• Convergence Security Journal
• /
• v.22 no.3
• /
• pp.101-110
• /
• 2022

As the cases of using artificial intelligence in various fields increase, attempts to solve various issues through artificial intelligence in the intrusion detection field are also increasing. However, the black box basis, which cannot explain or trace the reasons for the predicted results through machine learning, presents difficulties for security professionals who must use it. To solve this problem, research on explainable AI(XAI), which helps interpret and understand decisions in machine learning, is increasing in various fields. Therefore, in this paper, we propose an explanatory AI to enhance the reliability of machine learning-based intrusion detection prediction results. First, the intrusion detection model is implemented through XGBoost, and the description of the model is implemented using SHAP. And it provides reliability for security experts to make decisions by comparing and analyzing the existing feature importance and the results using SHAP. For this experiment, PKDD2007 dataset was used, and the association between existing feature importance and SHAP Value was analyzed, and it was verified that SHAP-based explainable AI was valid to give security experts the reliability of the prediction results of intrusion detection models.

• Journal of Intelligence and Information Systems
• /
• v.29 no.1
• /
• pp.249-263
• /
• 2023

Collaborative Filtering, a representative recommendation system methodology, consists of two approaches: neighbor methods and latent factor models. Among these, the latent factor model using matrix factorization decomposes the user-item interaction matrix into two lower-dimensional rectangular matrices, predicting the item's rating through the product of these matrices. Due to the factor vectors inferred from rating patterns capturing user and item characteristics, this method is superior in scalability, accuracy, and flexibility compared to neighbor-based methods. However, it has a fundamental drawback: the need to reflect the diversity of preferences of different individuals for items with no ratings. This limitation leads to repetitive and inaccurate recommendations. The Adaptive Deep Latent Factor Model (ADLFM) was developed to address this issue. This model adaptively learns the preferences for each item by using the item description, which provides a detailed summary and explanation of the item. ADLFM takes in item description as input, calculates latent vectors of the user and item, and presents a method that can reflect personal diversity using an attention score. However, due to the requirement of a dataset that includes item descriptions, the domain that can apply ADLFM is limited, resulting in generalization limitations. This study proposes a Generalized Adaptive Deep Latent Factor Recommendation Model, G-ADLFRM, to improve the limitations of ADLFM. Firstly, we use item ID, commonly used in recommendation systems, as input instead of the item description. Additionally, we apply improved deep learning model structures such as Self-Attention, Multi-head Attention, and Multi-Conv1D. We conducted experiments on various datasets with input and model structure changes. The results showed that when only the input was changed, MAE increased slightly compared to ADLFM due to accompanying information loss, resulting in decreased recommendation performance. However, the average learning speed per epoch significantly improved as the amount of information to be processed decreased. When both the input and the model structure were changed, the best-performing Multi-Conv1d structure showed similar performance to ADLFM, sufficiently counteracting the information loss caused by the input change. We conclude that G-ADLFRM is a new, lightweight, and generalizable model that maintains the performance of the existing ADLFM while enabling fast learning and inference.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.295-299
• /
• 2022

자원이 부족한 언어 환경에서 사전학습 언어모델 학습을 위한 대용량의 코퍼스를 구축하는데는 한계가 존재한다. 본 논문은 이러한 한계를 극복할 수 있는 Cross-lingual Post-Training (XPT) 방법론을 적용하여 비교적 자원이 부족한 한국어에서 해당 방법론의 효율성을 분석한다. 적은 양의 한국어 코퍼스인 400K와 4M만을 사용하여 다양한 한국어 사전학습 모델 (KLUE-BERT, KLUE-RoBERTa, Albert-kor)과 mBERT와 전반적인 성능 비교 및 분석 연구를 진행한다. 한국어의 대표적인 벤치마크 데이터셋인 KLUE 벤치마크를 사용하여 한국어 하위태스크에 대한 성능평가를 진행하며, 총 7가지의 태스크 중에서 5가지의 태스크에서 XPT-4M 모델이 기존 한국어 언어모델과의 비교에서 가장 우수한 혹은 두번째로 우수한 성능을 보인다. 이를 통해 XPT가 훨씬 더 많은 데이터로 훈련된 한국어 언어모델과 유사한 성능을 보일 뿐 아니라 학습과정이 매우 효율적임을 보인다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.45-46
• /
• 2024

본 논문에서는 Self-Attention 기반 딥러닝 기법인 Dense Prediction Transformer(DPT) 모델을 MVTec Anomaly Detection(MVTec AD) 데이터셋에 적용하여 실제 산업 제품 이미지 내 이상 부분을 분할하는 연구를 진행하였다. DPT 모델의 적용을 통해 기존 Convolutional Neural Network(CNN) 기반 이상 탐지기법의 한계점인 지역적 Feature 추출 및 고정된 수용영역으로 인한 문제를 개선하였으며, 실제 산업 제품 데이터에서의 이상 분할 시 기존 주력 기법인 U-Net의 구조를 적용한 최고 성능의 모델보다 1.14%만큼의 성능 향상을 보임에 따라 Self-Attention 기반 딥러닝 기법의 적용이 산업 제품 이상 분할에 효과적임을 입증하였다.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.18 no.5
• /
• pp.999-1008
• /
• 2023

