• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.11
• /
• pp.4502-4518
• /
• 2015

Fisher linear discriminant analysis (LDA) is one of the most popular projection techniques for feature extraction and has been widely applied in face recognition. However, it cannot be used when encountering the single sample per person problem (SSPP) because the intra-class variations cannot be evaluated. In this paper, we propose a novel method called local similarity based linear discriminant analysis (LS_LDA) to solve this problem. Motivated by the "divide-conquer" strategy, we first divide the face into local blocks, and classify each local block, and then integrate all the classification results to make final decision. To make LDA feasible for SSPP problem, we further divide each block into overlapped patches and assume that these patches are from the same class. To improve the robustness of LS_LDA to outliers, we further propose local similarity based median discriminant analysis (LS_MDA), which uses class median vector to estimate the class population mean in LDA modeling. Experimental results on three popular databases show that our methods not only generalize well SSPP problem but also have strong robustness to expression, illumination, occlusion and time variation.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2018.07a
• /
• pp.53-54
• /
• 2018

최근 화력 발전소 보일러 튜브의 노후화로 인해서 불시정지 빈도수 및 재가동 시간이 늦춰지고 있다. 이는 막대한 경제적, 사회적 손실로 이어지며, 이를 예방하기 위해서는 상태기반 정비가 필요하다. 현재의 상태기반 정비는 센서, 신호 수집장치, 신호 분석단계를 거쳐 전문가가 진단하기 때문에 즉각적으로 대응하기 어려운 문제점이 있어서 설비의 재가동 시간이 늦춰지고 있다. 따라서 본 논문에서는 전문가의 도움 없이 자동으로 상태를 진단하기 위해서 머신러닝 기법 중 하나인 서포트 벡터 머신(SVM)을 이용한 진단 알고리즘을 구현하고, 이를 탑재한 진단장치를 개발하여 비전문가들도 즉각적으로 대응할 수 있게 하여 불시정지 시간과 빈도수를 줄이고자 한다.

• ETRI Journal
• /
• v.43 no.4
• /
• pp.674-683
• /
• 2021

A trust-aware secure routing protocol (TSRP) for wireless sensor networks is proposed in this paper to defend against varieties of attacks. First, each node calculates the comprehensive trust values of its neighbors based on direct trust value, indirect trust value, volatilization factor, and residual energy to defend against black hole, selective forwarding, wormhole, hello flood, and sinkhole attacks. Second, any source node that needs to send data forwards a routing request packet to its neighbors in multi-path mode, and this continues until the sink at the end is reached. Finally, the sink finds the optimal path based on the path's comprehensive trust values, transmission distance, and hop count by analyzing the received packets. Simulation results show that TSRP has lower network latency, smaller packet loss rate, and lower average network energy consumption than ad hoc on-demand distance vector routing and trust based secure routing protocol.

• International Journal of Computer Science & Network Security
• /
• v.22 no.1
• /
• pp.167-174
• /
• 2022

In metaheuristic algorithms such as Genetic Algorithm (GA), initial population has a significant impact as it affects the time such algorithm takes to obtain an optimal solution to the given problem. In addition, it may influence the quality of the solution obtained. In the machine learning field, feature selection is an important process to attaining a good performance model; Genetic algorithm has been utilized for this purpose by scientists. However, the characteristics of Genetic algorithm, namely random initial population generation from a vector of feature elements, may influence solution and execution time. In this paper, the use of a statistical algorithm has been introduced (Chi2) for feature relevant checks where p-values of conditional independence were considered. Features with low p-values were discarded and subject relevant subset of features to Genetic Algorithm. This is to gain a level of certainty of the fitness of features randomly selected. An ensembled-based learning model for Hepatitis has been developed for Hepatitis C stage classification. 1385 samples were used using Egyptian-dataset obtained from UCI repository. The comparative evaluation confirms decreased in execution time and an increase in model performance accuracy from 56% to 63%.

• International Journal of Computer Science & Network Security
• /
• v.23 no.9
• /
• pp.192-198
• /
• 2023

Cyberbullying is a growing problem among adolescents and can have serious psychological and emotional consequences for the victims. In recent years, machine learning techniques have emerged as promising approach for detecting instances of cyberbullying in online communication. This research paper focuses on developing a machine learning models that are able to detect cyberbullying including support vector machines, naïve bayes, and random forests. The study uses a dataset of real-world examples of cyberbullying collected from Twitter and extracts features that represents the ideational metafunction, then evaluates the performance of each algorithm before and after considering the theory of systemic functional linguistics in terms of precision, recall, and F1-score. The result indicates that all three algorithms are effective at detecting cyberbullying with 92% for naïve bayes and an accuracy of 93% for both SVM and random forests. However, the study also highlights the challenges of accurately detecting cyberbullying, particularly given the nuanced and context-dependent nature of online communication. This paper concludes by discussing the implications of these findings for future research and the development of practical tool for cyberbullying prevention and intervention.

