• Proceedings of the Korea Multimedia Society Conference
• /
• 2004.05a
• /
• pp.636-639
• /
• 2004

본 논문은 웹 도큐먼트를 기반으로 사용자에게 의미 있는 정보를 찾아주기 위한 연관 객체 추출 기법인 PMPL(Personalized Multi-Strategey Pattern Loaming) 시스템을 제안하고자 한다. PMPL 모듈은 인터넷의 정보를 여과하여 필터링하고, 사용자 개인화의 키워드를 중심으로 연관된 객체를 추출한다. 이때 연관된 객체 추출 시 대용량 데이터에서 시간적, 공간적면에서 효율적인 연관 탐색 기법인 Fp-Tree와 Fp-Growth 알고리즘을 적용시켰으며, 연관규칙 탐색을 보완하기 위해 가중치 기법인 만유인력 기법을 적용시켰다. PMPL 시스템을 실행한 결과 개인화된 사용자 중심어 기초로 기존의 단일 학습 기법에 비해 더 많은 의미 있는 연관 지식을 추출한 결과가 보였다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2008.06c
• /
• pp.325-330
• /
• 2008

전자우편(email)은 인터넷의 급격한 보급으로 인하여 사용자들이 많이 사용하게 된 통신 메커니즘이다. 그러나 이러한 전자우편의 대중성을 상업적인 목적으로 이용한 스팸메일의 출현으로, 사용자들은 정신적 피해, 업무 방해, 메일서버의 트래픽 과부화로 인한 유지보수 비용 증가와 같은 문제점들을 접하게 되었다. 특히, 최근에는 광고성 이미지들을 첨부하는 등의 새로운 기법이 적용된 스팸메일의 발생으로 기존의 텍스트 기반의 스팸메일 필터링 기법들이 무의미하게 되었으며, 따라서 그로 인한 피해가 증가하는 추세이다. 이러한 이미지 기반의 스팸메일들의 필터링을 위하여 Support Vector Machine과 같은 기계학습 기법을 이용한 기법들이 제안되고 있으나, 여전히 그 성능은 만족스럽지 못하다. 본 논문은 전자우편으로부터 텍스트 및 시각적 의미를 분석하여 전자우편 온톨로지에 기술하고 스팸메일 판단을 위한 의미추론규칙을 적용함으로써 광고성 이미지가 첨부되어 있는 스팸메일을 효과적으로 필터링 하기 위한 시스템을 제안한다.

• ICROS
• /
• v.17 no.3
• /
• pp.22-26
• /
• 2011

신경회로망 기술은 다양한 공학적 및 과학적 문제에 적용되어 왔으며 복잡한 동특성을 갖는 시스템의 모델링에 특히 효율적인 것으로 알려져 있다. 신경회로망 학습은 신경회로망의 가중치 및 바이러스로서 주어지는 파라미터 벡터의 요소를 주어진 목적함수를 최소화하는 최적의 값으로 추정하는 연산과정을 의미한다. 따라서 신경회로망 파라미터 학습은 전체시스템의 성능을 직접적으로 좌우하는 매우 중요한 단계라 할 수 있으며 일반적으로 파라미터의 수정규칙 알고리즘을 도출한다. 이러한 수정규칙은 주로 최적화 기법을 적용하며 경사함수(gradient function)를 포함한다. 최근에는 이러한 경사함수를 포함하지 않는 학습 알고리즘이 많이 개발되고 있으며 특히 칼만 필터링 이론을 접목한 미분 신경회로망의 학습 알고리즘이 최근에 발표되었다.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.9
• /
• pp.101-108
• /
• 2020

Recommender system is a must-have feature of e-commerce, since it provides customers with convenience in selecting products. Collaborative filtering is a widely-used and representative technique, where it gives recommendation lists of products preferred by other users or preferred by the current user in the past. Recently, researches on the recommendation system using deep learning artificial intelligence technologies are actively being conducted to achieve performance improvement. This study develops a collaborative filtering based recommender system using restricted Boltzmann machines of the deep learning technology by utilizing user ratings. Moreover, a learning parameter update algorithm is proposed for learning efficiency and performance. Performance evaluation of the proposed system is made through experimental analysis and comparison with conventional collaborative filtering methods. It is found that the proposed algorithm yields superior performance than the basic restricted Boltzmann machines.

