• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.32 no.5
• /
• pp.817-826
• /
• 2022

AI technology is being successfully introduced in many fields, and models deployed as a service are deployed with black box environment that does not expose the model's information to protect intellectual property rights and data. In a black box environment, attackers try to steal data or parameters used during training by using model output. This paper proposes a method of inferring the type of model to directly find out the composition of layer of the target model, based on the fact that there is no attack to infer the information about the type of model from the deep learning model. With ResNet, VGGNet, AlexNet, and simple convolutional neural network models trained with MNIST datasets, we show that the types of models can be inferred using the output values in the gray box and black box environments of the each model. In addition, we inferred the type of model with approximately 83% accuracy in the black box environment if we train the big and small relationship feature that proposed in this paper together, the results show that the model type can be infrerred even in situations where only partial information is given to attackers, not raw probability vectors.

• Smart Media Journal
• /
• v.11 no.2
• /
• pp.9-17
• /
• 2022

The purpose of this study is to implement a predictive model to support decision-making to replace a starting pitcher before a crisis situation in a baseball game. To this end, using the Major League Statcast data provided by Baseball Savant, we implement a predictive model that preemptively replaces starting pitchers before a crisis situation. To this end, first, the crisis situation that the starting pitcher faces in the game was derived through data exploration. Second, if the starting pitcher was replaced before the end of the inning, learning was carried out by composing a label with a replacement in the previous inning. As a result of comparing the trained models, the model based on the ensemble method showed the highest predictive performance with an F1-Score of 65%. The practical significance of this study is that the proposed model can contribute to increasing the team's winning probability by replacing the starting pitcher before a crisis situation, and the coach will be able to receive data-based strategic decision-making support during the game.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.12 no.3
• /
• pp.451-456
• /
• 2017

Semantic segmentation, which is the most basic and complicated problem in computer vision, classifies each pixel of an image into a specific object and performs a task of specifying a label. MRF and CRF, which have been studied in the past, have been studied as effective methods for improving the accuracy of pixel level labeling. In this paper, we propose a semantic partitioning method that combines CNN, a kind of deep running, which is in the spotlight recently, and CRF, a probabilistic model. For learning and performance verification, Pascal VOC 2012 image database was used and the test was performed using arbitrary images not used for learning. As a result of the study, we showed better partitioning performance than existing semantic partitioning algorithm.

• Smart Media Journal
• /
• v.10 no.1
• /
• pp.39-46
• /
• 2021

This study is about a method of extracting a summary from a news article in consideration of the importance of each sentence constituting the article. We propose a method of calculating sentence importance by extracting the probabilities of topic sentence, similarity with article title and other sentences, and sentence position as characteristics that affect sentence importance. At this time, a hypothesis is established that the Topic Sentence will have a characteristic distinct from the general sentence, and a deep learning-based classification model is trained to obtain a topic sentence probability value for the input sentence. Also, using the pre-learned ELMo language model, the similarity between sentences is calculated based on the sentence vector value reflecting the context information and extracted as sentence characteristics. The topic sentence classification performance of the LSTM and BERT models was 93% accurate, 96.22% recall, and 89.5% precision, resulting in high analysis results. As a result of calculating the importance of each sentence by combining the extracted sentence characteristics, it was confirmed that the performance of extracting the topic sentence was improved by about 10% compared to the existing TextRank algorithm.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.17 no.4
• /
• pp.577-586
• /
• 2022

Detection and classification of steel surface defects are critical for product quality control in the steel industry. However, due to its low accuracy and slow speed, the traditional approach cannot be effectively used in a production line. The current, widely used algorithm (based on deep learning) has an accuracy problem, and there are still rooms for development. This paper proposes a method of steel surface defect detection combining EfficientNetV2 for image classification and YOLOv5 as an object detector. Shorter training time and high accuracy are advantages of this model. Firstly, the image input into EfficientNetV2 model classifies defect classes and predicts probability of having defects. If the probability of having a defect is less than 0.25, the algorithm directly recognizes that the sample has no defects. Otherwise, the samples are further input into YOLOv5 to accomplish the defect detection process on the metal surface. Experiments show that proposed model has good performance on the NEU dataset with an accuracy of 98.3%. Simultaneously, the average training speed is shorter than other models.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.26 no.2
• /
• pp.93-98
• /
• 2016

