With the rapid advancement of network technologies, cyberthreats have become increasingly sophisticated, posing significant challenges to traditional intrusion detection systems. Conventional machine learning and deep learning approaches frequently experience performance degradation when confronted with imbalanced datasets and novel attack vectors. To address these limitations, this study proposes a deep learning-based intrusion detection framework that employs feature fusion through incremental transfer learning between source and target domains. The proposed architecture integrates convolutional neural networks (CNNs) with an attention mechanism to extract and aggregate salient features, thereby enhancing the model’s discriminative capacity between normal traffic and various network attack categories. Experimental results demonstrate that the proposed model achieves a detection accuracy of 94.21% even when trained on only 33% of the available data, outperforming conventional models. These findings underscore the effectiveness of the proposed feature fusion strategy via transfer learning in improving detection capabilities within dynamic and evolving cyberthreat environments.
Recently, grid computing environment has being changed to the service-oriented grid integrated with Web Services for supporting more open, scalable and flexible grid architecture. In this architecture, a grid application is the service-oriented grid application in which Grid Services are executed according to the workflow of the grid application. This implies that the grid scheduling method needs to consider grid service characteristics. Although many grid scheduling methods have been developed, we need new grid scheduling method to reflect the service-oriented grid concept and to ensure performance of the grid application. Therefore, we propose the grid service performance model based on factors such as service availability, accessibility, reliability, throughput, and slowdown, and develop grid scheduling method to increase performance of the service-oriented grid application.
31 December 2021 in The Journal of Korean Institute of Information Technology
신경망 기반 딥러닝 기술은 이미지 감지, 분류 및 식별 분야에서 가장 많이 사용되고 있지만 많은 양의 이미지 데이터가 요구되기 때문에 군사 분야와 같이 데이터 획득이 제한적인 특정 분야에서는 활용이 제한될 수 있다. 이에 소규모 데이터를 이용하여 보다 높은 성능을 제공할 수 있는 방법들에 대한 연구들이 진행되고 있다. 본 논문에서는 군의 지상무기체계인 전차와 장갑차 이미지를 수집한 후 전이학습을 통해 도출된 각 모델 예측 값들의 평균을 구하는 앙상블을 수행하였다. 다수의 동종 및 이종모델들의 다양한 앙상블 조합에 대한 이미지 분류 정확도를 분석한 결과, 본 논문에서 고려한 이미지 데이터셋에 대하여 2개의 이종모델인 InceptionV3와 DenseNet121 앙상블 조합만으로도 98.5%의 분류 정확도를 제공할 수 있음을 확인할 수 있었다.
The aging of power plant pipelines has led to significant leaks worldwide, causing environmental damage, human safety risks, and economic losses. Rapid leak detection is critical for mitigating these issues, but challenges such as varying leak characteristics, ambient noise, and limited real-world data complicate their accurate detection and model development. To address these issues, we propose a leak detection model that integrates stepwise transfer learning and an attention mechanism. The proposed model utilizes a two-stage deep learning process. In Stage 1, one-dimensional convolutional neural networks (1D CNNs) are pre-trained to extract root mean square (RMS) and frequency-domain features from acoustic signals. In Stage 2, the classifier layers of the pre-trained models are removed, and the extracted features are fused and processed using a bidirectional long short-term memory (LSTM) network. An attention mechanism is incorporated within the LSTM to prioritize critical features, enhancing the ability of the model to distinguish leak signals from noise. The model achieved an accuracy of 99.99%, significantly outperforming the traditional methods considered in this study. By effectively addressing noise interference and data scarcity, this robust approach demonstrates its potential to enhance safety, reduce risks, and improve cost efficiency in industrial pipelines and critical infrastructure.
In a Kubernetes environment, the resource allocation for Pods has a direct impact on both performance and cost. When resource sizes are determined based on user experience, under-provisioning can lead to performance degradation and execution instability, while over-provisioning can result in resource waste and increased costs. To address these issues, this study proposes an adaptive scheduling method that employs the Design of Experiments (DoE) approach to determine the optimal resource size for each application with minimal experimentation and integrates the results into a custom Kubernetes scheduler. Experiments were conducted in a Kubernetes-based cloud environment using five applications with diverse workload characteristics, including CPU-intensive, memory-intensive, and AI inference workloads. The results show that the proposed method improved the performance score—calculated as the harmonic mean of execution time and cost—by an average of approximately 1.5 times (ranging from 1.15 to 1.59 times) compared with the conventional maximum resource allocation approach. Moreover, for all applications, the difference in mean scores before and after optimal resource allocation was statistically significant (p-value < 0.05). The proposed approach demonstrates scalability for achieving both resource efficiency and service-level agreement (SLA) compliance across various workload environments.