• The Korean Journal of Applied Statistics
• /
• 제37권2호
• /
• pp.191-207
• /
• 2024

Many studies have been conducted to quickly detect out-of-control situations in autocorrelated processes. The most traditionally used method is a residual control chart, which uses residuals calculated from a fitted time series model. However, many procedures for monitoring autocorrelated processes using statistical learning methods have recently been proposed. In this paper, we propose a monitoring procedure using the latent vector of LSTM Autoencoder, a deep learning-based unsupervised learning method. We compare the performance of this procedure with the LSTM Autoencoder procedure based on the reconstruction error, the RNN classification procedure, and the residual charting procedure through simulation studies. Simulation results show that the performance of the proposed procedure and the RNN classification procedure are similar, but the proposed procedure has the advantage of being useful in processes where sufficient out-of-control data cannot be obtained, because it does not require out-of-control data for training.

• Journal of Christian Education in Korea
• /
• 제63권
• /
• pp.255-281
• /
• 2020

The purpose of this study was to investigate the structural relationship among self-directed learning ability, learner-instructor interaction, learner-learner interaction, and class satisfaction in online learning environments by the structural equation modelling (SEM). Participants of the study consisted of 300 students (110 = high school students, 190 = college students). Through latent mean analysis (LMA), there was no significant difference of study variables between high school and college groups. However, thorough multi-group analysis, self-directed learning ability had a direct and indirect effect on class satisfaction for the college group via learner-instructor and learner-learner interactions, while learner-learner interaction played a full mediating role of the relationship between self-directed learning ability and class satisfaction for the high school group. In addition, self-directed learning ability had a stronger influence on learner-learner interaction for the college group than the high school group. These results would provide important implications for understanding the different mechanisms between high school and college online learning contexts.

• Proceedings of the Korea Information Processing Society Conference
• /
• 한국정보처리학회 2002년도 추계학술발표논문집 (상)
• /
• pp.409-412
• /
• 2002

본 논문에서는 실제 경영의 의사결정 등을 위한 활용가치가 있는 정보를 추출해 내는 방법론으로 SOM을 적용하였다. SOM은 자율(upsupervised)과 경쟁(competitive) 학습을 한다. 데이터를 입력하였을 때, SOM의 출력 노드중에서 다른 출력 노드과 비교해서 가장 강하게 반응하는 노드가 있을 것이며, 그러한 출력 노드를 더욱 더 강하게 반응하게끔 반복적으로 학습시키는 것이다. 입력에 대해 자연스럽게 반응하는 출력 노드를 선택하여 반복 학습을 시키면, 후에는 결과적으로 어떤 출력 노드가 반응되는지를 조사하면 거꾸로 입력을 알 수 있게 되는 것이다. 대량의 데이터, 잠재적으로 활용가치가 있는 데이터를 SOM을 통해 유용한 정보들을 추출할 수 있으며 이는 실제 경영의 의사결정을 위한 수단으로 충분히 활용될 수 있을 것이다.

• Annual Conference on Human and Language Technology
• /
• 한국정보과학회언어공학연구회 2010년도 제22회 한글 및 한국어 정보처리 학술대회
• /
• pp.127-129
• /
• 2010

본 논문에서는 다양한 코퍼스를 통해 언어를 학습하는 과정을 모델링하여 무감독학습(Unsupervised learning)으로 문법적 패턴을 습득하는 방법론을 소개한다. 제안 방법에서는 적은 수의 특성 조합으로 잠재적 패턴의 부분만을 표현한 후 표현된 규칙을 조합하여 유의미한 문법적 패턴을 탐색한다. 본 논문에서 제안한 방법은 베이지만 추론(Bayesian Inference)과 MCMC (Markov Chain Mote Carlo) 샘플링에 기반하여 특성 조합을 유의미한 문법적 패턴으로 정제하는 방법으로, 랜덤하이퍼그래프(Random Hypergraph) 모델을 이용하여 많은 수의 하이퍼에지를 생성한 후 생성된 하이퍼에지의 가중치를 조정하여 유의미한 문법적 패턴을 탈색하는 방법론이다. 우리는 본 논문에서 유아용 비디오의 스크립트를 이용하여 다양한 유아용 비디오 스크립트에서 문법적 패턴을 습득하는 방법론을 소개한다.

• Journal of IKEEE
• /
• 제25권4호
• /
• pp.728-734
• /
• 2021

In recent years, there has been a strong interest in the end-to-end autonomous driving based on deep reinforcement learning. In this paper, we present a reward function of latent SAC deep reinforcement learning to improve the longitudinal driving performance of an agent vehicle. While the existing reward function significantly degrades the driving safety and efficiency, the proposed reward function is shown to maintain an appropriate headway distance while avoiding the front vehicle collision.

