检索结果-内蒙古大学图书馆

Addressing Imbalance in Health Datasets: A New Method NR-Clustering SMOTE and Distance Metric Modification

Computers, Materials & Continua 2025年第2期82卷 2931-2949页

作者： Hairani Hairani Triyanna Widiyaningtyas Didik Dwi Prasetya Afrig Aminuddin Department of Electrical Engineering and Informatics Faculty of EngineeringUniversitas Negeri MalangMalang65145Indonesia Department of Computer Science Universitas BumigoraMataram83127Indonesia Department of Computer Graphic and Multimedia Faculty of ComputingCollege of Computing and Applied SciencesUniversiti Malaysia Pahang Al-Sultan AbdullahPekan26600Malaysia

An imbalanced dataset often challenges machine learning, particularly classification methods. Underrepresented minority classes can result in biased and inaccurate models. The Synthetic Minority Over-Sampling Technique (SMOTE) was developed to address the problem of imbalanced data. Over time, several weaknesses of the SMOTE method have been identified in generating synthetic minority class data, such as overlapping, noise, and small disjuncts. However, these studies generally focus on only one of SMOTE’s weaknesses: noise or overlapping. Therefore, this study addresses both issues simultaneously by tackling noise and overlapping in SMOTE-generated data. This study proposes a combined approach of filtering, clustering, and distance modification to reduce noise and overlapping produced by SMOTE. Filtering removes minority class data (noise) located in majority class regions, with the k-nn method applied for filtering. The use of Noise Reduction (NR), which removes data that is considered noise before applying SMOTE, has a positive impact in overcoming data imbalance. Clustering establishes decision boundaries by partitioning data into clusters, allowing SMOTE with modified distance metrics to generate minority class data within each cluster. This SMOTE clustering and distance modification approach aims to minimize overlap in synthetic minority data that could introduce noise. The proposed method is called “NR-Clustering SMOTE,” which has several stages in balancing data: (1) filtering by removing minority classes close to majority classes (data noise) using the k-nn method;(2) clustering data using K-means aims to establish decision boundaries by partitioning data into several clusters;(3) applying SMOTE oversampling with Manhattan distance within each cluster. Test results indicate that the proposed NR-Clustering SMOTE method achieves the best performance across all evaluation metrics for classification methods such as Random Forest, SVM, and Naїve Bayes, compared t

关键词： SMOTE modification Clustering-SMOTE manhattan distance

来源：评论

学校读者我要写书评

暂无评论

Backlit Keyboard Inspection Using Machine Vision

引用

Journal of Electronic Science and Technology 2015年第1期13卷 39-44页

作者： Der-Baau Perng Hsiao-Wei Liu Po-An Chen Department of Applied Informatics and Multimedia the International College Asia University Industrial Technology Research Institute Aurotek Corporation

A robust system for backlit keyboard inspection is revealed. The backlit keyboard not only has changeable diverse colors but also has the laser marking keys. The keys on the keyboard can be divided into regions of function keys, normal keys, and number keys. However, there might have some types of defects： incorrect illuminating area, non-uniform illumination of specified inspection region（IR）, and incorrect luminance and intensity of individual key. Since the illumination features of backlit keyboard are too complex to inspect for human inspector in the production line, an auto-mated inspection system for the backlit keyboard is proposed in this paper. The system was designed into the operation module and inspection module. A set of image processing methods were developed for these defects inspection. Some experimental results demonstrate the robustness and effectiveness of the proposed system.

关键词： Backlit keyboard illumination defect inspection machine vision uniformity

来源：评论

学校读者我要写书评

暂无评论

A novel cluster-based difference expansion transform for lossless data hiding

A novel cluster-based difference expansion transform for los...

引用

5th International Conference on Genetic and Evolutionary Computing, ICGEC2011

作者： Tsai, Yuan-Yu Chan, Chi-Shiang Department of Applied Informatics and Multimedia Asia University Taichung Taiwan

ISBN: (纸本)9780769544496

In this paper, we propose a lossless data hiding algorithm for grayscale images. Specifically, our technique is based on the cluster-based difference expansion transform. The main scenario behind our technique is that we use a recursive cluster construction technique to divide the input image into several clusters. In the data embedding process, a modified difference expansion transform is used to embed the secret message into the pixels cluster by cluster. Experimental results show that our technique can achieve high embedding capacity from 0.56 to 0.85 bpp while the PSNR value is over 30db. The technique provides a reversible method and has been demonstrated to be feasible in image data hiding. © 2011 IEEE.

