An Under-Sampling Algorithm Based on SVM

机译：一种基于SVM的抽样算法

获取原文

页面导航

摘要
著录项
相似文献
相关主题

摘要

Tradition classification algorithms often get poor performance in imbalanced datasets because they are proposed under the assumption that the datasets are nearly balanced. Random under-sampling(RUS) algorithm is a popular algorithm to solve imbalance problem through removing some majority class samples randomly. However, RUS algorithm may neglect some key information of datasets. A new under-sampling algorithm based on SVM is proposed in this paper. The proposed algorithm aims to reserve samples distribution information in undersampling process. The simulation results show that the proposed algorithm could achieve satisfying performance.

机译：传统分类算法通常在不平衡数据集中获得差的性能，因为它们是在假设数据集几乎平衡的情况下提出的。随机欠采样（RUS）算法是一种流行的算法，可以通过随机删除一些多数类样本来解决不平衡问题。但是，RUS算法可能会忽略数据集的一些关键信息。本文提出了一种基于SVM的新的采样算法。该算法的旨在在欠采样过程中预留样本分布信息。仿真结果表明，该算法可以实现令人满意的性能。

著录项

来源
《IEEE International Conference on Artificial Intelligence and Industrial Design》|2021年|64-68|共5页
会议地点
作者
Zheng Hengyu;
展开▼
作者单位

展开▼
会议组织
原文格式 PDF
正文语种
中图分类
关键词
Support vector machines; Interpolation; Simulation; Conferences; Classification algorithms; Artificial intelligence;

机译：支持向量机;插值;模拟;会议;分类算法;人工智能;

相似文献

外文文献
中文文献
专利

1. Classification of Medical Datasets Using SVMs with Hybrid Evolutionary Algorithms Based on Endocrine-Based Particle Swarm Optimization and Artificial Bee Colony Algorithms [J] . Lin Kuan-Cheng, Hsieh Yi-Hsiu Journal of medical systems . 2015,第10期

机译：基于基于内分泌粒子群优化和人工蜂群算法的混合进化算法的支持向量机支持的医学数据集分类
2. Enhancing Subcellular Localization Prediction of Apoptosis Proteins by Ensemble SVMs with Random Under-Sampling [J] . Xiao Wang, Xiaohe Li, Hui Li, International Journal of Performability Engineering . 2018,第7期

机译：通过随机欠下抽样通过集合SVM提高细胞凋亡蛋白的亚细胞定位预测
3. Improving protein-ATP binding residues prediction by boosting SVMs with random under-sampling [J] . Dong-Jun Yu, Jun Hu, Zhen-Min Tang, Neurocomputing . 2013,第Mara15期

机译：通过使用随机欠采样增强SVM来改善蛋白质-ATP结合残基的预测
4. Performance evaluation of SVM and iterative FSVM classifiers with bootstrapping-based over-sampling and under-sampling [C] . Zughrat A., Mahfouf M., Thornton S. IEEE International Conference on Fuzzy Systems . 2015

机译：基于自举的过采样和欠采样对SVM和迭代FSVM分类器的性能评估
5. Voting Nearest Neighbors: SVM Constraints Selection Algorithm Based on K-Nearest Neighbors [D] . Moreira da Costa, Leandro. 2019

机译：投票最近的邻居：基于K-Indect邻居的SVM约束选择算法
6. MD-SVM: a novel SVM-based algorithm for the motif discovery of transcription factor binding sites [O] . Jialu Hu, Jingru Wang, Jianan Lin, 2019

机译：MD-SVM：一种基于SVM的新颖算法用于转录因子结合位点的基序发现
7. Figure 4: (A) One conserved sequence, which occurs 79 times in 46,264 binding site peaks from the ChIP-seq data-set. The mutation profile of this conserved sequence is illustrated, where ’_ ’ indicates this base is unchanged; DEL indicates this base is lost; INS X indicates a new base X is inserted in front of this base. (B) Several repeated elements patterns are listed. (C) In the first column, the top five DNA motifs, mined by meme-chip tools (Machanick Bailey, 2011) are illustrated. The resemblant conserved sequences, found by the CFSP algorithm are listed in the second column. In the third column, the position-specific scoring matrices, which are transformed from mutational information are listed. The similarity between meme motif and resemblant conserved sequence with PSSM format was calculated via a stamp motif comparison tool (Mahony Benos, 2007). The E-values for the similarity of those pairs is displayed in the fourth column. (D) One motif is selected in each group clustered by gkmsvm descriptors, and the corresponding motif found by the CFSP algorithm is listed below. (E) There are additional datasets (File No: ENCFF100GRL, ENCFF616IRT, ENCFF870CER, Target: SREBF1) collected from https://www.encodeproject.org. The top two motifs are selected in each file using meme tools, and the corresponding motifs found by our algorithm are listed below. [O] . -1

机译：图4：（a）一种保守序列，其发生在芯片-SEQ数据集中的46,264个结合位点峰值中的79倍。说明了这种保守序列的突变分布，其中'_'表示该碱度不变; del表示此基础丢失; INS X表示新的基础X插入此基础前面。（b）列出了几种重复的元素模式。（c）在第一栏中，示出了由MEME芯片工具（Machanick＆Bailey，2011）开采的前五个DNA主题。由CFSP算法发现的相应保守序列列于第二列中。在第三列中，列出了从突变信息转换的特定位置的评分矩阵。 MEME主题与PSSM格式的相似性与PSSM格式之间的相似性通过邮票图章比较工具（Mahony＆Benos，2007）计算。这些对相似性的电子值显示在第四列中。（d）在由GKMSVM描述符聚集的每个组中选择了一个图案，下面列出了CFSP算法的相应主题。（e）从https://www.encodeproject.org收集的，有附加数据集（文件no：cernff100grl，cenf616irl，conf8.20cer，target：srebf1）。使用MEME工具在每个文件中选择前两个图案，并且我们的算法发现的相应主题如下所示。

An Under-Sampling Algorithm Based on SVM

摘要

著录项

相似文献

相关主题

期刊订阅