How you type tells who you are : Keystroke dynamics-based user authentication

Pilsung Kang(강필성) 한국경영과학회 2013 한국경영과학회 학술대회논문집 Vol.2013 No.5

이상치 탐지 방법론을 활용한 반도체 가상 계측 결과의 신뢰도 추정

강필성(Pilsung Kang),김동일(Dongil Kim),이승경(Seung-kyung Lee),도승용(Seungyong Doh),조성준(Sungzoon Cho) 대한산업공학회 2012 대한산업공학회지 Vol.38 No.1

The purpose of virtual metrology (VM) in semiconductor manufacturing is to predict every wafer’s metrological values based on its process equipment data without an actual metrology. In this paper, we propose novelty detection-based reliability estimation models for VM in order to support flexible utilization of VM results. Because the proposed model can not only estimate the reliability of VM, but also identify suspicious process variables lowering the reliability, quality control actions can be taken selectively based on the reliance level and its causes. Based on the preliminary experimental results with actual semiconductor manufacturing process data, our models can successfully give a high reliance level to the wafers with small prediction errors and a low reliance level to the wafers with large prediction errors. In addition, our proposed model can give more detailed information by identifying the critical process variables and their relative impacts on the low reliability.

기술가치평가를 위한 시장대체원가 접근법

강필성(Pilsung Kang),금영정(Youngjung Geum),박현우(Hyun-woo Park),김상국(Sang-gook Kim),성태응(Tae-eung Sung),이학연(Hakyeon Lee) 대한산업공학회 2015 대한산업공학회지 Vol.41 No.2

This paper proposes a new approach to technology valuation, the market-replacement cost approach which integrates the cost-based approach and market-based approach. The proposed approach estimates the market-replacement cost of a target technology using R&D costs of similar R&D projects previously conducted. Similar R&D projects are extracted from project database based on document similarity between project proposals and technology description of the target technology. R&D costs of similar R&D projects are adjusted by mirroring the rate of technological obsolescence and inflation. Market-replacement cost of the technology is then derived by calculating the weighted average of adjusted costs and similarity values of similar R&D projects. A case of “Prevention method and system for the diffusion of mobile malicious code” is presented to illustrate the proposed approach.

How you type tells who you are : Keystroke dynamics-based user authentication

Pilsung Kang(강필성) 대한산업공학회 2013 대한산업공학회 춘계학술대회논문집 Vol.2013 No.5

데이터 불균형 문제에서의 SVM 앙상블 기법의 적용

강필성(Pilsung Kang),이형주(Hyoung-joo Lee),조성준(Sungzoon Cho) 한국정보과학회 2004 한국정보과학회 학술발표논문집 Vol.31 No.2Ⅱ

대부분의 기계학습 알고리즘은 학습 데이터에서 각각의 범주간의 비율이 동일하거나 비슷하다는 가정 하에 문제를 풀게 된다. 그러나 실제 문제에서는 그 비율이 동일하지 않으며 매우 큰 차이를 보이기도 하는데 이는 분류 성능을 저하시키는 요인이기도 하다. 따라서 본 논문에서는 이러한 데이터의 불균형 문제를 해소하는 방안으로 SVM 앙상블 기법을 적용한 샘플링을 제안하고 이를 실제 불균형 데이터에 적용함으로써 제안된 방법이 기존의 방법들에 비해 향상된 성능을 나타내는 것을 보였다.

의원간 유사성에 기반한 18대 국회의원 투표행태 분석

강필성(Pilsung Kang),박영준(Youngjoon Park),조수곤(Sugon Cho),김성범(Seoung Bum Kim) 대한산업공학회 2014 대한산업공학회지 Vol.40 No.1

This paper aims to propose a research framework of analyzing voting activities of a national assembly on the basis of member-level voting similarity and provides a case study in the 18<SUP>th</SUP> national assembly in South Korea. First, we propose a bill contentiousness measure that gives a higher score to bills for which ayes and noes are more diversified in both conservative and progressive parties. Based on the bill contentiousness measure, the top 5%, 10%, and 20% bills were identified and used for further analyses. Moreover, we propose a member-level voting similarity measure that compensates for the lower frequency of noes, and evaluate the pair-wise voting similarities for all lawmakers. Then, voting similarity differences to the affiliated/non-affiliated parties were analyzed for the members in the two major parties according to some internal/external key factors. Finally, similar voting groups were identified and their affiliations were investigated based on the multi-dimensional scaling (MDS) and network analysis techniques. A case study on the 18<SUP>th</SUP> national assembly of South Korea showed that the cohesion of the members in the ‘Hanara’ party becomes higher than that of the ‘Minju’ party as the bill contentiousness increases, whereas the number of elected, local constituency versus proportional representation, and the competition intensity in a local constituency were found to be partially influential to the voting activities of lawmakers. In addition, MDS and network analysis showed that there is a distinctive difference between two parties when all bills are analyzed, whereas the diversity of parties increases in the same group as the bill contentiousness increases.

분류 성능 향상을 위한 지역적 선형 재구축 기반 결측치 대치

강필성(Pilsung Kang) 대한산업공학회 2012 대한산업공학회지 Vol.38 No.4

Classification algorithms generally assume that the data is complete. However, missing values are common in real data sets due to various reasons. In this paper, we propose to use locally linear reconstruction (LLR) for missing value imputation to improve the classification performance when missing values exist. We first investigate how much missing values degenerate the classification performance with regard to various missing ratios. Then, we compare the proposed missing value imputation (LLR) with three well-known single imputation methods over three different classifiers using eight data sets. The experimental results showed that (1) any imputation methods, although some of them are very simple, helped to improve the classification accuracy; (2) among the imputation methods, the proposed LLR imputation was the most effective over all missing ratios, and (3) when the missing ratio is relatively high, LLR was outstanding and its classification accuracy was as high as the classification accuracy derived from the compete data set.

