• 산업경영시스템학회지
• /
• 제47권1호
• /
• pp.9-19
• /
• 2024

Deep learning-based computer vision anomaly detection algorithms are widely utilized in various fields. Especially in the manufacturing industry, the difficulty in collecting abnormal data compared to normal data, and the challenge of defining all potential abnormalities in advance, have led to an increasing demand for unsupervised learning methods that rely on normal data. In this study, we conducted a comparative analysis of deep learning-based unsupervised learning algorithms that define and detect abnormalities that can occur when transparent contact lenses are immersed in liquid solution. We validated and applied the unsupervised learning algorithms used in this study to the existing anomaly detection benchmark dataset, MvTecAD. The existing anomaly detection benchmark dataset primarily consists of solid objects, whereas in our study, we compared unsupervised learning-based algorithms in experiments judging the shape and presence of lenses submerged in liquid. Among the algorithms analyzed, EfficientAD showed an AUROC and F1-score of 0.97 in image-level tests. However, the F1-score decreased to 0.18 in pixel-level tests, making it challenging to determine the locations where abnormalities occurred. Despite this, EfficientAD demonstrated excellent performance in image-level tests classifying normal and abnormal instances, suggesting that with the collection and training of large-scale data in real industrial settings, it is expected to exhibit even better performance.

• 한국학교수학회논문집
• /
• 제3권1호
• /
• pp.31-45
• /
• 2000

The idea to be changed first among the general ideas of mathematics is that, from the standpoint of teachers, one-sided teaching is efficient due to the progress of classwork of excessive quantity, and, from the standpoint of students, they say they gave up mathematics publicly feeling uneasiness in every class because of the deficiency of previous learning and interest, and they think all problems are due to the educational system, neglecting their studies, not participating actively in lessons. Therefore this study is, after giving learning points and problem papers of which the degree of difficulty differs to the groups of students divided into the advanced, intermediate, and elementary by personal abilities of learning and concerns, tried out to remove uniformity and integrity of lessons through solving the problems students can, to convert lessons to the learner-oriented that learners take part in voluntarily through discussing, solving the problems, and correcting the errors of them by group-cooperative actions, for the students to realize that they can solve everything only by their own endeavors.

• 한국지능시스템학회논문지
• /
• 제11권7호
• /
• pp.633-640
• /
• 2001

In this paper we examine the machine learning issues raised by the domain of the Intrusion Detection Systems(IDS), which have difficulty successfully classifying intruders. There systems also require a significant amount of computational overhead making it difficult to create robust real-time IDS. Machine learning techniques can reduce the human effort required to build these systems and can improve their performance. Genetic algorithms are used to improve the performance of search problems, while data mining has been used for data analysis. Data Mining is the exploration and analysis of large quantities of data to discover meaningful patterns and rules. Among the tasks for data mining, we concentrate the classification task. Since classification is the basic element of human way of thinking, it is a well-studied problem in a wide variety of application. In this paper, we propose a classifier system based on genetic algorithm, and the proposed system is evaluated by applying it to IDS problem related to classification task in data mining. We report our experiments in using these method on KDD audit data.

• 전자공학회논문지B
• /
• 제28B권12호
• /
• pp.27-37
• /
• 1991

In this paper, the novel PLD (Preceding Layer Driven) MLP (Multi Layer Perceptron) neural network model and its learning algorithm is described. This learning algorithm is different from the conventional. This integer weights and hard limit function are used for synaptic weight values and activation function, respectively. The entire learning process is performed by layer-by-layer method. the number of layers can be varied with difficulty of training data. Since the synaptic weight values are integers, the synapse circuit can be easily implemented with CMOS. PLD MLP neural network was applied to English Characters, arbitrary waveform generation and spiral problem.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2018년도 춘계학술발표대회
• /
• pp.183-186
• /
• 2018

In this paper, we propose the database design for skill learning service through the internet from the viewpoint of service engineering. This paper we describe the outlines for a design theory for skill learning service, which can lead to the satisfaction of both learner and instructor. Compared to other services, motion control learning takes a considerable amount of time, and this leads to a difficulty for learners to rate the quality of the service as well as for the instructors to provide consistent quality and standard of teaching. To solve these problems, we use a relational database with MongoDB which is an unstructured database allowing to flexibly incorporate the demands of both learner and instructor into the database itself.

• 한국멀티미디어학회논문지
• /
• 제22권10호
• /
• pp.1133-1141
• /
• 2019

Most of the previous studies for bark recognition have focused on the extraction of LBP-like statistical features. Deep learning approach was not well studied because of the difficulty of acquiring large volume of bark image dataset. To overcome the bark dataset problem, this study utilizes the MobileNet which was trained with the ImageNet dataset. This study proposes two approaches. One is to extract features by the pixel-wise convolution and classify the features with SVM. The other is to tune the weights of the MobileNet by flexibly freezing layers. The experimental results with two public bark datasets, BarkTex and Trunk12, show that the proposed methods are effective in bark recognition. Especially the results of the flexible tunning method outperform state-of-the-art methods. In addition, it can be applied to mobile devices because the MobileNet is compact compared to other deep learning models.