关键词： Steganography

来源：评论

学校读者我要写书评

暂无评论

Folksonomy-based indexing for retrieving tutoring resources

Folksonomy-based indexing for retrieving tutoring resources

引用

2012 17th IEEE International Conference on Wireless, Mobile and Ubiquitous Technology in Education, WMUTE 2012

作者： Shih, Wen-Chung Tseng, Shian-Shyong Department of Applied Informatics and Multimedia Asia University Taichung 41354 Taiwan

ISBN: (纸本)9780769546629

As more and more undergraduate students act as voluntary tutors to rural pupils after school, there is a growing need for a resource repository to support tutors during their tutoring process. However, when tutoring resources are not text-based, such as a clip of Flash animation, the technology of conventional information retrieval cannot be simply applied to retrieve these resources. Therefore, we propose a folksonomy-based indexing method to improve the performance of retrieving non-textual tutoring resources. The proposed approach consists of an initializing phase and a self-organizing phase. This study investigates the performance of constructing and maintaining the folksonomy-based index. In addition, the attitudes of tutors toward the folksonomy-based indexing method are addressed. A prototype of the tutoring resource repository has been designed and implemented, and experiments have been conducted to evaluate the proposed approach. The results show that the folksonomy-based index can be constructed and maintained efficiently. Also, survey on tutors shows the proposed approach can help them find relevant resources efficiently. © 2012 IEEE.

关键词： Indexing (of information)

来源：评论

学校读者我要写书评

暂无评论

PhenoProfiler: Advancing Phenotypic Learning for Image-based Drug Discovery

arXiv

引用

arXiv 2025年

作者： Li, Bo Zhang, Bob Zhang, Chengyang Zhou, Minghao Huang, Weiliang Wang, Shihang Wang, Qing Li, Mengran Zhang, Yong Song, Qianqian PAMI Research Group Department of Computer and Information Science University of Macau Taipa China Beijing Key Laboratory of Multimedia and Intelligent Software Technology Beijing Institute of Artificial Intelligence Beijing University of Technology Beijing China Department of Health Outcomes and Biomedical Informatics College of Medicine University of Florida Florida United States Faculty of Applied Sciences Macao Polytechnic University Taipa China School of Intelligent Systems Engineering Sun Yat-sen University Guangdong China Department of Cancer Biology Wake Forest School of Medicine Winston SalemNC United States

In the field of image-based drug discovery, capturing the phenotypic response of cells to various drug treatments and perturbations is a crucial step. This process involves transforming high-throughput cellular images into quantitative representations for downstream analysis. However, existing methods require computationally extensive and complex multi-step procedures, which can introduce inefficiencies, limit generalizability, and increase potential errors. To address these challenges, we present PhenoProfiler, an innovative model designed to efficiently and effectively extract morphological representations, enabling the elucidation of phenotypic changes induced by treatments. PhenoProfiler is designed as an end-to-end tool that processes whole-slide multi-channel images directly into low-dimensional quantitative representations, eliminating the extensive computational steps required by existing methods. It also includes a multi-objective learning module to enhance robustness, accuracy, and generalization in morphological representation learning. PhenoProfiler is rigorously evaluated on large-scale publicly available datasets, including over 230,000 whole-slide multi-channel images in end-to-end scenarios and more than 8.42 million single-cell images in non-end-to-end settings. Across these benchmarks, PhenoProfiler consistently outperforms state-of-the-art methods by up to 20%, demonstrating substantial improvements in both accuracy and robustness. Furthermore, PhenoProfiler uses a tailored phenotype correction strategy to emphasize relative phenotypic changes under treatments, facilitating the detection of biologically meaningful signals. UMAP visualizations of treatment profiles demonstrate PhenoProfiler’s ability to effectively cluster treatments with similar biological annotations, thereby enhancing interpretability. These findings establish PhenoProfiler as a scalable, generalizable, and robust tool for phenotypic learning, offering transformative advancement

关键词： Benchmarking

来源：评论

学校读者我要写书评

暂无评论

Retaining and exploring online remains on YouTube

Retaining and exploring online remains on YouTube

引用

3rd International Conference on Emerging Intelligent Data and Web Technologies, EIDWT 2012

作者： Akoumianakis, Demosthenes Kafousis, Ionannis Karadimitriou, Nikolas Tsiknakis, Manolis Department of Applied Informatics and Multimedia Technological Education Institution of Crete Heraklion Greece

ISBN: (纸本)9780769547343

The paper presents a method and a collection of techniques for conducting virtual excavations in online social networking services. YouTube and its Data API are used as a case study of a virtual settlement. The objective is to assess not only what is retained by YouTube but also what sense can be made of a designated set of YouTube online remains. The research focuses on defining, capturing and transforming digital trace data into interactive visualizations that unlock crucial dynamics of online activity. © 2012 IEEE.

