• The Journal of the Acoustical Society of Korea
• /
• v.21 no.1E
• /
• pp.3-11
• /
• 2002

Broadcast news transcription is one of the hardest tasks in speech recognition because broadcast speech signals have much variability in speech quality, channel and background conditions. We developed a Korean broadcast news speech recognizer. We used a morpheme-based dictionary and a language model to reduce the out-of·vocabulary (OOV) rate. We concatenated the original morpheme pairs of short length or high frequency in order to reduce insertion and deletion errors due to short morphemes. We used a lexicon with multiple pronunciations to reflect inter-morpheme pronunciation variations without severe modification of the search tree. By using the merged morpheme as recognition units, we achieved the OOV rate of 1.7% comparable to European languages with 64k vocabulary. We implemented a hidden Markov model-based recognizer with vocal tract length normalization and online speaker adaptation by maximum likelihood linear regression. Experimental results showed that the recognizer yielded 21.8% morpheme error rate for anchor speech and 31.6% for mostly noisy reporter speech.

• Proceedings of the IEEK Conference
• /
• summer
• /
• pp.259-262
• /
• 2004

In most of the character recognition systems, the method of template matching or statistical method using hidden Markov model is used to extract and recognize feature shapes. In this paper, we used modified chain-code which has 8-directions but 4-codes, and made the chain-code of hand-written character, after that, converted it into transition chain-code by applying to HMM(Hidden Markov Model). The transition chain code by HMM is analyzed as signal flow graph by Mason's theory which is generally used to calculate forward gain at automatic control system. If the specific forward gain and feedback gain is properly set, the forward gain of transition chain-code using Mason's theory can be distinguished depending on each object for recognition. This data of the gain is reorganized as tree structure, hence making it possible to distinguish different hand-written characters. With this method, $91\%$ recognition rate was acquired.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2001.07a
• /
• pp.319-322
• /
• 2001

The proposed system was composed of pre-processor which was executing binary/high-pass filtering and post-processor which ranged from statistic data to prediction. In post-processor work, step one was filter process of image, step two was image recognition, and step three was destruction degree/time prediction. After these processing, we could predict image of the last destruction timestamp. This research was produced variation value according to growth of tree pattern. This result showed improved correction, when this research was applied image Processing. Pre-processing step of original image had good result binary work after high pass- filter execution. In the case of using partial discharge of the image, our research could predict the last destruction timestamp. By means of experimental data, this Prediction system was acquired ${\pm}$3.2% error range.

• Journal of the Korean Data and Information Science Society
• /
• v.18 no.1
• /
• pp.81-96
• /
• 2007

The headline of advertisements holds the important position in advertisement recognition and receptivity. In this paper, we classified the headlines as several types with measuring the response effects for purpose of understanding the customer's behavior. Also we conducted the decision tree analysis and correspondence analysis. The decision tree shows the difference between responders and non-responders, and the correspondence analysis shows the relationship between the sales and the customer's demographic information.

• The Journal of the Korea Contents Association
• /
• v.13 no.11
• /
• pp.562-570
• /
• 2013

Image of emotional pleasure or displeasure, tension or emotional division of tranquility in the form of a tree is evaluated by weighting. Image representative evaluation of the sensitivity of the brightness contrast ratings 1 car pleasure, displeasure or stress or emotional tranquility and two cars are separated by image segmentation. Emotion Recognition of four compared to the numerical data is measured by brightness. OpenCV implementation through evaluation graph the stress intensity contrast, tranquility, pleasure, displeasure, depending on changes in the value of the computing is divided into four emotional. Contrast sensitivity of computing the brightness depending on the value entered 'nuisance' to 'excellent' or 'stress' to 'calm' the emotional changes can give. Calculate the sensitivity of the image regularity of localized computing system can control the future direction of industry on the application of emotion recognition will play a positive role.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.14 no.4
• /
• pp.348-359
• /
• 1989

The multitree pattern recognition algorithm proposed by [1] and [2] is modified in order to improve its performance. The basic idea of the multitree pattern classification algorithm is that the binary dceision tree used to classify an unknow pattern is constructed for each feature and that at each stage, classification rule decides whether to classify the unknown pattern or to extract the feature value according to the feature ordet. So the feature ordering needed in the calssification procedure is simple and the number of features used in the classification procedure is small compared with other classification algorithms. Thus the algorithm can be easily applied to real pattern recognition problems even when the number of features and that of the classes are very large. In this paper, the wighting factor assignment scheme in the decision procedure is modified and various classification rules are proposed by means of the weighting factor. And the branch and bound method is applied to feature subset selection and feature ordering. Several experimental results show that the performance of the multitree pattern classification algorithm is improved by the proposed scheme.

