• Speech Sciences
• /
• v.14 no.1
• /
• pp.93-103
• /
• 2007

The aims of this study were to examine the acoustic features of Korean and Japanese vowels, and to determine whether new phones that do not have counterparts in Japanese or similar phones that have counterparts improve more from learning. This study consisted of three parts. In Experiment I, a speech production test was performed to observe the acoustic features of Korean and Japanese vowels. In Experiment II, the speech production of Korean vowels produced by Koreans, advanced Japanese learners of Korean, and beginning Japanese learners of Korean was investigated. In Experiment III, a speech perception study of Korean vowels produced by the two Japanese learner groups was conducted to observe the effect of learning on acquiring L2 phones. The conclusion drawn from the study was that the similar phones produced by Japanese show more similarity with those of Koreans than new phones in terms of F1 and F2, but Japanese learners of Korean displayed more improvement in new phones from learning.

• Journal of Korea Multimedia Society
• /
• v.23 no.4
• /
• pp.595-602
• /
• 2020

We propose a generative probabilistic model with Dirichlet prior distribution for topic modeling and text similarity analysis. It assigns a topic and calculates text correlation between documents within a corpus. It also provides posterior probabilities that are assigned to each topic of a document based on the prior distribution in the corpus. We then present a Gibbs sampling algorithm for inference about the posterior distribution and compute text correlation among 50 abstracts from the papers published by IEEE. We also conduct a supervised learning to set a benchmark that justifies the performance of the LDA (Latent Dirichlet Allocation). The experiments show that the accuracy for topic assignment to a certain document is 76% for LDA. The results for supervised learning show the accuracy of 61%, the precision of 93% and the f1-score of 96%. A discussion for experimental results indicates a thorough justification based on probabilities, distributions, evaluation metrics and correlation coefficients with respect to topic assignment.

• The Transactions of the Korea Information Processing Society
• /
• v.7 no.3
• /
• pp.1025-1031
• /
• 2000

Pre-existing pattern recognition techniques, in the case of character recognition, have limited on the application field. But CAI character learning system and writer's recognition system are very important parts. The application field of pre-existing system can be expanded in the content that the learning of characters and the recognition of writers in the proposed paper. In order to achieve these goals, the development contents are the following: Firstly, pre-processing method by understanding the image structure is proposed, secondly, recognition of characters are accomplished b the histogram distribution characteristics. Finally, similarity measure functions are defined from standard character pattern for matching of the input character pattern. Also the effectiveness of this system is demonstrated by experimenting the standard primitive image.

• International Journal of Advanced Culture Technology
• /
• v.7 no.1
• /
• pp.64-69
• /
• 2019

With the advent of the fourth industry, the need for office automation with automatic character recognition capabilities is increasing day by day. Therefore, in this paper, we study a character recognition algorithm that effectively recognizes a new experimental data character by using learning data characters. The proposed algorithm computes the degree of similarity that the structural regions of learning data characters match the corresponding regions of the experimental data character. It has been confirmed that satisfactory results can be obtained by selecting the learning data character with the highest degree of similarity in the matching process as the final recognition result for a given experimental data character.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.13 no.5
• /
• pp.1043-1050
• /
• 2018

In this paper, we propose a reinforce learning based on cooperative sensing scheme to select optimal secondary users(SUs) to enhance the detection performance of spectrum sensing in Cognitive radio(CR) networks. The SU with high accuracy is identified based on the similarity between the global sensing result obtained through cooperative sensing and the local sensing result of the SU. A fusion center(FC) uses similarity of SUs as reward value for Q-learning to determine SUs which participate in cooperative sensing with accurate sensing results. The experimental results show that the proposed method improves the detection performance compared to conventional cooperative sensing schemes.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.2
• /
• pp.121-129
• /
• 2021

This paper deals with a super-resolution that improves the resolution of side scan sonar images using learning-based compressive sensing. Learning-based compressive sensing combined with deep learning and compressive sensing takes a structure of a feed-forward network and parameters are set automatically through learning. In particular, we propose a method that can effectively extract additional information required in the super-resolution process through various initialization methods. Representative experimental results show that the proposed method provides improved performance in terms of Peak Signal-to-Noise Ratio (PSNR) and Structure Similarity Index Measure (SSIM) than conventional methods.

