• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.628-629
• /
• 2022

딥러닝의 높은 예측 정확도를 위해서는 많은 양의 학습 데이터가 필요하다. 그러나 실세계에서 많은 양의 레이블이 붙은 데이터를 구하는 것은 어렵고 많은 비용이 든다. 때문에 레이블이 없이도 양질의 표현 학습이 가능한 자기지도학습이 각광을 받고 있다. 와서스타인 거리는 생성모델에도 쓰이지만 의사 레이블 (pseudo label) 을 만들어 레이블이 없는 데이터들을 분류 하는데도 좋은 성능을 보이고 있다. 따라서. 본 연구는 와서스타인 거리를 자기지도학습에 접목시키는 방법을 제안한다. 실험을 통해 연구의 가능성을 보인다.

• Journal of KIISE:Software and Applications
• /
• v.35 no.9
• /
• pp.547-553
• /
• 2008

We propose a two-stage document layout segmentation method. At the first stage, as top-down segmentation, morphological distance map algorithm extracts a collection of rectangular regions from a given input image. This preliminary result from the first stage is employed as input parameters for the process of next stage. At the second stage, a machine-learning algorithm is adopted RBF network, one of neural networks based on statistical model, is selected. In order for constructing the hidden layer of RBF network, a data clustering technique bared on the self-organizing property of Kohonen network is utilized. We present a result showing that the supervised neural network, trained by 300 number of sample data, improves the preliminary results of the first stage.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.4
• /
• pp.157-163
• /
• 2022

Recently, with the development of automation in the industrial field, research on anomaly detection is being actively conducted. An application for anomaly detection used in factory automation is camera-based defect inspection. Vision camera inspection shows high performance and efficiency in factory automation, but it is difficult to overcome the instability of lighting and environmental conditions. Although camera inspection using deep learning can solve the problem of vision camera inspection with much higher performance, it is difficult to apply to actual industrial fields because it requires a huge amount of normal and abnormal data for learning. Therefore, in this study, we propose a network that overcomes the problem of collecting abnormal data with 72 geometric transformation deep learning methods using only normal data and adds an outlier exposure method for performance improvement. By applying and verifying this to the MVTec data set, which is a database for auto-mobile parts data and outlier detection, it is shown that it can be applied in actual industrial sites.

• Korean Journal of Artificial Intelligence
• /
• v.11 no.4
• /
• pp.9-13
• /
• 2023

In this study, we explored the potential of integrating interactive AI callbot technology into the medical consultation domain as part of a broader service development initiative. Aimed at enhancing patient satisfaction, the AI callbot was designed to efficiently address queries from hospitals' primary users, especially the elderly and those using phone services. By incorporating an AI-driven callbot into the hospital's customer service center, routine tasks such as appointment modifications and cancellations were efficiently managed by the AI Callbot Agent. On the other hand, tasks requiring more detailed attention or specialization were addressed by Human Agents, ensuring a balanced and collaborative approach. The deep learning model for voice recognition for this study was based on the Transformer model and fine-tuned to fit the medical field using a pre-trained model. Existing recording files were converted into learning data to perform SSL(self-supervised learning) Model was implemented. The ANN (Artificial neural network) neural network model was used to analyze voice signals and interpret them as text, and after actual application, the intent was enriched through reinforcement learning to continuously improve accuracy. In the case of TTS(Text To Speech), the Transformer model was applied to Text Analysis, Acoustic model, and Vocoder, and Google's Natural Language API was applied to recognize intent. As the research progresses, there are challenges to solve, such as interconnection issues between various EMR providers, problems with doctor's time slots, problems with two or more hospital appointments, and problems with patient use. However, there are specialized problems that are easy to make reservations. Implementation of the callbot service in hospitals appears to be applicable immediately.

• Journal of Periodontal and Implant Science
• /
• v.48 no.2
• /
• pp.114-123
• /
• 2018

