• The Korean Journal of Applied Statistics
• /
• v.34 no.6
• /
• pp.905-922
• /
• 2021

Cluster analysis is one of unsupervised learning techniques used for discovering clusters when there is no prior knowledge of group membership. K-means, one of the commonly used cluster analysis techniques, may fail when the number of variables becomes large. In such high-dimensional cases, it is common to perform tandem analysis, K-means cluster analysis after reducing the number of variables using dimension reduction methods. However, there is no guarantee that the reduced dimension reveals the cluster structure properly. Principal component analysis may mask the structure of clusters, especially when there are large variances for variables that are not related to cluster structure. To overcome this, techniques that perform dimension reduction and cluster analysis simultaneously have been suggested. This study proposes probabilistic reduced K-means, the transition of reduced K-means (De Soete and Caroll, 1994) into a probabilistic framework. Simulation shows that the proposed method performs better than tandem clustering or clustering without any dimension reduction. When the number of the variables is larger than the number of samples in each cluster, probabilistic reduced K-means show better formation of clusters than non-probabilistic reduced K-means. In the application to a real data set, it revealed similar or better cluster structure compared to other methods.

• Database Research
• /
• v.34 no.3
• /
• pp.137-147
• /
• 2018

A key to a well-performing human activity recognition (HAR) system through machine learning technique is the availability of a substantial amount of labeled data. Collecting sufficient labeled data is an expensive and time-consuming task. To build a HAR system in a new environment (i.e., the target domain) with very limited labeled data, it is unfavorable to naively exploit the data or trained classifier model from the existing environment (i.e., the source domain) as it is due to the domain difference. While traditional machine learning approaches are unable to address such distribution mismatch, transfer learning approach leverages the utilization of knowledge from existing well-established source domains that help to build an accurate classifier in the target domain. In this work, we propose a transfer learning approach to create an accurate HAR classifier with very limited data through the multitask neural network. The classifier loss function minimization for source and target domain are treated as two different tasks. The knowledge transfer is performed by simultaneously minimizing the loss function of both tasks using a single neural network model. Furthermore, we utilize the unlabeled data in an unsupervised manner to help the model training. The experiment result shows that the proposed work consistently outperforms existing approaches.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2024.05a
• /
• pp.576-579
• /
• 2024

경제 위기 대비를 위해 인공지능을 활용한 주식시장 변동성 이상을 탐지하는 목적을 가지고 있다. 글로벌 이슈와 경제 위기 대비를 위해 주식시장 변동성 예측의 중요성이 부각되고 있으며, 기존의 주식시장 변동성 지수인 VIX 의 한계로 인해 더 복잡한 모델 및 인공지능을 활용한 연구에 관심이 집중되고 있다. 기존의 주식시장 변동성 예측에 관한 연구들은 통계적인 방법을 사용했으며 인공지능을 이용한 연구 또한 대부분 이상치 구간을 표시하여 예측을 목표로 하고 있으나 이러한 접근법은 라벨이 있는 데이터 수집 어려움, 클래스 불균형 문제가 있다. 본 연구는 인공지능을 활용한 주식시장 변동성 탐지에 기여하고 지도 학습 방식 대신 비지도 학습 기반의 이상탐지모델을 사용하여 주식시장 변동성을 예측하는 새로운 방법론을 제안한다. 본 연구에서 개발한 인공지능 모델은 IsolationForest 모델을 활용하며, 시계열 데이터를 전처리한 후 정상성을 확보하는 등의 과정을 거친다. 실험 결과로 인공지능 모델이 주요 경제이슈를 이상치로 검출하는 성능을 확인하였으며 재현율 약 93.6%, 정밀도 100%로 높은 성능을 달성했다.

• The KIPS Transactions:PartB
• /
• v.15B no.1
• /
• pp.61-68
• /
• 2008

This paper presents a spam filter system using co-training with spamminess features and URL features based on the maximum entropy model. Spamminess features are the emphasizing patterns or abnormal patterns in spam messages used by spammers to express their intention and to avoid being filtered by the spam filter system. Since spammers use URLs to give the details and make a change to the URL format not to be filtered by the black list, normal and abnormal URLs can be key features to detect the spam messages. Co-training with spamminess features and URL features uses two different features which are independent each other in training. The filter system can learn information from them independently. Experiment results on TREC spam test collection shows that the proposed approach achieves 9.1% improvement and 6.9% improvement in accuracy compared to the base system and bogo filter system, respectively. The result analysis shows that the proposed spamminess features and URL features are helpful. And an experiment result of the co-training shows that two feature sets are useful since the number of training documents are reduced while the accuracy is closed to the batch learning.

• Journal of Navigation and Port Research
• /
• v.42 no.4
• /
• pp.277-282
• /
• 2018

Recently, the Vessel Traffic Service (VTS) coverage has expanded to include coastal areas following the increased attention on vessel traffic safety. However, it has increased the workload on the VTS operators. In some cases, when the traffic volume increases sharply during the rush hour, the VTS operator may not be aware of the risks. Therefore, in this paper, we proposed a new method to recognize ship movement anomalies automatically to support the VTS operator's decision-making. The proposed method generated traffic pattern model without any category information using the unsupervised learning algorithm.. The anomaly score can be calculated by classification and comparison of the trained model. Finally, we reviewed the experimental results using a ship-handling simulator and the actual trajectory data to verify the feasibility of the proposed method.

