shows the difficulty related to understand the interface of collaborators each other. We verify theoretically the suggested metrics for Weyuker's nine properties. Moreover, we show the computation results for analysis classes of the system which automatically respond to questions of the user using the text mining technique. As a result of the comparison of CC and CBO and WMC suggested by Chidamber and Kemerer, the class that have highly the proposed metric value maintain the high complexity at the design phase too. And the complexity can be represented by CC and IC more than CBO and WMC. We can expect that our metrics may provide us the earlier feedback and hence possible to predict the efforts, costs and time required to remainder processes. As a result, we expect to develop the cost-effective OO software by reviewing the complexity of analysis classes in the first stage of SDLC(Software Development Life Cycle).

• The KIPS Transactions:PartD
• /
• v.8D no.1
• /
• pp.71-80
• /
• 2001

Object-Oriented (OO) methodology to use the concept like encapsulation, inheritance, polymorphism, and message passing demands metrics that are different from structured methodology. There are many studies for OO software metrics such as program complexity or design metrics. But the metrics for the analysis class need to decrease the complexity in the analysis phase so that greatly reduce the effort and the cost of system development. In this paper, we propose new metrics to measure the complexity of analysis classes which draw out in the analysis phase based on Unified Process. By the collaboration complexity, is denoted by CC, we mean the maximum number of the collaborations can be achieved with each of the collaborator and detennine the potential complexity. And the interface complexity, is denoted by IC, shows the difficulty related to understand the interface of collaborators each other. We prove mathematically that the suggested metrics satisfy OO characteristics such as class size and inheritance. And we verify it theoretically for Weyuker' s nine properties. Moreover, we show the computation results for analysis classes of the system which automatically respond to questions of the it's user using the text mining technique. As we compared CC and IC to CBO and WMC, the complexity can be represented by CC and IC more than CBO and WMC. We expect to develop the cost-effective OO software by reviewing the complexity of analysis classes in the first stage of SDLC (Software Development Life Cycle).

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.8 no.8
• /
• pp.1769-1776
• /
• 2004

Effective keyword extraction is important in the information search system and there are several ways to select proper keyword in many keywords. Among them, DER Structure for AC Algorithm to search single keyword, can search multiple keywords but it has time complexity problem. In this paper, we developed a algorithm, "EDER structure" by expanding standalone search table based on DER structure search method to improve time complexity. We tested the algorithm using 500 text files and found that EDER structure is more efficient than DER structure for AC for keyword posting result and time complexity that 0.2 second for EDER and 0.6 second for DER structure,structure,

• Journal of Korea Society of Digital Industry and Information Management
• /
• v.9 no.4
• /
• pp.103-110
• /
• 2013

In Information Technology era, various amazing IT technologies, for example Big Data, are appearing and are available as the amount of information increase. The number of counselling for violation of personal data protection is also increasing every year that it amounts to over 160,000 in 2012. According to Korean Privacy Act, in the case of treating unique personal identification information, appropriate measures like encipherment should be taken. The technologies of encipherment are the most basic countermeasures for personal data invasion and the base elements in information technology. So various cryptography algorithms exist and are used for encipherment technology. Therefore studies on safer new cryptography algorithms are executed. Cryptography algorithms started from classical replacement enciphering and developed to computationally secure code to increase complexity. Nowadays, various mathematic theories such as 'factorization into prime factor', 'extracting square root', 'discrete lognormal distribution', 'elliptical interaction curve' are adapted to cryptography algorithms. RSA public key cryptography algorithm which was based on 'factorization into prime factor' is the most representative one. This paper suggests algorithm utilizing telescoping series as a safer cryptography algorithm which can maximize the complexity. Telescoping series is a type of infinite series which can generate various types of function for given value-the plain text. Among these generated functions, one can be selected as a original equation. Some part of this equation can be defined as a key. And then the original equation can be transformed into final equation by improving the complexity of original equation through the command of "FullSimplify" of "Mathematica" software.

• Proceedings of the KSPS conference
• /
• 2005.11a
• /
• pp.103-106
• /
• 2005

In this paper, we propose a language modeling approach to improve the performance of a large vocabulary continuous speech recognition system. The proposed approach is based on the active learning framework that helps to select a text corpus from a plenty amount of text data required for language modeling. The perplexity is used as a measure for the corpus selection in the active learning. From the recognition experiments on the task of continuous Korean speech, the speech recognition system employing the language model by the proposed language modeling approach reduces the word error rate by about 6.6 % with less computational complexity than that using a language model constructed with randomly selected texts.

• Journal of Information Technology Services
• /
• v.17 no.1
• /
• pp.79-89
• /
• 2018

Sentiment analysis is a technique of text mining that extracts feelings of the person who wrote the sentence like movie review. The preliminary researches of sentiment analysis identify sentiments by using the dictionary which contains negative and positive words collected in advance. As researches on deep learning are actively carried out, sentiment analysis using deep learning model with morpheme or word unit has been done. However, this model has disadvantages in that the word dictionary varies according to the domain and the number of morphemes or words gets relatively larger than that of phonemes. Therefore, the size of the dictionary becomes large and the complexity of the model increases accordingly. We construct a sentiment analysis model using recurrent neural network by dividing input data into phoneme-level which is smaller than morpheme-level. To verify the performance, we use 30,000 movie reviews from the Korean biggest portal, Naver. Morpheme-level sentiment analysis model is also implemented and compared. As a result, the phoneme-level sentiment analysis model is superior to that of the morpheme-level, and in particular, the phoneme-level model using LSTM performs better than that of using GRU model. It is expected that Korean text processing based on a phoneme-level model can be applied to various text mining and language models.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.40 no.3
• /
• pp.194-202
• /
• 2003

To convert printed images into electronic ones automatically, it requires region analysis of document images and character recognition. In these, regional analysis segments document image into detailed regions and classifies thee regions into the types of text, picture, table and so on. But it is difficult to classify the text and the picture exactly, because the size, density and complexity of pixel distribution of some of these are similar. Thu, misclassification in region analysis is the main reason that makes automatic conversion difficult. In this paper, we propose region analysis method that segments document image into text and picture regions. The proposed method solves the referred problems using one dimensional median filter based method in text and picture classification. And the misclassification problems of boldface texts and picture regions like graphs or tables, caused by using median filtering, are solved by using of skin peeling filter and maximal text length. The performance, therefore, is better than previous methods containing commercial softwares.