Purpose: The aim of the current study was to develop a computer-assisted detection system based on a deep convolutional neural network (CNN) algorithm and to evaluate the potential usefulness and accuracy of this system for the diagnosis and prediction of periodontally compromised teeth (PCT). Methods: Combining pretrained deep CNN architecture and a self-trained network, periapical radiographic images were used to determine the optimal CNN algorithm and weights. The diagnostic and predictive accuracy, sensitivity, specificity, positive predictive value, negative predictive value, receiver operating characteristic (ROC) curve, area under the ROC curve, confusion matrix, and 95% confidence intervals (CIs) were calculated using our deep CNN algorithm, based on a Keras framework in Python. Results: The periapical radiographic dataset was split into training (n=1,044), validation (n=348), and test (n=348) datasets. With the deep learning algorithm, the diagnostic accuracy for PCT was 81.0% for premolars and 76.7% for molars. Using 64 premolars and 64 molars that were clinically diagnosed as severe PCT, the accuracy of predicting extraction was 82.8% (95% CI, 70.1%-91.2%) for premolars and 73.4% (95% CI, 59.9%-84.0%) for molars. Conclusions: We demonstrated that the deep CNN algorithm was useful for assessing the diagnosis and predictability of PCT. Therefore, with further optimization of the PCT dataset and improvements in the algorithm, a computer-aided detection system can be expected to become an effective and efficient method of diagnosing and predicting PCT.

• Journal of Intelligence and Information Systems
• /
• v.27 no.1
• /
• pp.65-82
• /
• 2021

Stock market investors are generally split into foreign investors, institutional investors, and individual investors. Compared to individual investor groups, professional investor groups such as foreign investors have an advantage in information and financial power and, as a result, foreign investors are known to show good investment performance among market participants. The purpose of this study is to propose an investment strategy that combines investor-specific transaction information and machine learning, and to analyze the portfolio investment performance of the proposed model using actual stock price and investor-specific transaction data. The Korea Exchange offers daily information on the volume of purchase and sale of each investor to securities firms. We developed a data collection program in C# programming language using an API provided by Daishin Securities Cybosplus, and collected 151 out of 200 KOSPI stocks with daily opening price, closing price and investor-specific net purchase data from January 2, 2007 to July 31, 2017. The self-organizing map model is an artificial neural network that performs clustering by unsupervised learning and has been introduced by Teuvo Kohonen since 1984. We implement competition among intra-surface artificial neurons, and all connections are non-recursive artificial neural networks that go from bottom to top. It can also be expanded to multiple layers, although many fault layers are commonly used. Linear functions are used by active functions of artificial nerve cells, and learning rules use Instar rules as well as general competitive learning. The core of the backpropagation model is the model that performs classification by supervised learning as an artificial neural network. We grouped and transformed investor-specific transaction volume data to learn backpropagation models through the self-organizing map model of artificial neural networks. As a result of the estimation of verification data through training, the portfolios were rebalanced monthly. For performance analysis, a passive portfolio was designated and the KOSPI 200 and KOSPI index returns for proxies on market returns were also obtained. Performance analysis was conducted using the equally-weighted portfolio return, compound interest rate, annual return, Maximum Draw Down, standard deviation, and Sharpe Ratio. Buy and hold returns of the top 10 market capitalization stocks are designated as a benchmark. Buy and hold strategy is the best strategy under the efficient market hypothesis. The prediction rate of learning data using backpropagation model was significantly high at 96.61%, while the prediction rate of verification data was also relatively high in the results of the 57.1% verification data. The performance evaluation of self-organizing map grouping can be determined as a result of a backpropagation model. This is because if the grouping results of the self-organizing map model had been poor, the learning results of the backpropagation model would have been poor. In this way, the performance assessment of machine learning is judged to be better learned than previous studies. Our portfolio doubled the return on the benchmark and performed better than the market returns on the KOSPI and KOSPI 200 indexes. In contrast to the benchmark, the MDD and standard deviation for portfolio risk indicators also showed better results. The Sharpe Ratio performed higher than benchmarks and stock market indexes. Through this, we presented the direction of portfolio composition program using machine learning and investor-specific transaction information and showed that it can be used to develop programs for real stock investment. The return is the result of monthly portfolio composition and asset rebalancing to the same proportion. Better outcomes are predicted when forming a monthly portfolio if the system is enforced by rebalancing the suggested stocks continuously without selling and re-buying it. Therefore, real transactions appear to be relevant.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.3
• /
• pp.389-395
• /
• 2022

Because of improvement of deep learning techniques, deep learning using computer vision such as classification, detection and segmentation has also been used widely at many fields. Expecially, automatic driving is one of the major fields that applies computer vision systems. Also there are a lot of works and researches to combine multiple tasks in a single network. In this study, we propose the network that predicts the individual depth of pedestrians and vehicles. Proposed model is constructed based on YOLOv3 for object detection and Monodepth for depth estimation, and it process object detection and depth estimation consequently using encoder and decoder based on hard parameter sharing. We also used attention module to improve the accuracy of both object detection and depth estimation. Depth is predicted with monocular image, and is trained using self-supervised training method.