• Journal of Information Processing Systems
• /
• 제17권2호
• /
• pp.385-398
• /
• 2021

The complexity of deep learning models affects the real-time performance of gesture recognition, thereby limiting the application of gesture recognition algorithms in actual scenarios. Hence, a residual learning neural network based on a deep convolutional neural network is proposed. First, small convolution kernels are used to extract the local details of gesture images. Subsequently, a shallow residual structure is built to share weights, thereby avoiding gradient disappearance or gradient explosion as the network layer deepens; consequently, the difficulty of model optimisation is simplified. Additional convolutional neural networks are used to accelerate the refinement of deep abstract features based on the spatial importance of the gesture feature distribution. Finally, a fully connected cascade softmax classifier is used to complete the gesture recognition. Compared with the dense connection multiplexing feature information network, the proposed algorithm is optimised in feature multiplexing to avoid performance fluctuations caused by feature redundancy. Experimental results from the ISOGD gesture dataset and Gesture dataset prove that the proposed algorithm affords a fast convergence speed and high accuracy.

• Journal of Information Processing Systems
• /
• 제15권5호
• /
• pp.1119-1130
• /
• 2019

With the wide spread of Social Network Services (SNS), fake news-which is a way of disguising false information as legitimate media-has become a big social issue. This paper proposes a deep learning architecture for detecting fake news that is written in Korean. Previous works proposed appropriate fake news detection models for English, but Korean has two issues that cannot apply existing models: Korean can be expressed in shorter sentences than English even with the same meaning; therefore, it is difficult to operate a deep neural network because of the feature scarcity for deep learning. Difficulty in semantic analysis due to morpheme ambiguity. We worked to resolve these issues by implementing a system using various convolutional neural network-based deep learning architectures and "Fasttext" which is a word-embedding model learned by syllable unit. After training and testing its implementation, we could achieve meaningful accuracy for classification of the body and context discrepancies, but the accuracy was low for classification of the headline and body discrepancies.

• Nuclear Engineering and Technology
• /
• 제55권10호
• /
• pp.3515-3524
• /
• 2023

A reinforcement learning framework is proposed for the control problem of outlet steam pressure of the once-through steam generator(OTSG) in this paper. The double-layer controller using Proximal Policy Optimization(PPO) algorithm is applied in the control structure of the OTSG. The PPO algorithm can train the neural networks continuously according to the process of interaction with the environment and then the trained controller can realize better control for the OTSG. Meanwhile, reinforcement learning has the characteristic of difficult application in real-world objects, this paper proposes an innovative pretraining method to solve this problem. The difficulty in the application of reinforcement learning lies in training. The optimal strategy of each step is summed up through trial and error, and the training cost is very high. In this paper, the LSTM model is adopted as the training environment for pretraining, which saves training time and improves efficiency. The experimental results show that this method can realize the self-adjustment of control parameters under various working conditions, and the control effect has the advantages of small overshoot, fast stabilization speed, and strong adaptive ability.

• 산경연구논집
• /
• 제9권8호
• /
• pp.35-40
• /
• 2018

Purpose - The purpose of this study is to conduct a monitoring study on the learning process of e-learning contents. This study has two research objectives. First, by conducting monitoring research on the learning process, we aim to explore the implications for content development that reflects future student needs. Second, we want to collect empirical basic data on the estimation of appropriate amount of learning. Research design, data, and methodology - This study is a case study of learner's learning process in e-learning. After completion of the study, an in-depth interview was made after conducting a test to measure the total amount of cognitive load and the level of engagement that occurred during the learning process. The tool used to measure cognitive load is NASA-TLX, a subjective cognitive load measurement method. In the monitoring process, we observe external phenomena such as page movement and mouse movement path, and identify cognitive activities such as Think-Aloud technique. Results - In the total of three research subjects, the two courses showed excess learning time compared to the learning time, and one course showed less learning time than the learning time. This gives the following implications for content development. First, it is necessary to consider the importance of selecting the target and contents level according to the level of the subject. Second, it is necessary to design the learner participation activity that meets the learning goal level and to calculate the appropriate time accordingly. Third, it is necessary to design appropriate learning support strategy according to the learning task. This should be considered in designing lessons. Fourth, it is necessary to revitalize contents design centered on learning activities such as simulation. Conclusions - The implications of the examination system are as follows. First, it can be confirmed that there is difficulty in calculating the amount of learning centered on learning time and securing objective objectivity. Second, it can be seen that there are various variables affecting the actual learning time in addition to the content amount. Third, there is a need for reviewing the system of examination of learning amount centered on 'learning time'.