• Journal of IKEEE
• /
• v.25 no.1
• /
• pp.180-192
• /
• 2021

One widely used algorithm for image recognition and pattern detection is the convolution neural network (CNN). To efficiently handle convolution operations, which account for the majority of computations in the CNN, we use hardware accelerators to improve the performance of CNN applications. In using these hardware accelerators, the CNN fetches data from the off-chip DRAM, as the massive computational volume of data makes it difficult to derive performance improvements only from memory inside the hardware accelerator. In other words, data communication between off-chip DRAM and memory inside the accelerator has a significant impact on the performance of CNN applications. In this paper, a simulator for the CNN is developed to analyze the main memory or DRAM access with respect to the size of the on-chip memory or global buffer inside the CNN accelerator. For AlexNet, one of the CNN architectures, when simulated with increasing the size of the global buffer, we found that the global buffer of size larger than 100kB has 0.8x as low a DRAM access count as the global buffer of size smaller than 100kB.

• Journal of the Korean Institute of Gas
• /
• v.24 no.6
• /
• pp.91-97
• /
• 2020

Lithium-ion batteries (LIBs) have a longer lifespan, higher energy density, and lower self-discharge rates than other batteries, therefore, they are preferred as an Energy Storage System (ESS). However, during years 2017-2019, 28 ESS fire accidents occurred in Korea, and accurate capacity estimation of LIB is essential to ensure safety and reliability during operations. In this study, data-driven modeling that predicts capacity changes according to the charging cycle of LIB was conducted, and developed models were compared their performance for the selection of the optimal machine learning model, which includes the Decision Tree, Ensemble Learning Method, Support Vector Regression, and Gaussian Process Regression (GPR). For model training, lithium battery test data provided by NASA was used, and GPR showed the best prediction performance. Based on this study, we will develop an enhanced LIB capacity prediction and remaining useful life estimation model through additional data training, and improve the performance of anomaly detection and monitoring during operations, enabling safe and stable ESS operations.

• Journal of Internet Computing and Services
• /
• v.23 no.4
• /
• pp.11-19
• /
• 2022

This paper is a study to evaluate the performance when classifying explosive components by capacity using a convolutional neural network (CNN). Among the existing explosive classification methods, the IMS steam detector method determines the presence or absence of an explosive only when the explosive concentration exceeds the threshold set by the user. The IMS steam detector has a problem of determining that even if an explosive exists, the explosive does not exist in an amount that does not exceed the threshold. Therefore, it is necessary to detect the explosive component even when the concentration of the explosive component does not exceed the threshold. Accordingly, in this paper, after imaging explosive time series data with the Gramian Angular Field (GAF) algorithm, it is possible to determine whether there are explosive components and the amount of explosive components even when the concentration of explosive components does not exceed a threshold.

• Convergence Security Journal
• /
• v.24 no.2
• /
• pp.9-17
• /
• 2024

This paper examines the security threats to enterprise Generative Artificial Intelligence systems and proposes countermeasures. As AI systems handle vast amounts of data to gain a competitive edge, security threats targeting AI systems are rapidly increasing. Since AI security threats have distinct characteristics compared to traditional human-oriented cybersecurity threats, establishing an AI-specific response system is urgent. This study analyzes the importance of AI system security, identifies key threat factors, and suggests technical and managerial countermeasures. Firstly, it proposes strengthening the security of IT infrastructure where AI systems operate and enhancing AI model robustness by utilizing defensive techniques such as adversarial learning and model quantization. Additionally, it presents an AI security system design that detects anomalies in AI query-response processes to identify insider threats. Furthermore, it emphasizes the establishment of change control and audit frameworks to prevent AI model leakage by adopting the cyber kill chain concept. As AI technology evolves rapidly, by focusing on AI model and data security, insider threat detection, and professional workforce development, companies can improve their digital competitiveness through secure and reliable AI utilization.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.20 no.1
• /
• pp.163-169
• /
• 2020