• The KIPS Transactions:PartB
• /
• v.10B no.3
• /
• pp.311-320
• /
• 2003

The Optimal filter yielding optimal texture feature separation is a most effective technique for extracting the texture objects from multiple textures images. But, most optimal filter design approaches are restricted to the issue of supervised problems. No full-unsupervised method is based on the recognition of texture objects in image. We propose a novel approach that uses unsupervised learning schemes for efficient texture image analysis, and the band-pass feature of Gabor-filter is used for the optimal filter design. In our approach, the self-organizing neural network for multiple texture image identification is based on block-based clustering. The optimal frequency of Gabor-filter is turned to the optimal frequency of the distinct texture in frequency domain by analyzing the spatial frequency. In order to show the performance of the designed filters, after we have attempted to build a various texture images. The texture objects extraction is achieved by using the designed Gabor-filter. Our experimental results show that the performance of the system is very successful.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.09b
• /
• pp.71-75
• /
• 2003

오류 역전파 학습 알고리즘을 이용하여 영상 인식에 적용 할 경우에는 은닉층의 노드 수를 경험적으로 설정하므로, 학습시간과 지역최소화 및 정체현상이 발생한다. 그리고 ARTI 알고리즘은 입력 패턴과 저장 패턴간의 측정 방법인 유사성 검증 방법과 경계 변수의 설정에 따라 인식률이 좌우된다. 경계 변수의 값이 크면 입력 패턴과 저장 패턴사이에 약간의 차이만 있어도 새로운 카테고리(Category)로 분류하고, 반대로 경계 변수의 값이 적으면 입력 패턴과 저장 패턴 사이에 많은 차이가 있더라도 유사성이 인정되어 입력 패턴들을 대략적으로 분류한다. 따라서 ART1 알고리즘을 영상 인식에 적용하기 위해서는 경계 변수를 경험적으로 설정하므로 인식률에 부정적인 영향을 갖는 문제점이 있다. 따라서 본 논문에서는 개선된 ART1 알고리즘과 지도 학습 방법을 결합하여 신경망의 은닉층 노드를 동적으로 변화시키는 자가 생성지도 학습 알고리즘을 제안한다. 제안된 신경망에서 입력층과 은닉층의 학습 구조에는 ART1 알고리즘을 개선하여 적용하고, 은닉층과 출력층의 학습 구조에는 은닉층에서 승자로 선택된 노드와 출력층 노드와 연결된 가중치만을 조정하고 Delta-Bar-Delta 알고리즘을 적용한다. 제안된 방법의 학습 성능을 분석하기 위하여 학생증 영상에서 추출한 학번 패턴 분류에 적용한 결과, 기존의 신경망 학습 알고리즘보다 학습 성능이 개선됨을 확인하였다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2006.05a
• /
• pp.313-318
• /
• 2006

본 논문에서는 ART-1 기반 자가 생성 지도학습 알고리즘과 얼굴 인증을 이용한 주민등록증 인식방법을 제안한다. 본 논문에서는 주민등록증 영상에서 주민등록번호와 발행일을 추출하기 위해, 획득된 주민등록증의 영상에서 Sobel Mask와 Median Filter를 이용하여 윤곽선을 추출하고 잡음을 제거한 후, 수평 스미어링을 적용하여 주민등록번호와 발행일 영역을 각각 추출한다. 그리고 고주파 필터링을 적용하여 추출된 영역을 이진화하고 4방향 윤곽선 추적 알고리즘을 적용하여 개별 코드를 추출한다. 추출된 개별 코드는 ART-1 기반 자가 생성 지도학습 알고리즘을 적용하여 인식한다. 얼굴 인증은 Template Matching 방법을 적용하여 Face Template Database를 구축하고, 획득된 주민등록증의 얼굴 영역과의 유사도를 측정하여 주민등록증의 사진 위조 여부를 판별한다. 제안된 주민등록증 인식 방법의 성능을 평가하기 위해 10개의 주민등록증을 대상으로 실험하였고 원본 주민등록증 영상에서 사진과 얼굴 부분을 위조한 주민등록증에 대해 얼굴 인증 실험을 하였다. 실험을 통해 제안된 방법이 주민등록번호 인식 및 얼굴 인증에 있어서 우수한 성능이 있음을 확인하였다.