• Journal of IKEEE
• /
• v.23 no.4
• /
• pp.1393-1399
• /
• 2019

With the rapid growth of the Internet, millions of indiscriminate advertising SMS are sent every day because of the convenience of sending and receiving data. Although we still use methods to block spam words manually, we have been actively researching how to filter spam in a various ways as machine learning emerged. However, spam words and patterns are constantly changing to avoid being filtered, so existing machine learning mechanisms cannot detect or adapt to new words and patterns. Recently, the concept of Lifelong Learning emerged to overcome these limitations, using existing knowledge to keep learning new knowledge continuously. In this paper, we propose a method of spam filtering system using ensemble techniques of naive bayesian which is most commonly used in document classification and LLML(Lifelong Machine Learning). We validate the performance of lifelong learning by applying the model ELLA and the Naive Bayes most commonly used in existing spam filters.

• Journal of KIISE
• /
• v.43 no.12
• /
• pp.1325-1333
• /
• 2016

Software classification has several applications such as copyright infringement detection, malware classification, and software automatic categorization in software repositories. It can be also employed by software filtering systems to prevent the transmission of illegal software. If illegal software is identified by measuring software similarity in software filtering systems, the average number of comparisons can be reduced by shrinking the search space. In this study, we focused on the classification of Windows executables using API call information and machine learning. We evaluated the classification performance of machine learning-based classifier according to the refinement method for API information and machine learning algorithm. The results showed that the classification success rate of SVM (Support Vector Machine) with PolyKernel was higher than other algorithms. Since the API call information can be extracted from binary executables and machine learning-based classifier can identify tampered executables, API call information and machine learning-based software classifiers are suitable for software filtering systems.

• Annual Conference of KIPS
• /
• 2019.10a
• /
• pp.907-910
• /
• 2019

본 연구에서는 X-ray 영상에서 비산란 그리드 장치의 영향으로 인한 아티팩트를 제거하기 위하여 이산코사인변환(DCT: discrete cosine transform) 기반의 주파수 분석 기법과 딥러닝 네트워크의 학습 기법을 상호 보완적으로 결합하는 방법론을 제안한다. 피사체의 특성에 따라 다양하게 나타나는 그리드 라인의 억제 기능을 학습하기 위하여 서로 다른 특성을 반영하는 3 종류의 학습데이터를 생성한다. 학습에 사용되는 그리드 라인 영상의 타겟 데이터를 산출하기 위하여 DCT 기반의 밴드스톱 필터링 기법을 사용하였으며 학습데이터의 양적인 부족을 해결하기 위하여 패치 기반의 학습 방법을 적용하였다. 제안된 방법에 대해 기존의 방법과 비교하여 피사체 경계선 영역에서 발생하는 성능저하 현상, 분할의 가장자리에서 발생하는 블로킹 현상, 배경 영상에서의 성능저하 현상 등을 상대적으로 개선할 수 있음을 실험적으로 평가하였다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2018.07a
• /
• pp.63-66
• /
• 2018

스팸 스미싱 SMS의 차단을 위하여 지금까지 다양한 차단기법이 개발되어 사용되고 있다. 그 중에서도 대부분을 차지하는 방법들이 기계학습을 통한 내용 기반의 차단과 사용자의 스팸신고를 통한 송신자 차단 방법이다. 그러나 이러한 방법들은 공통적으로 스팸 스미싱을 식별하기 위해 학습 데이터가 필요하다는 문제점을 갖고 있기 때문에, 신종 스팸 공격들은 차단이 불가능하여 차단율의 한계를 보인다. 본 논문에서는 오늘날 사용되고 있는 스팸 스미싱 차단 기법들의 근본적인 문제점들을 규명하고, 이를 해결할 수 있는 스팸차단 기법 중 하나인 송신자 자가인증 기법을 소개한다. 그리고 송신자 자가인증 기법을 적용한 안드로이드애플리케이션의 구현 및 동작과정을 설명한다.