기계학습을 활용한 레이더 공기건조기 이상 탐지

김요진(Yojin Kim),강필성(Pilsung Kang),염동원(Dong-Won Yeom),김건우(Kun-Woo Kim) 한국산학기술학회 2023 한국산학기술학회논문지 Vol.24 No.3

무기체계가 첨단화됨에 따라 기존의 사후정비나 예방정비를 통해 무기체계를 운용 및 정비하는 것은 선제적 관리측면에서 한계점이 존재한다. 이를 극복하기 위해 국방부에서는 「국방개혁 2020」을 통해 빅데이터 기반의 총수명주기관리체계의 중요성을 강조하였고, 국방업무 전분야에서 상태기반정비(CBM: Condition Based Maintenance, 이하 CBM)를 적용하는 사업이 추진되고 있다. 하지만 보안규제 및 아날로그식 장비운용 등으로 인해 데이터 수집 및 활용에 제약이 있고 이로 인해 빅데이터 기반의 CBM을 적용하는데 어려움을 겪고 있다. 본 연구에서는 이러한 제약으로 인해 충분히 확보되지 않은 장비의 운용데이터를 이용하여 레이더의 고장방지를 위한 공기건조기의 이상탐지 모델 구축 사례를 제시하며, 운용데이터와 고장이력이 충분하지 않은 여건에서의 무기체계의 이상탐지 방법을 제안한다. 6개 부대의 운용데이터로 확보한 정상데이터를 이용하여 비지도학습 알고리듬 기반의 여러 가지 이상탐지 모델을 구축한 후 최적의 모델을 선정하는 작업을 수행하였다. 또한, 부대별로 정상상태에서의 오알람율을 낮추고 최대한 빠르게 이상을 탐지할 수 있도록 최적의 임계치 선정 및 고장알람기준을 정의하는 작업을 진행하였다. 본 연구결과는 무기체계의 운용데이터를 이용하여 기계학습 기반의 CBM 모델을 확보하는 실증사례를 확보함으로써 추후 다른 무기체계에서도 인공지능 적용을 통한 기계학습 방법론의 적용 가능성을 확인하고 기반을 마련하였다. As weapon systems become advanced, it is difficult to operate and maintain them through existing post-maintenance or preventive maintenance methods. To overcome this, the Ministry of National Defense has emphasized the importance of management of the total life cycle system through Defense Reform 2020, and a project is being promoted to apply big-data-based state-based maintenance (CBM) in all areas of defense work. However, due to security regulations and the operation of analog facilities, data collection and use are limited, and it is difficult to apply big-data-based CBM. This research shows a case of constructing an abnormal detection model of an air dryer to prevent radar failure using the operation data of facilities that are not sufficiently secured due to the restrictions mentioned. An abnormal detection method is proposed for a weapon system under conditions with insufficient operation data and failure history. After acquiring normal operational data from six military units, an anomaly detection model was made using various anomaly detection techniques from unsupervised learning, and an optimal model was selected. In addition, an optimal threshold and alarm criteria were selected to decrease the false alarm rate in normal conditions for each unit and to detect abnormalities as quickly as possible. The results of this study confirmed the possibility of applying anomaly detection through AI application to another weapon system and laid the foundation of securing empirical cases, CBM models using AI, and data analysis techniques with actual operation data.

품질기능전개를 이용한 인지재활 ICT 서비스 항목 설계 및 우선순위 도출

이종찬(Jongchan Lee),강필성(Pilsung Kang) 한국정보기술학회 2014 한국정보기술학회논문지 Vol.12 No.6

This paper configures information and communication technology (ICT) service attributes for cognitive rehabilitation based on quality function deployment (QFD). Initially, quality assessment criteria for general medical service are identified. Then, a survey on the current cognitive rehabilitation service is conducted for both therapists as service providers and patients as customers. Their responses are analyzed to assign the weight of each customer requirement. Then, service components are identified and the relationship between the customer requirements and service components are mapped through thorough interviews with ICT experts in medical service industry. Finally, the house of quality is built through the process of quality function deployment and the priority of ICT service components are determined. ‘Mobile’ is found to be the most important attributes, followed by ‘Monitoring’ and ‘Early warning’.

자유로운 문자열의 키스트로크 다이나믹스와 일범주 분류기를 활용한 사용자 인증

서동민(Dongmin Seo),강필성(Pilsung Kang) 대한산업공학회 2016 대한산업공학회지 Vol.42 No.5

User authentication is an important issue on computer network systems. Most of the current computer network systems use the ID-password string match as the primary user authentication method. However, in password-based authentication, whoever acquires the password of a valid user can access the system without any restrictions. In this paper, we present a keystroke dynamics-based user authentication to resolve limitations of the password-based authentication. Since most previous studies employed a fixed-length text as an input data, we aims at enhancing the authentication performance by combining four different variable creation methods from a variable-length free text as an input data. As authentication algorithms, four one-class classifiers are employed. We verify the proposed approach through an experiment based on actual keystroke data collected from 100 participants who provided more than 17,000 keystrokes for both Korean and English. The experimental results show that our proposed method significantly improve the authentication performance compared to the existing approaches.