• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.38B no.12
• /
• pp.954-961
• /
• 2013

In this paper, we propose a new learning method using a variable learning to improve pattern recognition in the FCSR(Fast Commit Slow Recode) learning method of the Fuzzy ART. Traditional learning methods have used a fixed learning rate in updating weight vector(representative pattern). In the traditional method, the weight vector will be updated with a fixed learning rate regardless of the degree of similarity of the input pattern and the representative pattern in the category. In this case, the updated weight vector is greatly influenced from the input pattern where it is on the boundary of the category. Thus, in noisy environments, this method has a problem in increasing unnecessary categories and reducing pattern recognition capacity. In the proposed method, the lower similarity between the representative pattern and input pattern is, the lower input pattern contributes for updating weight vector. As a result, this results in suppressing the unnecessary category proliferation and improving pattern recognition capacity of the Fuzzy ART in noisy environments.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.618-620
• /
• 2023

최근 자연어처리 분야에서 의미론적 유사성을 반영하기 위한 대조 학습 (contrastive learning) 관련 연구가 활발히 이뤄지고 있다. 이러한 대조 학습의 핵심은 의미론적으로 가까워져야 하는 쌍과 멀어져야 하는 쌍을 잘 구축하는 것이지만, 기존의 손실 함수는 문장의 상대적인 유사성을 풍부하게 반영하는데 한계가 있다. 이를 해결하기 위해, 이전 연구에서는 삼중 항 손실 함수 (triplet loss)를 도입하였으며, 본 논문에서는 이러한 삼중 항을 구성하기 위해 대조 학습에서의 효과적인 토큰 컷오프(cutoff) 데이터 증강 기법을 제안한다. BERT, RoBERTa 등 널리 활용되는 언어 모델을 이용한 실험을 통해 제안하는 방법의 우수한 성능을 보인다.

• Journal of the Korea Convergence Society
• /
• v.9 no.10
• /
• pp.21-28
• /
• 2018

To help users who are experiencing difficulties finding the right learning course corresponding to their level of proficiency, we developed a recommendation model for personalized learning course for Intelligence Tutoring System(ITS). The Personalized Learning Course Recommendation model for ITS analyzes the learner profile and extracts the keyword by calculating the weight of each word. The similarity of vector between extracted words is measured through the cosine similarity method. Finally, the three courses of top similarity are recommended for learners. To analyze the effects of the recommendation model, we applied the recommendation model to the Women's ability development center. And mean, standard deviation, skewness, and kurtosis values of question items were calculated through the satisfaction survey. The results of the experiment showed high satisfaction levels in accuracy, novelty, self-reference and usefulness, which proved the effectiveness of the recommendation model. This study is meaningful in the sense that it suggested a learner-centered recommendation system based on machine learning, which has not been researched enough both in domestic, foreign domains.

• Phonetics and Speech Sciences
• /
• v.12 no.1
• /
• pp.11-17
• /
• 2020

Non-tonal language speakers may have difficulty learning second language lexical tones. In the present study, we explored this issue with Korean-speaking learners of Mandarin Chinese (i.e., non-tonal first language speakers) by examining their perception and production of Mandarin lexical tones. In the perception experiment, the Korean learners were asked to listen to the tone of each stimulus and assign it to one of four Mandarin lexical tones using the response keys; in the production experiment, the learners provided speech production data for the lexical tones and then their productions were identified by native listeners of Mandarin Chinese. Our results showed that the Korean learners of Mandarin Chinese had difficulty in perceptually distinguishing Tone 2 and Tone 3, with the most frequent production error being the mispronunciation of Tone 3 as Tone 2. We also investigated whether unfamiliar non-native phonemes (i.e., Chinese phonemes) that do not exist in the native language phonemic inventory (i.e., Korean) may hinder the processing of the non-native lexical tones. We found no evidence for such effects, neither for the perception nor for the production of the tones.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.18 no.3
• /
• pp.449-458
• /
• 2023

Many people think that sign language is only for people who are deaf and cannot speak, but of course it is necessary for people who want to talk with them. One of the biggest challenges in ASL(American Sign Language) alphabet recognition is the high inter-class similarities and high intra-class variance. In this paper, we proposed an architecture that can overcome these two problems, which performs similarity learning to reduces inter-class similarities and intra-class variance between images. The proposed architecture consists of the same convolutional neural network with a double configuration that shares parameters (weights and biases) and also applies the Keras API to reduce similarity learning and variance through this pathway. The similarity learning results the use of the dual CNN shows that the accuracy is improved by reducing the similarity and variability between classes by not including the poor results of the two classes.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.7
• /
• pp.127-135
• /
• 2011