Recently, research on automation and unmanned operation of machines in the industrial field has been conducted with the advent of AI, Big data, and the IoT, which are the core technologies of the Fourth Industrial Revolution. The machines for these automation processes are controlled based on the data collected from the sensors attached to them, and further, the processes are managed. Conventionally, the abnormalities of sensors are periodically checked and managed. However, due to various environmental factors and situations in the industrial field, there are cases where the inspection due to the failure is not missed or failures are not detected to prevent damage due to sensor failure. In addition, even if a failure occurs, it is not immediately detected, which worsens the process loss. Therefore, in order to prevent damage caused by such a sudden sensor failure, it is necessary to identify the failure of the sensor in an embedded system in real-time and to diagnose the failure and determine the type for a quick response. In this paper, a deep neural network-based fault diagnosis system is designed and implemented using Raspberry Pi to classify typical sensor fault types such as erratic fault, hard-over fault, spike fault, and stuck fault. In order to diagnose sensor failure, the network is constructed using Google's proposed Inverted residual block structure of MobilieNetV2. The proposed scheme reduces memory usage and improves the performance of the conventional CNN technique to classify sensor faults.

• Journal of the Korean Institute of Gas
• /
• v.12 no.2
• /
• pp.38-41
• /
• 2008

In this research, we propose a new method for tracking the release rate using the concentration data obtained from the sensor. We used a sensor network that has already been set surrounding the area where hazardous gas releases can occur. From the real-time sensor data, we detected and analyzed releases of harmful materials and their concentrations. Based on the results, the release rate is estimated using the neural network. This model consists of 14 input variables (sensor data, material properties, process information, meteorological conditions) and one output (release rate). The dispersion model then performs the simulation of the expected dispersion consequence by combining the sensor data, GIS data and the diagnostic result of the source term. The result of this study will improve the safety-concerns of residents living next to storage facilities containing hazardous materials by providing the enhanced emergency response plan and monitoring system for toxic gas releases.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2013.08a
• /
• pp.82.2-82.2
• /
• 2013

반도체 및 디스플레이 소자를 생산 하기 위하여 다양하고 많은 공정 기술이 사용 되며 그 중에서 플라즈마를 이용하는 제조공정이 차지 하는 부분은 상당한 부분을 차지 하고 있습니다. 전체 반도체 공정 중 48%가 진공공정이며, 진공공정 중 68% 이상이 플라즈마를 이용하고 있으며, 식각과 증착 장비 뿐만 아니라 세정과 이온증착 에 이르기 까지 다양하며 앞으로도 더욱 범위가 늘어 날 것으로 보입니다. 이러한 플라즈마를 이용한 제조 공정들은 제품의 생산성을 향상 하기 위하여 오염제어 기술을 비롯한 공정관리기술 그리고 고기능 센서기술을 이용한 공정 모니터링 및 제어 기술에 이르기 까지 다양한 기술들을 필요로 합니다. 플라즈마를 이용한 제조 장비는 RF파워모듈, 진공제어모듈, 공정가스제어모듈, 웨이퍼 및 글래스의 반송장치, 그리고 온도제어 모듈과 같이 다양한 장치의 집합체라 할 수 있습니다. 플라즈마의 생성과 이를 제어 하기 위한 기술은 제조장비의 국산화를 위한 부단한 노력의 결실로 많은 부분 기술이 축적되어 왔고 성과를 거두고 있습니다. 그러나 고기능 모니터링 센서 기술 개발은 그 동안 활발 하게 이루어져 오고 있지 않았으며 대부분 외산 기술에 의존해 왔습니다. 세계 반도체 시장은 현재 300 mm 웨이퍼 가공에서, 추후 450 mm 시장으로 패러다임이 변화될 예정이며, 미세화 공정이 더욱 진행 됨에 따라 반도체 제조사들의 관심사가 "성능 중심의 반도체 제조기술"로부터 "오류 최소를 통한 생산성 향상"에 더욱 주목 하고 있습니다. 공정미세화 및 웨이퍼 대구경화로 인해 실시간 복합 센서를 이용한 데이터 처리 알고리즘 및 자동화 소프트웨어의 기능이 탑재된 장비를 요구하고 있습니다. 주식회사 레인보우 코퍼레이션은 플라즈마 Chemistry상태를 정성 분석 가능한 OES (Optical Emission Spectroscopy)를 이용한 EPD System을 상용화 하여 고객사에 공급 중이며, 플라즈마의 광 신호를 실시간으로 고속 계측함과 동시에 최적화된 알고리즘을 이용하여 플라즈마의 이상 상태를 감지하며 이를 통하여 제조 공정 및 장비의 개선을 가능하게 하여 고객 제품의 생산성을 향상 하도록 하는 기술을 개발 하고 있습니다. 본 심포지엄에서는 주식회사 레인보우 코퍼레이션이 개발 중인 "실시간 고속 플라즈마 광 모니터링 기술" 의 개념을 소개하고, 제품의 응용 범위와 응용 방법에 대하여 설명을 하고자 합니다.