• Journal of Intelligence and Information Systems
• /
• v.21 no.1
• /
• pp.119-142
• /
• 2015

With the explosive growth in the volume of information, Internet users are experiencing considerable difficulties in obtaining necessary information online. Against this backdrop, ever-greater importance is being placed on a recommender system that provides information catered to user preferences and tastes in an attempt to address issues associated with information overload. To this end, a number of techniques have been proposed, including content-based filtering (CBF), demographic filtering (DF) and collaborative filtering (CF). Among them, CBF and DF require external information and thus cannot be applied to a variety of domains. CF, on the other hand, is widely used since it is relatively free from the domain constraint. The CF technique is broadly classified into memory-based CF, model-based CF and hybrid CF. Model-based CF addresses the drawbacks of CF by considering the Bayesian model, clustering model or dependency network model. This filtering technique not only improves the sparsity and scalability issues but also boosts predictive performance. However, it involves expensive model-building and results in a tradeoff between performance and scalability. Such tradeoff is attributed to reduced coverage, which is a type of sparsity issues. In addition, expensive model-building may lead to performance instability since changes in the domain environment cannot be immediately incorporated into the model due to high costs involved. Cumulative changes in the domain environment that have failed to be reflected eventually undermine system performance. This study incorporates the Markov model of transition probabilities and the concept of fuzzy clustering with CBCF to propose predictive clustering-based CF (PCCF) that solves the issues of reduced coverage and of unstable performance. The method improves performance instability by tracking the changes in user preferences and bridging the gap between the static model and dynamic users. Furthermore, the issue of reduced coverage also improves by expanding the coverage based on transition probabilities and clustering probabilities. The proposed method consists of four processes. First, user preferences are normalized in preference clustering. Second, changes in user preferences are detected from review score entries during preference transition detection. Third, user propensities are normalized using patterns of changes (propensities) in user preferences in propensity clustering. Lastly, the preference prediction model is developed to predict user preferences for items during preference prediction. The proposed method has been validated by testing the robustness of performance instability and scalability-performance tradeoff. The initial test compared and analyzed the performance of individual recommender systems each enabled by IBCF, CBCF, ICFEC and PCCF under an environment where data sparsity had been minimized. The following test adjusted the optimal number of clusters in CBCF, ICFEC and PCCF for a comparative analysis of subsequent changes in the system performance. The test results revealed that the suggested method produced insignificant improvement in performance in comparison with the existing techniques. In addition, it failed to achieve significant improvement in the standard deviation that indicates the degree of data fluctuation. Notwithstanding, it resulted in marked improvement over the existing techniques in terms of range that indicates the level of performance fluctuation. The level of performance fluctuation before and after the model generation improved by 51.31% in the initial test. Then in the following test, there has been 36.05% improvement in the level of performance fluctuation driven by the changes in the number of clusters. This signifies that the proposed method, despite the slight performance improvement, clearly offers better performance stability compared to the existing techniques. Further research on this study will be directed toward enhancing the recommendation performance that failed to demonstrate significant improvement over the existing techniques. The future research will consider the introduction of a high-dimensional parameter-free clustering algorithm or deep learning-based model in order to improve performance in recommendations.

• The Journal of the Korea Contents Association
• /
• v.7 no.4
• /
• pp.66-73
• /
• 2007

Internet is a part of our common life and tremendous information is cumulated. In these trends, the personalization becomes a very important technology which could find exact information to present users. Previous personalized services use content based filtering which is able to recommend by analyzing the content and collaborative filtering which is able to recommend contents according to preference of users group. But, collaborative filtering needs the evaluation of some amount of data. Also, It cannot reflect all data of users because it recommends items based on data of some users who have similar inclination. Therefore, we need a new recommendation method which can recommend prefer items without preference data of users. In this paper, we proposed the apparel coordination system using personalized preference on the semantic web. This paper provides the results which this system can reduce the searching time and advance the customer satisfaction measurement according to user's feedback to system.