• Convergence Security Journal
• /
• v.23 no.4
• /
• pp.61-70
• /
• 2023

Clustering is an unsupervised learning method that involves grouping data based on features such as distance metrics, using data without known labels or ground truth values. This method has the advantage of being applicable to various types of data, including images, text, and audio, without the need for labeling. Traditional clustering techniques involve applying dimensionality reduction methods or extracting specific features to perform clustering. However, with the advancement of deep learning models, research on deep clustering techniques using techniques such as autoencoders and generative adversarial networks, which represent input data as latent vectors, has emerged. In this study, we propose a deep clustering technique based on deep learning. In this approach, we use an autoencoder to transform the input data into latent vectors, and then construct a vector space according to the cluster structure and perform k-means clustering. We conducted experiments using the MNIST and Fashion-MNIST datasets in the PyTorch machine learning library as the experimental environment. The model used is a convolutional neural network-based autoencoder model. The experimental results show an accuracy of 89.42% for MNIST and 56.64% for Fashion-MNIST when k is set to 10.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.32 no.4
• /
• pp.691-708
• /
• 2022

As the technology of machine learning and deep learning became common, it began to be applied to research on anomaly(abnormal) detection of industrial control systems. In Korea, the HAI dataset was developed and published to activate artificial intelligence research for abnormal detection of industrial control systems, and an AI contest for detecting industrial control system security threats is being conducted. Most of the anomaly detection studies have been to create a learning model with improved performance through the ensemble model method, which is applied either by modifying the existing deep learning algorithm or by applying it together with other algorithms. In this study, a study was conducted to improve the performance of anomaly detection with a post-processing method that detects abnormal data and corrects the labeling results, rather than the learning algorithm and data pre-processing process. Results It was confirmed that the results were improved by about 10% or more compared to the anomaly detection performance of the existing model.

• Journal of Advanced Navigation Technology
• /
• v.22 no.5
• /
• pp.472-477
• /
• 2018

In this paper, we develop interactive content services for preventing depression of users through an intelligent Internet of Things(IoT) mirror system. For interactive content services, an IoT mirror device measures attention and meditation data from an EEG headset device and also measures facial expression data such as "sad", "angery", "disgust", "neutral", " happy", and "surprise" classified by a multi-layer perceptron algorithm through an webcam. Then, it sends the measured data to an oneM2M-compliant IoT server. Based on the collected data in the IoT server, a machine learning model is built to classify three levels of depression (RED, YELLOW, and GREEN) given by a proposed merge labeling method. It was verified that the k-nearest neighbor (k-NN) model could achieve about 93% of accuracy by experimental results. In addition, according to the classified level, a social network service agent sent a corresponding alert message to the family, friends and social workers. Thus, we were able to provide an interactive content service between users and caregivers.

• KIPS Transactions on Software and Data Engineering
• /
• v.4 no.7
• /
• pp.283-290
• /
• 2015

In this paper, we propose a hybrid method for the recognition of answer types in the WiseQA system. The answer types are classified into two categories: the lexical answer type (LAT) and the semantic answer type (SAT). This paper proposes two models for the LAT detection. One is a rule-based model using question focuses. The other is a machine learning model based on sequence labeling. We also propose two models for the SAT classification. They are a machine learning model based on multiclass classification and a filtering-rule model based on the lexical answer type. The performance of the LAT detection and the SAT classification shows F1-score of 82.47% and precision of 77.13%, respectively. Compared with IBM Watson for the performance of the LAT, the precision is 1.0% lower and the recall is 7.4% higher.

• Korean Journal of Culture and Social Issue
• /
• v.28 no.4
• /
• pp.661-690
• /
• 2022

Adolescent delinquency is a substantial social problem that occurs in both offline and online domains. The current study utilized random forest algorithms to identify predictors of adolescents' online and offline delinquency. Further, we explored the applicability of classic delinquency theories (social learning, strain, social control, routine activities, and labeling theory). We used the first-grade and fourth-grade elementary school panels as well as the first-grade middle school panel (N=4,137) among the sixth wave of the nationally-representative Korean Children and Youth Panel Survey 2010 for analysis. Random forest algorithms were used instead of the conventional regression analysis to improve the predictive performance of the model and possibly consider many predictors in the model. Random forest algorithm results showed that classic delinquency theories designed to explain offline delinquency were also applicable to online delinquency. Specifically, salient predictors of online delinquency were closely related to individual factors(routine activities and labeling theory). Social factors(social control and social learning theory) were particularly important for understanding offline delinquency. General strain theory was the commonly important theoretical framework that predicted both offline and online delinquency. Findings may provide evidence for more tailored prevention and intervention strategies against offline and online adolescent delinquency.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.6
• /
• pp.700-705
• /
• 2020

Generative adversarial networks are methods of generating images by opposing two neural networks. When generating the image, randomly generated noise is rearranged to generate the image. The image generated by this method is not generated well depending on the noise, and it is difficult to generate a proper image when the number of pixels of the image is small In addition, the speed and size of data accumulation in data classification increases, and there are many difficulties in labeling them. In this paper, to solve this problem, we propose a technique to generate noise based on random noise using real data. Since the proposed system generates an image based on the existing image, it is confirmed that it is possible to generate a more natural image, and if it is used for learning, it shows a higher hit rate than the existing method using the hostile neural network respectively.