• ETRI Journal
• /
• v.46 no.4
• /
• pp.648-659
• /
• 2024

Author verification is challenging because of the diversity in writing styles. We propose an enhanced handwriting verification method that combines handcrafted and automatically extracted features. The method uses a genetic algorithm to reduce the dimensionality of the feature set. We consider offline Bangla handwriting content and evaluate the proposed method using handcrafted features with a simple logistic regression, radial basis function network, and sequential minimal optimization as well as automatically extracted features using a convolutional neural network. The handcrafted features outperform the automatically extracted ones, achieving an average verification accuracy of 94.54% for 100 writers. The handcrafted features include Radon transform, histogram of oriented gradients, local phase quantization, and local binary patterns from interwriter and intrawriter content. The genetic algorithm reduces the feature dimensionality and selects salient features using a support vector machine. The top five experimental results are obtained from the optimal feature set selected using a consensus strategy. Comparisons with other methods and features confirm the satisfactory results.

• Journal of Broadcast Engineering
• /
• v.25 no.3
• /
• pp.313-324
• /
• 2020

In this paper, we propose a motion vector prediction method that increases the coding efficiency at the boundary of an image by utilizing the 360 video characteristic. In the current VVC method, the location of a neighbor block is excluded from the candidate list for inter prediction in case that it is out of boundary. This can reduce coding efficiency as well as subject quality. To solve this problem, we construct new candidates adding the location of the neighbor block at the boundary of the picture from already decoded information based on the projection method for 360 video coding. To evaluate the performance of the proposed method, we compare with VTM6.0 and 360Lib9.1 under Random Access condition of JVET-360 CTC. As a result, the coding performance shows a BD-rate reduction of 0.02% on average in luma component and 0.05%, 0.06% on average in chroma components respectively, without additional computational complexity. The coding performance at the picture boundary shows a BD-rate reduction of 0.29% on average in luma component and 0.45%, 0.43% on average in chroma components, respectively. Furthermore, we perform subjective quality test with the DSCQS method and obtain MOS values. The MOS value is improved by 0.03 value, and we calculate BD-MOS using MOS value and bit-rate. As a result, the proposed method improved performance by up to 8.78% and 5.18% on average.

• Journal of KIISE:Software and Applications
• /
• v.29 no.9
• /
• pp.676-683
• /
• 2002

The success of XML(eXtensible Markup Language) is primarily based on its flexibility : everybody can define the structure of XML documents that represent information in the form he or she desires. XML is so flexible that XML documents cannot be automatically provided with an underlying semantics. Different tag sets, different names for elements or attributes, or different document structures in general mislead the task of classifying and clustering XML documents precisely. In this paper, we design and implement a system that allows checking the semantic-based similarity between XML tags. First, this system extracts the underlying semantics of tags and then expands the synonym set of tags using an WordNet thesaurus and user-defined word library which supports the abbreviation forms and compound words for XML tags. Seconds, considering the relative importance of XML tags in the XML documents, we extend a conventional vector space model which is the most generally used for document model in Information Retrieval field. Using this method, we have been able to check the similarity between XML tags which are represented different tags.

• Asia-pacific Journal of Multimedia Services Convergent with Art, Humanities, and Sociology
• /
• v.7 no.3
• /
• pp.329-337
• /
• 2017

This paper proposes a search method that explores parameters C and σ values of support vector machines (SVM) to improve performance while maintaining search accuracy. A traditional grid search method requires tremendous computational times because it searches all available combinations of C and σ values to find optimal combinations which provide the best performance of SVM. To address this issue, this paper proposes a deep search method that reduces computational time. In the first stage, it divides C-σ- accurate metrics into four regions, searches a median value of each region, and then selects a point of the highest accurate value as a start point. In the second stage, the selected start points are re-divided into four regions, and then the highest accurate point is assigned as a new search point. In the third stage, after eight points near the search point. are explored and the highest accurate value is assigned as a new search point, corresponding points are divided into four parts and it calculates an accurate value. In the last stage, it is continued until an accurate metric value is the highest compared to the neighborhood point values. If it is not satisfied, it is repeated from the second stage with the input level value. Experimental results using normal and defect bearings show that the proposed deep search algorithm outperforms the conventional algorithms in terms of performance and search time.

• ETRI Journal
• /
• v.26 no.3
• /
• pp.269-272
• /
• 2004

We propose an efficient scheme for camera motion characterization in MPEG-compressed video. The proposed scheme detects six types of basic camera motions through threshold-based qualitative interpretation, in which fixed thresholds are applied to motion model parameters estimated from MPEG motion vectors (MVs). The efficiency and robustness of the scheme are validated by the experiment with real compressed video sequences.