It is required that an e-learning system has a content recommendation component which helps a learner choose an item. In order to predict items concerning learner's interest, collaborative filtering and content-based filtering methods have been most widely used. The methods recommend items for a learner based on other learner's interests without considering the knowledge level of the learner. So, the effectiveness of the recommendation can be reduced when the number of overall users are relatively small. Also, it is not easy to recommend a newly added item. In order to address the problem, we propose a content recommendation method based on the similarity and the difficulty of an item. By using a recommendation function that reflects both characteristics of items, a higher-level leaner can choose more difficult but less similar items, while a lower-level learner can select less difficult but more similar items, Thus, a learner can be presented items according to his or her level of achievement, which is irrelevant to other learner's interest.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.442-444
• /
• 2023

그래프 임베딩 방법은 그래프 구조를 이용하여 그래프의 노드를 저차원 임베딩 공간에서 벡터로 매핑하여 각 노드를 벡터로 표현하는 것을 목표로 한다. 다양한 방법들이 제시되었지만 기존의 방법들은 그래프에서 노드 간의 유사성을 잘 보존할 수 없어 다양한 기계 학습에 대해 부정확한 벡터를 생성하였다. 이러한 문제를 해결하기 위해 노드 사이의 유사성을 이용한 방법이 제안되었다. 본 논문에서, 우리는 여섯 가지 실세계 데이터셋을 사용하여 세 가지 기계 학습 작업시 그래프 임베딩 방법들의 성능을 비교하여 유사성 기반의 그래프 임베딩 방법의 우수성을 확인했다.

• Proceedings of the Korean Association for Survey Research Conference
• /
• 2002.06a
• /
• pp.115-133
• /
• 2002

최근 우리사회는 제반운영기반이 지식에 의존하고 있고 지식의 폭발적인 증가와 더불어 지식의 생명주기 단축이 매우 빠름을 특징으로 하고 있다. 지식의 중요성은 생활의 곳곳에까지도 영향을 미치게 되므로 많은 사람들이 지식사회에 부응하는 학습의 필요성을 절실히 느끼고 있지만 학습내용이 다양하게 개발되어 있지 않고, 또한 어느 학습에 어떻게 참여할지 모르는 경우도 있다. 교육인적자원부에서는 평생학습과 관련된 $\ulcorner$국가인적자원개발 기본계획$\lrcorner$(2001. 12. 7.)과 연계된 평생학습분야의 구체적 실행계획을 마련하기도 하였다. 본 연구에서는 고령화사회에서의 평생학습에 대한 수요는 계속적으로 증가할 것으로 전망하여 “사회통계조사”의 평생학습 참여실태를 검토하였고 그룹간 유사성 모델을 살펴 본 후 향후에는 평생교육이 어디에 중점을 두어야하는지를 밝혔다.

• Korean Journal of Cognitive Science
• /
• v.12 no.4
• /
• pp.29-39
• /
• 2001

Kaplan(2000) reported that instances were categorized more accurate in the aligned condition than in the non-aligned condition irrespective of similarity between instances[16]. This study investigated wether Kaplan(2000)\\`s results could be explained by stimulus types she used and alignment effects in categorization were due to selective attention to aligned attributes. In Experiment 1. I examined whether attribute alignment produced significant effects on similarity and categorization and aligned attributes were recalled more than non-aligned ones. Results showed that instances were rated more similar and categories were learned more rapidly in the aligned condition than in the non-aligned condition. It can be explained that categories are learned rapidly in the aligned condition because attribute alignment increases within-category similarity. But. the result that aligned attributes were recalled more than non-aliened ones in the attribute recall test implies that alignment effects in categorization can be independent of similarity between instances partially. In Experiment 2. I used equal numbed of attributes defining two categories and instructed subjects to pay their attention to categorization-relevant dimensions only. Results showed that dimension instruction facilitated category learning in the non-aligned condition only but categories were learned more rapidly in the aligned condition than in the non-aliened condition irrespective of instruction types. In conclusion. attribute alignment in categorization may facilitate paying selective attention to categorization-relevant attributes.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.3
• /
• pp.51-55
• /
• 2020

In this paper, an improved FAM is proposed by adopting similarity learning in the existing FAM (Fuzzy Associative Memory) used in image restoration. Image restoration refers to the recovery of the latent clean image from its noise-corrupted version. In serious application like face recognition, this process should be noise-tolerant, robust, fast, and scalable. The existing FAM is a simple single layered neural network that can be applied to this domain with its robust fuzzy control but has low capacity problem in real world applications. That similarity measure is implied to the connection strength of the FAM structure to minimize the root mean square error between the recovered and the original image. The efficacy of the proposed algorithm is verified with significant low error magnitude from random noise in our experiment.