For accurate precipitation forecasts the choice of weather factors and prediction method is very important. Recently, machine learning has been widely used for forecasting precipitation, and artificial neural network, one of machine learning techniques, showed good performance. In this paper, we suggest a new method for forecasting precipitation using DBN, one of deep learning techniques. DBN has an advantage that initial weights are set by unsupervised learning, so this compensates for the defects of artificial neural networks. We used past precipitation, temperature, and the parameters of the sun and moon's motion as features for forecasting precipitation. The dataset consists of observation data which had been measured for 40 years from AWS in Seoul. Experiments were based on 8-fold cross validation. As a result of estimation, we got probabilities of test dataset, so threshold was used for the decision of precipitation. CSI and Bias were used for indicating the precision of precipitation. Our experimental results showed that DBN performed better than MLP.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.3
• /
• pp.133-140
• /
• 2022

We investigate the performance of deep learning-based Korean language models on a task of predicting the score range of Korean essays written by foreign students. We construct a data set containing a total of 304 essays, which include essays discussing the criteria for choosing a job ('job'), conditions of a happy life ('happ'), relationship between money and happiness ('econ'), and definition of success ('succ'). These essays were labeled according to four letter grades (A, B, C, and D), and a total of eleven essay score range prediction experiments were conducted (i.e., five for predicting the score range of 'job' essays, five for predicting the score range of 'happiness' essays, and one for predicting the score range of mixed topic essays). Three deep learning-based Korean language models, KoBERT, KcBERT, and KR-BERT, were fine-tuned using various training data. Moreover, two traditional probabilistic machine learning classifiers, naive Bayes and logistic regression, were also evaluated. Experiment results show that deep learning-based Korean language models performed better than the two traditional classifiers, with KR-BERT performing the best with 55.83% overall average prediction accuracy. A close second was KcBERT (55.77%) followed by KoBERT (54.91%). The performances of naive Bayes and logistic regression classifiers were 52.52% and 50.28% respectively. Due to the scarcity of training data and the imbalance in class distribution, the overall prediction performance was not high for all classifiers. Moreover, the classifiers' vocabulary did not explicitly capture the error features that were helpful in correctly grading the Korean essay. By overcoming these two limitations, we expect the score range prediction performance to improve.

• The Journal of the Korea Contents Association
• /
• v.22 no.3
• /
• pp.71-80
• /
• 2022

In the past, fish diseases were bacterial in aqua farms, but in recent years, the frequency of fish diseases has increased as they have become viral and mixed. Viral diseases in an enclosed space called a aqua farm have a high spread rate, so it is very likely to lead to mass death. Fast identification of fish diseases is important to prevent group death. However, diagnosis of fish diseases requires a high level of expertise and it is difficult to visually check the condition of fish every time. In order to prevent the spread of the disease, an automatic identification system of diseases or fish is needed. In this paper, in order to improve the performance of the disease identification system of Paralichthys olivaceus based on deep learning, the existing pre-processing method is compared and tested. Target diseases were selected from three most frequent diseases such as Scutica, Vibrio, and Lymphocystis in Paralichthys olivaceus. The RGB, HLS, HSV, LAB, LUV, XYZ, and YCRCV were used as image pre-processing methods. As a result of the experiment, HLS was able to get the best results than using general RGB. It is expected that the fish disease identification system can be advanced by improving the recognition rate of diseases in a simple way.

• Journal of Digital Convergence
• /
• v.20 no.3
• /
• pp.325-331
• /
• 2022

As IoT devices are used in various ways in an edge network environment, multiple studies are being conducted that utilizes the information collected from IoT devices in various applications. However, it is not easy to apply accurate IoT data immediately as IoT data collected according to network environment (interference, interference, etc.) are frequently missed or error occurs. In order to minimize mistakes in IoT data collected in an edge network environment, this paper proposes a management technique that ensures the reliability of IoT data by randomly generating signature values of IoT data and allocating only Security Information (SI) values to IoT data in bit form. The proposed technique binds IoT data into a blockchain by applying multiple hash chains to asymmetrically link and process data collected from IoT devices. In this case, the blockchainized IoT data uses a probability function to which a weight is applied according to a correlation index based on deep learning. In addition, the proposed technique can expand and operate grouped IoT data into an n-layer structure to lower the integrity and processing cost of IoT data.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.16 no.12
• /
• pp.2785-2791
• /
• 2012

In this paper, a method for automatic classification of learning objects is proposed for effective management and reuse of learning contents. Proposed method will create cohesion of learning objects using cases of learning objects and perform automatic classification of learning objects by measuring their relationship based on cohesion. Application of proposed method to learning management system has the advantage of reducing the costs for developing learning contents. Simulation has shown the average accuracy of 28.20% with probability-based method and 56.38% with cohesion-based method. Simulation has proved that the method proposed in this paper is effective in automatic classification of learning objects.