• Smart Media Journal
• /
• v.12 no.1
• /
• pp.24-31
• /
• 2023

Meta-learning is metacognition that instantly distinguishes between knowing and unknown. It is a learning method that adapts and solves new problems by self-learning with a small amount of data.A few-shot learning method is a type of meta-learning method that accurately predicts query data even with a very small support set. In this study, we propose a method to solve the limitations of the prototype created with the mean-point vector of each class. For this purpose, we use the few-shot learning method that created the prototype used in the few-shot learning method as the median prototype. For quantitative evaluation, a handwriting recognition dataset and mini-Imagenet dataset were used and compared with the existing method. Through the experimental results, it was confirmed that the performance was improved compared to the existing method.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.31 no.6
• /
• pp.1215-1225
• /
• 2021

Although Artificial Intelligence (AI) techniques have shown impressive performance in various fields, they are vulnerable to adversarial examples which induce misclassification by adding human-imperceptible perturbations to the input. Previous studies to defend the adversarial examples can be classified into three categories: (1) model retraining methods; (2) input transformation methods; and (3) adversarial examples detection methods. However, even though the defense methods against adversarial examples have constantly been proposed, there is no research to classify the type of adversarial attack. In this paper, we proposed an adversarial attack family classification method based on dimensionality reduction and clustering. Specifically, after extracting adversarial perturbation from adversarial example, we performed Linear Discriminant Analysis (LDA) to reduce the dimensionality of adversarial perturbation and performed K-means algorithm to classify the type of adversarial attack family. From the experimental results using MNIST dataset and CIFAR-10 dataset, we show that the proposed method can efficiently classify five tyeps of adversarial attack(FGSM, BIM, PGD, DeepFool, C&W). We also show that the proposed method provides good classification performance even in a situation where the legitimate input to the adversarial example is unknown.

• Journal of The Korean Association For Science Education
• /
• v.44 no.3
• /
• pp.231-248
• /
• 2024

This study developed a chatbot for first-year high school students, employing open-source software and the Korean Sentence-BERT model for AI-powered document classification. The chatbot utilizes the Sentence-BERT model to find the six most similar Q&A pairs to a student's query and presents them in a carousel format. The initial dataset, built from online resources, was refined and expanded based on student feedback and usability throughout over the operational period. By the end of the 2023 academic year, the chatbot integrated a total of 30,819 datasets and recorded 3,457 student interactions. Analysis revealed students' inclination to use the chatbot when prompted by teachers during classes and primarily during self-study sessions after school, with an average of 2.1 to 2.2 inquiries per session, mostly via mobile phones. Text mining identified student input terms encompassing not only science-related queries but also aspects of school life such as assessment scope. Topic modeling using BERTopic, based on Sentence-BERT, categorized 88% of student questions into 35 topics, shedding light on common student interests. A year-end survey confirmed the efficacy of the carousel format and the chatbot's role in addressing curiosities beyond integrated science learning objectives. This study underscores the importance of developing chatbots tailored for student use in public education and highlights their educational potential through long-term usage analysis.

• Journal of The Korean Association For Science Education
• /
• v.43 no.6
• /
• pp.573-582
• /
• 2023

This study aimed to explore the potential use of artificial intelligence in science education for gifted students by analyzing the structure of abstracts written by students at a gifted science academy and comparing the performance of various elements extracted using AI. The study involved an analysis of 263 graduation theses from S Science High School over five years (2017-2021), focusing on the frequency and types of background, objectives, methods, results, and discussions included in their abstracts. This was followed by an evaluation of their accuracy using AI classification methods with fine-tuning and prompts. The results revealed that the frequency of elements in the abstracts written by gifted students followed the order of objectives, methods, results, background, and discussions. However, only 57.4% of the abstracts contained all the essential elements, such as objectives, methods, and results. Among these elements, fine-tuned AI classification showed the highest accuracy, with background, objectives, and results demonstrating relatively high performance, while methods and discussions were often inaccurately classified. These findings suggest the need for a more effective use of AI, through providing a better distribution of elements or appropriate datasets for training. Educational implications of these findings were also discussed.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.9
• /
• pp.1305-1311
• /
• 2022

With the recent rapid development of mobile terminals and personal computers and the advent of neural network technology, real-time face swapping using images has become possible. In particular, the cycle generative adversarial network made it possible to replace faces using uncorrelated image data. In this paper, we propose an input data processing scheme that can improve the quality of face swapping with less training data and time. The proposed scheme can improve the image quality while preserving facial structure and expression information by combining facial landmarks extracted through a pre-trained neural network with major information that affects the structure and expression of the face. Using the blind/referenceless image spatial quality evaluator (BRISQUE) score, which is one of the AI-based non-reference quality metrics, we quantitatively analyze the performance of the proposed scheme and compare it to the conventional schemes. According to the numerical results, the proposed scheme obtained BRISQUE scores improved by about 4.6% to 14.6%, compared to the conventional schemes.