关键词： Excavation

来源：评论

学校读者我要写书评

暂无评论

Exploring LSB substitution and pixel-value differencing for block-based adaptive data hiding

引用

International Journal of Network Security 2014年第5期16卷 363-368页

作者： Tsai, Yuan-Yu Chen, Jian-Ting Chan, Chi-Shiang Department of Applied Informatics and Multimedia Asia University No. 500 Lioufeng Rd. Wufeng Dist. Taichung City 41354 Taiwan

Khodaei and Faez proposed a new adaptive data hiding technique based on LSB substitution and pixel-value differencing. Their algorithm can embed a large amount of secret data while maintaining acceptable image quality. However, their proposed algorithm only has fixed embedding capacity. In addition, the derivation for three consecutive pixels in the boundary region is poorly manipulated using raster scan order, resulting in inaccurate pixel differences. Finally, an overflow problem may occur for some embedding cases. In this study, we adopt non-overlapping blocks with m-by-n pixels to address the above problems. The cover image is first partitioned into non-overlapping blocks. The LSB substitution and optimal pixel adjustment process are then employed to embed the secret message into the central pixel of each block. The residual pixels within the same block are with message embedded using a pixel-value differencing scheme. The experimental results show that our proposed algorithm can achieve an adjustable embedding capacity according to the block size. The proposed technique is feasible in adaptive data hiding.

关键词： Pixels

来源：评论

学校读者我要写书评

暂无评论

An efficient and distortion-controllable information hiding algorithm for 3D polygonal models with adaptation

引用

International Journal of Network Security 2015年第1期17卷 79-84页

作者： Tsai, Yuan-Yu Huang, Wen-Ching Peng, Bo-Feng Department of Applied Informatics and Multimedia Asia University No. 500 Lioufeng Rd. Wufeng Dist. Taichung City41354 Taiwan

We present an efficient information hiding algorithm for polygonal models. The decision to referencing neighbors for each embeddable vertex is based on a modified breadth first search, starting from the initial polygon determining by principal component analysis. The surface complexity is then estimated by the distance between the embedding vertex and the center of its referencing neighbors. Different amounts of secret messages are adaptively embedded according to the surface properties of each vertex. A constant threshold is employed to control the maximum embedding capacity for each vertex and decrease the model distortion simultaneously. The experimental results show the proposed algorithm is efficient and can provide higher robustness, higher embedding capacity, and lower model distortion than previous work, with acceptable estimation accuracy. The proposed technique is feasible in 3D adaptive information hiding.

关键词： Principal component analysis

来源：评论

学校读者我要写书评

暂无评论

NCXplore: A design space exploration framework of temporal encoding for on-chip serial interconnects

引用

International Journal of High Performance Systems Architecture 2010年第3-4期2卷 177-186页

作者： Kornaros, George Electronics and Computer Engineering Department Technical University of Crete Kounoupidiana Chania Greece Applied Informatics and Multimedia Department Technological Educational Institute of Crete Heraklion Crete Greece

Multi-processor systems-on-chip (MPSoC) seek for high performance, scalable and power efficient communication infrastructures. Recent research considers on-chip serial links for communication fabrics as a solution to reduce routing congestion and design complexity. This paper describes a methodology and a tool-chain for design space exploration of temporal encoding schemes, thereafter referred to as NCXplore. NCXplore assists the designer to achieve the best fit as regards both switching activity combined with reduction of crosstalk effects, and performance. A novel class of temporal encoding schemes is also presented to manage switching activity and crosstalk induced delays. NCXplore accepts any encoding technique as a mapping function to investigate crosstalk effects. Copyright © 2010 Inderscience Enterprises Ltd.

关键词： Electric power utilization

来源：评论

学校读者我要写书评

暂无评论

Clinical practice Guideline Management Information Systems: Cancer guidelines as boundary spanning transformable objects of practice

Clinical practice Guideline Management Information Systems: ...

引用

International Workshop on Computational Intelligence in Networks and Systems

作者： Akoumianakis, Demosthenes Milolidakis, Giannis Akrivos, Anargyros Panteris, Zacharias Ktistakis, Giorgos Department of Applied Informatics and Multimedia Technological Education Institution of Crete Heraklion Crete Greece

ISBN: (纸本)9780769542782

The paper elaborates on the concept of transformable boundary artifacts and their role in fostering knowledge-based work in cross-organization virtual communities of practice. The domain of investigation is clinical practice guidelines development for cancer. By reviewing the social worlds involved, we claim that guideline development is a boundary spanning activity which can be facilitated through social networking tools of a Guideline Management Information System. Such a system is then described, focusing on the way it appropriates 'plasticity' to make guidelines cross social, institutional and technological boundaries. Contrasting earlier efforts, a key contribution of the research presented is that it emphasizes the dialogue and the intrinsic properties of developing (rather than interpreting) guidelines in collaborative settings. © 2010 IEEE.

关键词： Knowledge based systems

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：