• Journal of Korea Society of Industrial Information Systems
• /
• v.25 no.2
• /
• pp.19-27
• /
• 2020

Recently, rapid and accurate 3D models creation is required in various applications using virtual reality and augmented reality technology. In this paper, we propose an on-site learning based shape deformation method which transforms the clothed 3D human model into the shape of an input point cloud. The proposed algorithm consists of two main parts: one is pre-learning and the other is on-site learning. Each learning consists of encoder, template transformation and decoder network. The proposed network is learned by unsupervised method, which uses the Chamfer distance between the input point cloud form and the template vertices as the loss function. By performing on-site learning on the input point clouds during the inference process, the high accuracy of the inference results can be obtained and presented through experiments.

• Journal of Internet Computing and Services
• /
• v.23 no.3
• /
• pp.35-45
• /
• 2022

Even in recent years, treatment of first-aid patients is still often delayed due to a shortage of medical resources in marginalized areas. Research on automating the analysis of medical data to solve the problems of inaccessibility for medical services and shortage of medical personnel is ongoing. Computer vision-based medical inspection automation requires a lot of cost in data collection and labeling for training purposes. These problems stand out in the works of classifying lesion that are rare, or pathological features and pathogenesis that are difficult to clearly define visually. Anomaly detection is attracting as a method that can significantly reduce the cost of data collection by adopting an unsupervised learning strategy. In this paper, we propose methods for detecting abnormal images on chest X-RAY images as follows based on existing anomaly detection techniques. (1) Normalize the brightness range of medical images resampled as optimal resolution. (2) Some feature vectors with high representative power are selected in set of patch features extracted as intermediate-level from lesion-free images. (3) Measure the difference from the feature vectors of lesion-free data selected based on the nearest neighbor search algorithm. The proposed system can simultaneously perform anomaly classification and localization for each image. In this paper, the anomaly detection performance of the proposed system for chest X-RAY images of PA projection is measured and presented by detailed conditions. We demonstrate effect of anomaly detection for medical images by showing 0.705 classification AUROC for random subset extracted from the PadChest dataset. The proposed system can be usefully used to improve the clinical diagnosis workflow of medical institutions, and can effectively support early diagnosis in medically poor area.

• Journal of Advanced Navigation Technology
• /
• v.12 no.5
• /
• pp.502-508
• /
• 2008

With the increased use of the internet and the tremendous amount of data it transfers, it is more necessary to summarize documents. We propose a new method using the Non-negative Semantic Variable Matrix (NSVM) and the generic important of semantic features obtained by Non-negative Matrix Factorization (NMF) to extract the sentences for automatic generic summarization. The proposed method use non-negative constraints which is more similar to the human's cognition process. As a result, the proposed method selects more meaningful sentences for summarization than the unsupervised method used the Latent Semantic Analysis (LSA) or clustering methods. The experimental results show that the proposed method archives better performance than other methods.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• v.9 no.2
• /
• pp.370-376
• /
• 2005

본 논문에서는 HSI 정보와 신경망의 비지도 학습 방법인 ART2 알고리즘을 이용하여 신 차량 번호판을 인식하는 방법을 제안한다. 제안된 방법은 차량의 영상에서 번호판 영역을 추출하는 부분과 추출된 번호판 영역의 문자를 인식하는 부분으로 구성된다. 본 논문에서는 차량 번호판 영역을 추출하기 위해 HSI 컬러 모형의 Hue 정보를 이용하여 차량 번호판 영역을 추출하고 개선된 퍼지 이진화 방법을 적용하여 추출된 차량 번호판 영역으로부터 문자를 포함한 특징 영역을 이치화 한 후에 4방향 윤곽선 추적 알고리즘을 적용하여 개별 코드를 추출한다. 추출된 개별 코드를 인식하기 위해 잡음과 훼손에 비교적 강한 ART2 알고리즘을 적용한다. 제안된 방법의 차량 번호판 추출 및 인식 성능을 평가하기 위하여 실제 비영업용 차량 번호판에 적용한 결과, 기존의 차량 번호판의 추출 방법보다 번호판 영역의 추출률이 개선되었다. 또한 ART2 알고리즘을 적용하여 신 차량 번호판을 인식하는 것이 효율적임을 확인하였다.

• Journal of KIISE:Software and Applications
• /
• v.37 no.5
• /
• pp.378-385
• /
• 2010

The explosive increase in the use of email has made to need email classification efficiently and accurately. Current work on the email classification method have mainly been focused on a binary classification that filters out spam-mails. This methods are based on Support Vector Machines, Bayesian classifiers, rule-based classifiers. Such supervised methods, in the sense that the user is required to manually describe the rules and keyword list that is used to recognize the relevant email. Other unsupervised method using clustering techniques for the multi-category classification is created a category labels from a set of incoming messages. In this paper, we propose a new automatic email multi-category classification method using NMF for automatic category label construction method and dynamic category hierarchy method for the reorganization of email messages in the category labels. The proposed method in this paper, a large number of emails are managed efficiently by classifying multi-category email automatically, email messages in their category are reorganized for enhancing accuracy whenever users want to classify all their email messages.