• Culinary science and hospitality research
• /
• v.24 no.1
• /
• pp.130-139
• /
• 2018

The purpose of this study was to grasp the profitability of coffee franchise companies using ROIC tree, ROA and ROE. As a result, ROA was using assets efficiently, and some companies are in desperate need to improve their management based on past brand recognition, and a new paradigm management strategy is urgently required. In particular, the franchise company's current cost burden is greater than comparable firms. In ROE, the capital was operating efficiently. Since ROIC is very difficult to classify business related and operating related amounts accurately based on financial statement information, there is a limit to accurate ROIC calculation. Therefore, ROIC is estimated to be 7.6~38.29% as of 2016 based on ROIC calculation. In the case of continuous growth companies, investments are made to improve steady sales, but some companies seem to be unable to escape from past paradigm. In order to continue the growth of the company beyond the accounting profit in the future, the ROIC Tree can be used to measure the subdivided management performance and propose efficiency plan.

• Speech Sciences
• /
• v.8 no.4
• /
• pp.17-33
• /
• 2001

In this paper, we implement the large variable vocabulary speech recognition system, which is characterized by no additional pre-training process and no limitation of recognized word list. We have designed the system in order to achieve the high recognition rate using the decision tree based state tying algorithm and in order to reduce the processing time using the gaussian selection based variable flooring algorithm, the limitation algorithm of the number of nodes and ENNS algorithm. The gaussian selection based variable flooring algorithm shows that it can reduce the total processing time by more than half of the recognition time, but it brings about the reduction of recognition rate. In other words, there is a trade off between the recognition rate and the processing time. The limitation algorithm of the number of nodes shows the best performance when the number of gaussian mixtures is a three. Both of the off-line and on-line experiments show the same performance. In our experiments, there are some differences of the recognition rate and the average recognition time according to the distinction of genders, speakers, and the number of vocabulary.

• Journal of the Institute of Convergence Signal Processing
• /
• v.2 no.4
• /
• pp.9-15
• /
• 2001

In this paper we investigate context dependent acoustic models to improve the performance of the Korean speech recognition . The algorithm are using the Korean phonological rules and decision tree, By Successive State Splitting(SSS) algorithm the Hidden Merkov Netwwork(HM-Net) which is an efficient representation of phoneme-context-dependent HMMs, can be generated automatically SSS is powerful technique to design topologies of tied-state HMMs but it doesn't treat unknown contexts in the training phoneme contexts environment adequately In addition it has some problem in the procedure of the contextual domain. In this paper we adopt a new state-clustering algorithm of SSS, called Phonetic Decision Tree-based SSS (PDT-SSS) which includes contexts splits based on the Korean phonological rules. This method combines advantages of both the decision tree clustering and SSS, and can generated highly accurate HM-Net that can express any contexts To verify the effectiveness of the adopted methods. the experiments are carried out using KLE 452 word database and YNU 200 sentence database. Through the Korean phoneme word and sentence recognition experiments. we proved that the new state-clustering algorithm produce better phoneme, word and continuous speech recognition accuracy than the conventional HMMs.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1998.06a
• /
• pp.674-679
• /
• 1998

In this paepr, we proposed a new way to make a pre-pruned searching tree for GO game moves from macroscopic strategy described in linguistic expression. The strategy was a consequence of macroscopic recognition of GO game situations. The definitions of fuzzy macroscopic strategy, fuzzy tactics and tactical sequences using fuzzy set are shown and its family, so called "multi level fuzzy set". Some examples are also shown.