• Journal of the Korean Institute of Gas
• /
• v.1 no.1
• /
• pp.27-32
• /
• 1997

Continual monitoring of abnormal operating conditions i a key issue in maintaining high product quality and safe operation, since the undetected process abnormality may lead to the undesirable operations, finally producing low quality products, or breakdown of equipment. The statistical projection method recently highlighted has the advantage of easily building reference model with the historical measurement data in the statistically in-control state and not requiring any detailed mathematical model or knowledge-base of process. As the complexity of process increases, however, we have more measurement variables and recycle streams. This situation may not only result in the frequent occurrence of process Perturbation, but make it difficult to pinpoint trouble-making causes or at most assignable source unit due to the confusing candidates. Consequently, an ad hoc skill to monitor and diagnose in plat-wide scale is needed. In this paper, we propose a hierarchical plant-wide monitoring methodology based on hierarchical decomposition and principal component analysis for handling the complexity and interactions among process units. This have the effect of preventing special events in a specific sub-block from propagating to other sub-blocks or at least delaying the transfer of undesired state, and so make it possible to quickly detect and diagnose the process malfunctions. To prove the performance of the proposed methodology, we simulate the Tennessee Eastman benchmark process which is operated continuously with 41 measurement variables of five major units. Simulation results have shown that the proposed methodology offers a fast and reliable monitoring and diagnosis for a large scale chemical plant.

• Journal of the Korean Society for information Management
• /
• v.37 no.2
• /
• pp.47-69
• /
• 2020

This research provides a quantitative analysis on research articles published in PLoS ONE, a multidisciplinary open access journal, by authors affiliated with Korean institutions. Korean authors published more than 6,500 research ariticles in the mega journal between 2006 and 2019. Korea is ranked the top 11th place in terms of article publishing in the journal. Most articles by Korean authors are concentrated in the biomedical fields. In recent years, the overall production of PLoS ONE has decreased as authors migrated to competing mega journals such as Scientific Reports and BMJ Open. The change might have been affected in part by the delay in the review period and the dropping impact factor score. The open access share of the Korean PLoS ONE authors of more than 10 articles hovers around 30%. However, there is a significant variation among researchers reaching up to 50% discrepancies. Among altmetrics provided by PLoS ONE, the saves are highly correlated with the views and the citations. On the contrary, the shares show low correlation with other use metrics. A follow up, survey questionnarie based research involving researchers who have published in PLoS ONE is planned in order to investigate author motivation and experience in the review process.

• Informatization Policy
• /
• v.26 no.4
• /
• pp.20-35
• /
• 2019

We empirically analyze the effects of psychological factors, such as the fear of parenting, on fertility rates. An index is calculated based on the share of negative news articles on child care in all social articles from 2000 to 2018. The analysis result shows that as the index increases, the fertility rate after three years falls. This result is repeated in the correlation analysis, simple regression, and VAR analysis. According to Granger causality analysis, it is found that the relation between the index and the fertility rate after three years is not just a simple correlation but a causal relationship. There are differences among age groups. The fertility rate of women in their 20s and 30s shows a significant response to the index, but that of the 40s does not. The index affects the birthrate of first child, but do not affect the birthrate of second or more children. These results are consistent with the intuition that younger women are more likely to be affected by the negative articles about parenting, but not to those who have already experienced childbirth. This study is meaningful in that a significant index for predicting social phenomena is extracted beyond the limited use of news big data such as a simple keyword mention volume monitoring. Also, this big data-based index is a 3-year leading indicator for fertility, which provides the advantage of providing information that helps early detection.