• Proceedings of the Korean Information Science Society Conference
• /
• 한국정보과학회 2007년도 한국컴퓨터종합학술대회논문집 Vol.34 No.1 (A)
• /
• pp.87-88
• /
• 2007

• Journal of Intelligence and Information Systems
• /
• 제29권1호
• /
• pp.249-263
• /
• 2023

Collaborative Filtering, a representative recommendation system methodology, consists of two approaches: neighbor methods and latent factor models. Among these, the latent factor model using matrix factorization decomposes the user-item interaction matrix into two lower-dimensional rectangular matrices, predicting the item's rating through the product of these matrices. Due to the factor vectors inferred from rating patterns capturing user and item characteristics, this method is superior in scalability, accuracy, and flexibility compared to neighbor-based methods. However, it has a fundamental drawback: the need to reflect the diversity of preferences of different individuals for items with no ratings. This limitation leads to repetitive and inaccurate recommendations. The Adaptive Deep Latent Factor Model (ADLFM) was developed to address this issue. This model adaptively learns the preferences for each item by using the item description, which provides a detailed summary and explanation of the item. ADLFM takes in item description as input, calculates latent vectors of the user and item, and presents a method that can reflect personal diversity using an attention score. However, due to the requirement of a dataset that includes item descriptions, the domain that can apply ADLFM is limited, resulting in generalization limitations. This study proposes a Generalized Adaptive Deep Latent Factor Recommendation Model, G-ADLFRM, to improve the limitations of ADLFM. Firstly, we use item ID, commonly used in recommendation systems, as input instead of the item description. Additionally, we apply improved deep learning model structures such as Self-Attention, Multi-head Attention, and Multi-Conv1D. We conducted experiments on various datasets with input and model structure changes. The results showed that when only the input was changed, MAE increased slightly compared to ADLFM due to accompanying information loss, resulting in decreased recommendation performance. However, the average learning speed per epoch significantly improved as the amount of information to be processed decreased. When both the input and the model structure were changed, the best-performing Multi-Conv1d structure showed similar performance to ADLFM, sufficiently counteracting the information loss caused by the input change. We conclude that G-ADLFRM is a new, lightweight, and generalizable model that maintains the performance of the existing ADLFM while enabling fast learning and inference.

• The Mathematical Education
• /
• 제50권1호
• /
• pp.103-118
• /
• 2011

The study examined whether the relation between mathematics self-efficacy and mathematics achievement was partially mediated by the learning strategies, using latent growth model analyses. It was also examined the auto-regressive, cross-lagged (ARCL) panel model for testing the stability and change in the relation of mathematics self-efficacy and learning strategy over time. The study analyzed the first-year to the third-year data of the Korean Educational Longitudinal Survey (KELS). The result of ARCL panel model analysis showed that earlier mathematics self-efficacy could predict later learning strategy use. There were linear trends in mathematics self-efficacy, learning strategy, and mathematics achievement. Specifically, mathematics achievement was increased over the three time points, whereas mathematics self-efficacy and learning strategies were significantly decreased. In the analyses of latent growth models, the mediating effects of learning strategies were overall supported. That is, both of initial status and change rate of rehearsal strategy partially mediated the relation of mathematics self-efficacy and mathematics achievement. However, in elaboration and meta-cognitive strategies, only the initial status of each variable showed the indirect relationship.

• Journal of Digital Contents Society
• /
• 제14권1호
• /
• pp.111-116
• /
• 2013

Instructive lessons in the language classrooms are extremely constrictive to language learners in improving their English skills, thus, the use of technology in education plays an important role in a language classroom. Results of language proficiency improvement of the learners vary depending on the extent of supplementary materials offered by the internet delivered to the learners. The purpose of the present study is to explore and propose a new approach of language acquisition offered by the internet. This study presents effective methods of using the internet in carrying out the written work. In this paper, we show that the relationship between participation of the learners in the class activities and improvement of English writing skills are mutually proportional.

• Annual Conference on Human and Language Technology
• /
• 한국정보과학회언어공학연구회 2022년도 제34회 한글 및 한국어 정보처리 학술대회
• /
• pp.284-288
• /
• 2022

최근 다양한 언어모델의 구축 및 발전으로 개체명 인식 시스템의 성능은 최고 수준에 도달했다. 하지만 이와 관련된 대부분의 연구는 데이터가 충분한 언어에 대해서만 다루기 때문에, 양질의 지도학습 데이터의 존재를 가정한다. 대부분의 언어에서는 개체 유형에 대한 언어의 잠재적 특성을 충분히 학습할 수 있는 지도학습 데이터가 부족하기 때문에, 종종 자원 부족의 어려움에 직면한다. 본 논문에서는 Masked language modeling 기반 언어 간 전이학습을 이용한 개체명 인식 방법론에 대한 분석을 수행한다. 이를 위해 전이를 수행하는 소스 언어는 고자원 언어로 가정하며, 전이를 받는 타겟 언어는 저자원 언어로 가정한다. 본 논문에서는 언어모델의 토큰 사전에 언어 독립적인 가상의 자질인 개체 유형에 대한 프롬프트 토큰을 추가하고 이를 소스 언어로 학습한 뒤, 타겟 언어로 전이하는 상황에서 제안하는 방법론에 대한 평가를 수행한다. 실험 결과, 제안하는 방법론은 일반적인 미세조정 방법론보다 높은 성능을 보였으며, 한국어에서 가장 큰 영향을 받은 타겟 언어는 네덜란드어, 한국어로 전이할 때 가장 큰 영향을 준 소스 언어는 중국어인 결과를 보였다.