• Journal of the Institute of Convergence Signal Processing
• /
• v.23 no.3
• /
• pp.166-172
• /
• 2022

In this study, we photograph driving obstacle objects such as crosswalks, side spheres, manholes, braille blocks, partial ramps, temporary safety barriers, stairs, and inclined curb that hinder or cause inconvenience to the movement of the vulnerable using electric mobility aids. We develop an optimal AI model that classifies photographed objects and automatically recognizes them, and implement an algorithm that can efficiently determine obstacles in front of electric mobility aids. In order to enable object detection to be AI learning with high probability, the labeling form is labeled as a polygon form when building a dataset. It was developed using a Mask R-CNN model in Detectron2 framework that can detect objects labeled in the form of polygons. Image acquisition was conducted by dividing it into two groups: the general public and the transportation weak, and image information obtained in two areas of the test bed was secured. As for the parameter setting of the Mask R-CNN learning result, it was confirmed that the model learned with IMAGES_PER_BATCH: 2, BASE_LEARNING_RATE 0.001, MAX_ITERATION: 10,000 showed the highest performance at 68.532, so that the user can quickly and accurately recognize driving risks and obstacles.

• The Journal of the Korea Contents Association
• /
• v.21 no.12
• /
• pp.896-905
• /
• 2021

In the field of SIEM(Security information and event management), many studies try to use a feedback system to solve lack of completeness of training data and false positives of new attack events that occur in the actual operation. However, the current feedback system requires too much human inputs to improve the running model and even so, those feedback from inexperienced analysts can affect the model performance negatively. Therefore, we propose "active model improving feedback technology" to solve the shortage of security analyst manpower, increasing false positive rates and degrading model performance. First, we cluster similar predicted events during the operation, calculate feedback priorities for those clusters and select and provide representative events from those highly prioritized clusters using XAI (eXplainable AI)-based event visualization. Once these events are feedbacked, we exclude less analogous events and then propagate the feedback throughout the clusters. Finally, these events are incrementally trained by an existing model. To verify the effectiveness of our proposal, we compared three distinct scenarios using PKDD2007 and CSIC2012. As a result, our proposal confirmed a 30% higher performance in all indicators compared to that of the model with no feedback and the current feedback system.

• The Journal of the Acoustical Society of Korea
• /
• v.41 no.3
• /
• pp.359-366
• /
• 2022

In this paper, we present the Korean menu-ordering Sentence Text-to-Speech (TTS) system using conformer-based FastSpeech2. Conformer is the convolution-augmented transformer, which was originally proposed in Speech Recognition. Combining two different structures, the Conformer extracts better local and global features. It comprises two half Feed Forward module at the front and the end, sandwiching the Multi-Head Self-Attention module and Convolution module. We introduce the Conformer in Korean TTS, as we know it works well in Korean Speech Recognition. For comparison between transformer-based TTS model and Conformer-based one, we train FastSpeech2 and Conformer-based FastSpeech2. We collected a phoneme-balanced data set and used this for training our models. This corpus comprises not only general conversation, but also menu-ordering conversation consisting mainly of loanwords. This data set is the solution to the current Korean TTS model's degradation in loanwords. As a result of generating a synthesized sound using ParallelWave Gan, the Conformer-based FastSpeech2 achieved superior performance of MOS 4.04. We confirm that the model performance improved when the same structure was changed from transformer to Conformer in the Korean TTS.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.25 no.6
• /
• pp.555-567
• /
• 2023

The cracks in the tunnel are currently determined through visual inspections conducted by inspectors based on images acquired using tunnel imaging acquisition systems. This labor-intensive approach, relying on inspectors, has inherent limitations as it is subject to their subjective judgments. Recently research efforts have actively explored the use of deep learning to automatically detect tunnel cracks. However, most studies utilize public datasets or lack sufficient objectivity in the analysis process, making it challenging to apply them effectively in practical operations. In this study, we selected test datasets consisting of images in the same format as those obtained from the actual inspection system to perform an objective evaluation of deep learning models. Additionally, we introduced ensemble techniques to complement the strengths and weaknesses of the deep learning models, thereby improving the accuracy of crack detection. As a result, we achieved high recall rates of 80%, 88%, and 89% for cracks with sizes of 0.2 mm, 0.3 mm, and 0.5 mm, respectively, in the test images. In addition, the crack detection result of deep learning included numerous cracks that the inspector could not find. if cracks are detected with sufficient accuracy in a more objective evaluation by selecting images from other tunnels that were not used in this study, it is judged that deep learning will be able to be introduced to facility safety inspection.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.20 no.1
• /
• pp.219-223
• /
• 2020

To guarantee AI model's prominent recognition rate and recognition precision, obtaining the large number of data is essential. In this paper, we propose transfer learning-based object detection algorithm for maintaining outstanding performance even when the volume of training data is small. Also, we proposed a tranfer learning network combining Resnet-50 and YOLO(You Only Look Once) network. The transfer learning network uses the Leeds Sports Pose dataset to train the network that detects the person who occupies the largest part of each images. Simulation results yield to detection rate as 84% and detection precision as 97%.