• Journal of the Korea Society of Computer and Information
• /
• v.22 no.12
• /
• pp.63-68
• /
• 2017

In this paper, we propose a bounding box prediction algorithm using multiple probability maps to improve object detection result of object detector. Although the performance of object detectors has been significantly improved, it is still not perfect due to technical problems and lack of learning data. Therefore, we use the result correction method to obtain more accurate object detection results. In the proposed algorithm, the preprocessed bounding box created as a result of object detection by the object detector is clustered in various form, and a conditional probability is given to each cluster to make multiple probability map. Finally, multiple probability map create new bounding box of object using morphological elements. Experiment results show that the newly predicted bounding box reduces the error in ground truth more than 45% on average compared to the previous bounding box.

• Nuclear Engineering and Technology
• /
• v.50 no.4
• /
• pp.582-588
• /
• 2018

Accident diagnosis is one of the complex tasks for nuclear power plant (NPP) operators. In abnormal or emergency situations, the diagnostic activity of the NPP states is burdensome though necessary. Numerous computer-based methods and operator support systems have been suggested to address this problem. Among them, the recurrent neural network (RNN) has performed well at analyzing time series data. This study proposes an algorithm for accident diagnosis using long short-term memory (LSTM), which is a kind of RNN, which improves the limitation for time reflection. The algorithm consists of preprocessing, the LSTM network, and postprocessing. In the LSTM-based algorithm, preprocessed input variables are calculated to output the accident diagnosis results. The outputs are also postprocessed using softmax to determine the ranking of accident diagnosis results with probabilities. This algorithm was trained using a compact nuclear simulator for several accidents: a loss of coolant accident, a steam generator tube rupture, and a main steam line break. The trained algorithm was also tested to demonstrate the feasibility of diagnosing NPP accidents.

• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.522-522
• /
• 2000

In this paper, two pre-filtering techniques are presented for accurately estimating the impact location of a loose part. The reason why a pre-filterng technique Is necessary in a Loose Part Monitoring System is that the effects of background noise on the signal to noise ratio (SNR) can be reduced considerably resulting in improved estimation accuracy. The first method is to take d moving average operation in the time domain. The second one is to adopt band-pass filters designed in the frequency domain such as a Butterworth filter, Chebyshev filter I & II and an Elliptic Filter. To show the effectiveness, the impact test data (signals) from the YGN3 power plant are first preprocessed and then used to estimate the loose pan impact position. Resultantly. we observed that SNR is much improved and the average estimation error is below 7.5%.

• Proceedings of the KOSOMBE Conference
• /
• v.1998 no.11
• /
• pp.205-206
• /
• 1998

In this paper, we present the detection algorithm of interictal epileptic waveform using LVQ network and wavelet transform. First wavelet coefficients is used to represent the characteristics of a single channel EEG wave, and make a number of neural network input node smaller. Then, three-layer neural network employing LVQ network is trained and tested using parameters obtained from the first stage. This study showed that preprocessed EEG data can be successfully used to train ANNs to detect epileptogenic discharges with a high success.

• Proceedings of the IEEK Conference
• /
• 2006.06a
• /
• pp.423-424
• /
• 2006

This paper describes a 3D face recognition using different devices for 3D faces and input faces which include several different pose. Before the recognition stage, through the EC-SVD, all data have to be preprocessed and normalized. At recognition stage, we propose the multi-point signature method for measuring facial surface information. And we use the root mean square error for matching. From the experiment results, we have 92.5% recognition rate.

• ETRI Journal
• /
• v.41 no.5
• /
• pp.560-573
• /
• 2019

Two supervised learning algorithms, a basic neural network and a long short-term memory recurrent neural network, are applied to traffic including DDoS attacks. The joint effects of preprocessing methods and hyperparameters for machine learning on performance are investigated. Values representing attack characteristics are extracted from datasets and preprocessed by two methods. Binary classification and two optimizers are used. Some hyperparameters are obtained exhaustively for fast and accurate detection, while others are fixed with constants to account for performance and data characteristics. An experiment is performed via TensorFlow on three traffic datasets. Three scenarios are considered to investigate the effects of learning former traffic on sequential traffic analysis and the effects of learning one dataset on application to another dataset, and determine whether the algorithms can be used for recent attack traffic. Experimental results show that the used preprocessing methods, neural network architectures and hyperparameters, and the optimizers are appropriate for DDoS attack detection. The obtained results provide a criterion for the detection accuracy of attacks.

• Journal of Information Processing Systems
• /
• v.18 no.5
• /
• pp.628-636
• /
• 2022

Automatic program repair (APR) techniques focus on automatically repairing bugs in programs and providing correct patches for developers, which have been investigated for decades. However, most studies have limitations in repairing complex bugs. To overcome these limitations, we developed an approach that augments datasets by utilizing token tagging and applying machine learning techniques for APR. First, to alleviate the data insufficiency problem, we augmented datasets by extracting all the methods (buggy and non-buggy methods) in the program source code and conducting token tagging on non-buggy methods. Second, we fed the preprocessed code into the model as an input for training. Finally, we evaluated the performance of the proposed approach by comparing it with the baselines. The results show that the proposed approach is efficient for augmenting datasets using token tagging and is promising for APR.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.26 no.4
• /
• pp.310-322
• /
• 2022

In this report, we apply an anomaly detection algorithm to a mobile oral health care application. In particular, we have investigated one class YOLOv3 as an anomaly detection model to classify pictures of mouths which will be used as inputs in the following machine learning model. We have achieved outstanding performances by proposing appropriate annotation strategies for our data sets and modifying the loss function. Moreover, the model can classify not only oral and non-oral pictures but also output preprocessed pictures that only contain the area around the lips by using the predicted bounding box. Thus, the model performs prediction and preprocessing simultaneously.

• Journal of Information Processing Systems
• /
• v.18 no.3
• /
• pp.311-318
• /
• 2022

Tobacco sales enterprises often need to summarize and verify the daily sales bills, which may consume substantial manpower, and manual verification is prone to occasional errors. The use of artificial intelligence technology to realize the automatic identification and verification of such bills offers important practical significance. This study presents a novel multi-branch residual network for tobacco sales bills to improve the efficiency and accuracy of tobacco sales. First, geometric correction and edge alignment were performed on the input sales bill image. Second, the multi-branch residual network recognition model is established and trained using the preprocessed data. The comparative experimental results demonstrated that the correct recognition rate of the proposed method reached 98.84% on the China Tobacco Bill Image dataset, which is superior to that of most existing recognition methods.

• Journal of Digital Contents Society
• /
• v.19 no.2
• /
• pp.377-383
• /
• 2018

Internet of Things (IoT) is producing various data as the smart environment comes. The IoT data collection is used as important data to judge systems's status. Therefore, it is important to monitor the anomaly state of the sensor in real-time and to detect anomaly data. However, it is necessary to convert the IoT data into a normalized data structure for anomaly detection because of the variety of data structures and protocols. Thus, we can expect a good quality effect such as accurate analysis data quality and service quality. In this paper, we propose an anomaly detection system based on big data from collected sensor data. The proposed system is applied to ensure anomaly detection and keep data quality. In addition, we applied the machine learning model of support vector machine using anomaly detection based on time-series data. As a result, machine learning using preprocessed data was able to accurately detect and predict anomaly.