Early prediction of chronic diseases such as diabetes is an important issue, and improving the accuracy of diabetes prediction is especially important. Various machine learning and deep learning-based methodologies are being introduced for diabetes prediction, but these technologies require large amounts of data for better performance than other methodologies, and the learning cost is high due to complex data models. In this study, we aim to verify the claim that DNN using the pima dataset and k-fold cross-validation reduces the efficiency of diabetes diagnosis models. Machine learning classification methods such as decision trees, SVM, random forests, logistic regression, KNN, and various ensemble techniques were used to determine which algorithm produces the best prediction results. After training and testing all classification models, the proposed system provided the best results on XGBoost classifier with ADASYN method, with accuracy of 81%, F1 coefficient of 0.81, and AUC of 0.84. Additionally, a domain adaptation method was implemented to demonstrate the versatility of the proposed system. An explainable AI approach using the LIME and SHAP frameworks was implemented to understand how the model predicts the final outcome.

• Journal of The Korean Institute of Defense Technology
• /
• v.5 no.2
• /
• pp.1-9
• /
• 2023

The purpose of this study is to derive the need to improve the test and evaluation(T&E) performance system of the weapon systems to which advanced science and technology is applied, evaluate priorities, and present development plans. T&E of Al-based weapon systems through a literature research and case analysis on changes in the T&E environment for weapon system, 11 detailed evaluation items were derived from the in terms of the T&E system, structure, and technology. Al-based weapon system test evaluation should be performed in parallel with data-based performance evaluation and actual T&E, and performance measurement using a separate T&E data set is required for AI models performance evaluation. As a result of analyzing the importance of T&E through AHP analysis, the order of T&E system-technology-structure was evaluated, and the priority of detailed evaluation items was evaluated in the order of T&E result judgment-T&E organization and expert training-scientific T&E. For evaluation items with high priority, measures to improve the T&E performance system were presented.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.11a
• /
• pp.592-594
• /
• 2023

차량 탐지와 차량 추적 기술은 교통관리 시스템, 자율주행 자동차 시스템 및 이를 응용한 보안 감시 시스템, 군사 작전 및 안전 등 다양한 산업 분야에서 활용되고 있다. 본 논문에서는 차량 탐지는 YOLOv7 모델을, 차량 추적은 DeepSORT 알고리즘을 사용하여 도로의 차량들에 대해 탐지 및 추적을 순차적으로 진행하였다. 실험환경은 차량 탐지 데이터 셋(dataset)을 직접 라벨링(labeling) 하여 실험하였고, 차량 추적은 차량 탐지에서 학습해서 얻은 체크포인트(checkpoint) 모델을 가중치로 설정하여 실험을 진행하였다. 차량 탐지 실험결과는 validation 과 test 에서 높은 정확도를 확인할 수 있었고, 차량 추적은 Namsa 비디오 및 Seohaegyo 비디오에서도 차량 추적이 잘 되고 있음을 확인할 수 있었다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.12
• /
• pp.1872-1879
• /
• 2022

Recently, with the development of artificial intelligence technology, research to use artificial intelligence to detect hacking attacks is being actively conducted. However, the fact that security data is a representative imbalanced data is recognized as a major obstacle in composing the learning data, which is the key to the development of artificial intelligence models. Therefore, in this paper, we propose a W-VAE oversampling technique that applies VAE, a deep learning generation model, to data extraction for oversampling, and sets the number of oversampling for each class through weight calculation using K-NN for sampling. In this paper, a total of five oversampling techniques such as ROS, SMOTE, and ADASYN were applied through NSL-KDD, an open network security dataset. The oversampling method proposed in this paper proved to be the most effective sampling method compared to the existing oversampling method through the F1-Score evaluation index.

• The Transactions of the Korea Information Processing Society
• /
• v.13 no.2
• /
• pp.76-90
• /
• 2024

Although artificial intelligence (AI) can be utilized in various domains such as smart city, healthcare, it is limited due to concerns about the exposure of personal and sensitive information. In response, the concept of distributed machine learning has emerged, wherein learning occurs locally before training a global model, mitigating the concentration of data on a central server. However, overall learning phase in a collaborative way among multiple participants poses threats to data privacy. In this paper, we systematically analyzes recent trends in privacy protection within the realm of distributed machine learning, considering factors such as the presence of a central server, distribution environment of the training datasets, and performance variations among participants. In particular, we focus on key distributed machine learning techniques, including horizontal federated learning, vertical federated learning, and swarm learning. We examine privacy protection mechanisms within these techniques and explores potential directions for future research.

• Journal of the Korea Society of Computer and Information
• /
• v.24 no.12
• /
• pp.17-24
• /
• 2019

The biggest problem in the AI field, which has become a hot topic in recent years, is how to deal with the lack of training data. Since manual data construction takes a lot of time and efforts, it is non-trivial for an individual to easily build the necessary data. On the other hand, automatic data construction needs to handle data quality issue. In this paper, we introduce a method to automatically extract the data required to develop Korean speech command recognizer from the web and to automatically select the data that can be used for training data. In particular, we propose a modified ResNet model that shows modest performance for the automatically constructed Korean speech command data. We conducted an experiment to show the applicability of the command set of the health and daily life domain. In a series of experiments using only automatically constructed data, the accuracy of the health domain was 89.5% in ResNet15 and 82% in ResNet8 in the daily lives domain, respectively.