• Journal of the Semiconductor & Display Technology
• /
• v.20 no.3
• /
• pp.113-116
• /
• 2021

As artificial intelligence technologies, including deep learning, develop, these technologies are being introduced to code similarity analysis. In the traditional analysis method of calculating the graph edit distance (GED) after converting the source code into a control flow graph (CFG), there are studies that calculate the GED through a trained graph neural network (GNN) with the converted CFG, Methods for analyzing code similarity through CNN by imaging CFG are also being studied. In this paper, to determine which approach will be effective and efficient in researching code similarity analysis methods using artificial intelligence in the future, code similarity is measured through funcGNN, which measures code similarity using GNN, and Siamese Network, which is an image similarity analysis model. The accuracy was compared and analyzed. As a result of the analysis, the error rate (0.0458) of the Siamese network was bigger than that of the funcGNN (0.0362).

• Phonetics and Speech Sciences
• /
• v.7 no.4
• /
• pp.49-55
• /
• 2015

It is argued in this paper that, in determining semantic similarity, Korean words should be recategorized with a focus on the semantic relation to ontology in light of cross-linguistic morphological variations. It is proposed, in particular, that Korean semantic similarity should be measured on three tracks, human judgements track, relatedness track, and cross-part-of-speech relations track. As demonstrated in Yang et al. (2015), GloVe, the unsupervised learning machine on semantic similarity, is applicable to Korean with its performance being compared with human judgement results. Based on this compatability, it was further thought that the model's performance might most likely vary with different kinds of specific relations in different languages. An attempt was made to analyze them in terms of two major Korean-specific categories involved in their lexical and cross-POS-relations. It is concluded that languages must be analyzed by varying methods so that semantic components across languages may allow varying semantic distance in the vector space models.

• Journal of the Korean Society for information Management
• /
• v.29 no.2
• /
• pp.155-171
• /
• 2012

In text categorization, core terms of an input document are hardly selected as classification features if they do not occur in a training document set. Besides, synonymous terms with the same concept are usually treated as different features. This study aims to improve text categorization performance by integrating synonyms into a single feature and by replacing input terms not in the training document set with the most similar term occurring in training documents using Wikipedia. For the selection of classification features, experiments were performed in various settings composed of three different conditions: the use of category information of non-training terms, the part of Wikipedia used for measuring term-term similarity, and the type of similarity measures. The categorization performance of a kNN classifier was improved by 0.35~1.85% in $F_1$ value in all the experimental settings when non-learning terms were replaced by the learning term with the highest similarity above the threshold value. Although the improvement ratio is not as high as expected, several semantic as well as structural devices of Wikipedia could be used for selecting more effective classification features.

• Journal of Science Education
• /
• v.47 no.2
• /
• pp.180-191
• /
• 2023

The purpose of this study was to investigate the preferred SSI topics of students and teachers in elementary, middle, and high schools. It analyzed the similarity of students' and teachers' preferred SSI topics by school level using the cosine similarity measure. A total of 566 students and 327 teachers from elementary, middle, and high schools participated in the study. Sixty topics were identified and listed in the areas of environment, science and technology, health and medicine, and other social issues based on the literature and SSI programs. Students and teachers were asked to select five of their favorite topics. The data was collected online using SurveyMonkey. The collected data was divided into six groups of students and teachers, and the frequency of topic selection was analyzed within each group. The topic preference similarity was analyzed by calculating vector values based on the frequency of the selected topics and measuring the cosine similarity between students, teachers, and teachers and students by school level. The results are as follows: First, the cosine similarity of SSI Preferred Topics between students' school-level cohorts was higher between middle and high school students (0.982) than between elementary and middle school students (0.651) or between elementary and high school students (0.662). Second, the cosine similarity of SSI Preferred Topics between teachers' school-level cohorts was similar for all comparison groups between elementary, middle, and high school. Third, the SSI topic preference similarity between students and teachers by school level had a higher cosine similarity between the elementary student and teacher cohorts (0.974) than the other school level comparisons, middle school (0.621) or high school (0.645). Access to topics of interest to students in SSI education is strongly associated with motivation and persistence in learning, as well as an enjoyable learning experience and positive attitudes toward learning. Therefore, when designing SSI lessons, it is important to examine topics from the perspective of student interest, especially if the teacher has selected SSI topics that are different from students' preferences. Careful instructional design will be needed to overcome the gap.