• Clinical and Experimental Pediatrics
• /
• v.63 no.3
• /
• pp.88-95
• /
• 2020

Accurate localization of the seizure onset zone is important for better seizure outcomes and preventing deficits following epilepsy surgery. Recent advances in neuroimaging techniques have increased our understanding of the underlying etiology and improved our ability to noninvasively identify the seizure onset zone. Using epilepsy-specific magnetic resonance imaging (MRI) protocols, structural MRI allows better detection of the seizure onset zone, particularly when it is interpreted by experienced neuroradiologists. Ultra-high-field imaging and postprocessing analysis with automated machine learning algorithms can detect subtle structural abnormalities in MRI-negative patients. Tractography derived from diffusion tensor imaging can delineate white matter connections associated with epilepsy or eloquent function, thus, preventing deficits after epilepsy surgery. Arterial spin-labeling perfusion MRI, simultaneous electroencephalography (EEG)-functional MRI (fMRI), and magnetoencephalography (MEG) are noinvasive imaging modalities that can be used to localize the epileptogenic foci and assist in planning epilepsy surgery with positron emission tomography, ictal single-photon emission computed tomography, and intracranial EEG monitoring. MEG and fMRI can localize and lateralize the area of the cortex that is essential for language, motor, and memory function and identify its relationship with planned surgical resection sites to reduce the risk of neurological impairments. These advanced structural and functional imaging modalities can be combined with postprocessing methods to better understand the epileptic network and obtain valuable clinical information for predicting long-term outcomes in pediatric epilepsy.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.602-605
• /
• 2021

최근 인공지능 연구가 활발히 진행되고 있는 가운데 국내외에서 오픈 데이터셋을 제공하고 있어 기술개발이 가속화되고 있다. 데이터셋은 지도학습을 위한 학습데이터로 라벨링 데이터를 포함하고 있어 다양한 라벨링 기능이 적용된 도구 개발이 필요하다. 본 논문에서는 의료영상의 라벨링 데이터를 정교하고 빠르게 생성하기 위한 라벨링 웹 애플리케이션에 대해서 기술한다. 이를 구현하기 위해서 Back Projection, Grabcut 기법을 이용한 반자동 방식과 기계학습 모델을 통해서 예측한 자동 방식의 라벨링 기능을 구현하였다. 이와 관련하여 라벨링 기능별 수행 결과를 근감소증 진단을 위한 영상 라벨링 수행결과와 정량분석 결과를 보였다.

• Journal of Navigation and Port Research
• /
• v.47 no.6
• /
• pp.358-366
• /
• 2023

This study aimed to provide a solution for improving ship collision alert of the 'accident vulnerable ship monitoring service' among the 'intelligent marine traffic information system' services of the Ministry of Oceans and Fisheries. The current ship collision alert uses a supervised learning (SL) model with survey labels based on large ship-oriented data and its operators. Consequently, the small ship data and the operator's opinion are not reflected in the current collision-supervised learning model, and the effect is insufficient because the alarm is provided from a longer distance than the small ship operator feels. In addition, the supervised learning (SL) method requires a large number of labeled data, and the labeling process requires a lot of resources and time. To overcome these limitations, in this paper, the classification model of collision alerts for small ships using unlabeled data with the semi-supervised learning (SSL) algorithms (Label Propagation and TabNet) was studied. Results of real-time experiments on small ship operators using the classification model of collision alerts showed that the satisfaction of operators increased.

• Journal of KIISE:Computer Systems and Theory
• /
• v.33 no.4
• /
• pp.231-241
• /
• 2006

Creating controllable, responsive avatars is an important problem in computer games and virtual environments. Recently, large collections of motion capture data have been exploited for increased realism in avatar animation and control. Large motion sets have the advantage of accommodating a broad variety of natural human motion. However, when a motion set is large, the time required to identify an appropriate sequence of motions is the bottleneck for achieving interactive avatar control. In this paper, we present a novel method for training avatar behaviors from unlabelled motion data in order to animate and control avatars at minimal runtime cost. Based on machine learning technique, called Q-teaming, our training method allows the avatar to learn how to act in any given situation through trial-and-error interactions with a dynamic environment. We demonstrate the effectiveness of our approach through examples that include avatars interacting with each other and with the user.