RGAN-EL: A GAN and ensemble learning-based hybrid approach for imbalanced data classification     

《Information processing & management》2023,60(2):103235

Imbalanced sample distribution is usually the main reason for the performance degradation of machine learning algorithms. Based on this, this study proposes a hybrid framework (RGAN-EL) combining generative adversarial networks and ensemble learning method to improve the classification performance of imbalanced data. Firstly, we propose a training sample selection strategy based on roulette wheel selection method to make GAN pay more attention to the class overlapping area when fitting the sample distribution. Secondly, we design two kinds of generator training loss, and propose a noise sample filtering method to improve the quality of generated samples. Then, minority class samples are oversampled using the improved RGAN to obtain a balanced training sample set. Finally, combined with the ensemble learning strategy, the final training and prediction are carried out. We conducted experiments on 41 real imbalanced data sets using two evaluation indexes: F1-score and AUC. Specifically, we compare RGAN-EL with six typical ensemble learning; RGAN is compared with three typical GAN models. The experimental results show that RGAN-EL is significantly better than the other six ensemble learning methods, and RGAN is greatly improved compared with three classical GAN models.  相似文献   

基于反向随机投影的神经网络改进算法          下载免费PDF全文

许邓艳卢民荣 《福建工程学院学报》2020,(4):358-364

提出基于反向随机局部投影的神经网络效率改进算法，通过降低深度学习中的网络规模，重点解决了从“局部连接”到“全连接”和随机节点抽取时输入端节点信息丢失的问题，从而提升网络的效率。在算法中设置缩减参数，提升了算法的可伸展性，以适用于不同数据集的学习。通过数据集ISO-LET进行实验，结果表明，基于反向随机局部投影的神经网络效率改进算法的准确率、效率分别平均提升了3.48%和105.21%?在迭代20次的实验中进行了缩减参数调节实验，当参数设置为1.4时其准确率则优于传统全连接网络2.61%，效率提升了272.78%，具有明显的优势。  相似文献   

The effect of toxicity on COVID-19 news network formation in political subcommunities on Reddit: An affiliation network approach     

《International Journal of Information Management》2021

Political polarization remains perhaps the “greatest barrier” to effective COVID-19 pandemic mitigation measures in the United States. Social media has been implicated in fueling this polarization. In this paper, we uncover the network of COVID-19 related news sources shared to 30 politically biased and 2 neutral subcommunities on Reddit. We find, using exponential random graph modeling, that news sources associated with highly toxic – “rude, disrespectful” – content are more likely to be shared across political subreddits. We also find homophily according to toxicity levels in the network of online news sources. Our findings suggest that news sources associated with high toxicity are rewarded with prominent positions in the resultant network. The toxicity in COVID-19 discussions may fuel political polarization by denigrating ideological opponents and politicizing responses to the COVID-19 pandemic, all to the detriment of mitigation measures. Public health practitioners should monitor toxicity in public online discussions to familiarize themselves with emerging political arguments that threaten adherence to public health crises management. We also recommend, based on our findings, that social media platforms algorithmically promote neutral and scientific news sources to reduce toxic discussion in subcommunities and encourage compliance with public health recommendations in the fight against COVID-19.  相似文献   

Digital Module 16: Longitudinal Data Analysis     

Jeffrey R. Harring  Tessa L. Johnson 《Educational Measurement》2020,39(3):137-138

In this digital ITEMS module, Dr. Jeffrey Harring and Ms. Tessa Johnson introduce the linear mixed effects (LME) model as a flexible general framework for simultaneously modeling continuous repeated measures data with a scientifically defensible function that adequately summarizes both individual change as well as the average response. The module begins with a nontechnical overview of longitudinal data analyses drawing distinctions with cross-sectional analyses in terms of research questions to be addressed. Nuances of longitudinal designs, timing of measurements, and the real possibility of missing data are then discussed. The three interconnected components of the LME model—(1) a model for individual and mean response profiles, (2) a model to characterize the covariation among the time-specific residuals, and (3) a set of models that summarize the extent that individual coefficients vary—are discussed in the context of the set of activities comprising an analysis. Finally, they demonstrate how to estimate the linear mixed effects model within an open-source environment (R). The digital module contains sample R code, diagnostic quiz questions, hands-on activities in R, curated resources, and a glossary.  相似文献   

Measuring Textbook Content Coverage: Efficient Content Analysis with Lesson Sampling     

Jiahui Zhang  Leland S. Cogan  William H. Schmidt 《Educational Measurement》2020,39(2):74-84

This study addresses measurement issues around a standards-based content analysis of mathematics textbooks’ coverage of standards for use in large-scale monitoring of standards implementation as proposed in a 2013 report by the National Research Council. An earlier study produced an exhaustive content analysis of textbooks using the 2012 Common Core State Standards for Mathematics. This yielded an accurate and reliable portrait of a textbook's coverage of standards. However, such an in-depth analysis is not feasible for large-scale standards-implementation monitoring in which a large number of textbooks may need to be analyzed. To provide such a portrait with sufficient accuracy while also substantially reducing the associated resources needed to produce such a portrait, a simulation study was conducted with the exhaustively coded database to compare different sampling schemes. Results indicated that sampling 1 day from each week and coding the corresponding lessons led to sufficiently accurate representations of the overall content of the textbook. The results provide empirical evidence for large-scale standards-based content analyses of mathematics textbooks for monitoring standards implementation which could be adapted for other subject areas.  相似文献   

2001-2017年尼泊尔中部地区滑坡变化及其影响因素     

孙国庆  陈方  于博  王宁 《中国科学院大学学报》2020,37(3):308-316

滑坡是造成尼泊尔巨大经济损失和人员伤亡的主要地质灾害。目前，遥感方法对滑坡信息的识别、提取和监测具有巨大的潜力。基于随机森林模型的平均不纯度减少算法，选用NDBI、MNDWI、NDVI和红波段4个参数对Landsat遥感数据进行滑坡信息的初提取，同时借助目视解译得到滑坡的精提取结果。采用数据格网的方法将研究区划分为0.05°×0.05°的滑坡格网，并在不同土地覆盖类型中，对滑坡与降雨和温度进行偏相关分析。研究表明：1）尼泊尔中部地区的滑坡主要分布在海拔1 000~2 500 m，坡度20°~40°的范围内；2）相比其他土地覆盖类型，林地中滑坡数目与降雨的偏相关系数为0.671 5，而在草地中滑坡数目与温度的偏相关性最强，其偏相关系数为0.436 1；3）研究区片麻岩、页岩分布广泛，加之地震、断裂带、人类活动的影响，破坏边坡的稳定性，极易造成滑坡灾害。  相似文献   

随机需求条件下库存控制策略研究   总被引：1，自引：0，他引：1  

张杰 《科技创业月刊》2007,20(11):78-79

在库存控制问题中,客户的需求常常带有一定的随机性,分析了随机需求下的库存成因,提出了随机需求下的库存控制策略。  相似文献   

Introductory statistics in the garden          下载免费PDF全文

John C. Wagaman 《Teaching Statistics》2017,39(2):52-56

This article describes four semesters of introductory statistics courses that incorporate service learning and gardening into the curriculum with applications of the binomial distribution, least squares regression and hypothesis testing. The activities span multiple semesters and are iterative in nature.  相似文献   

基于半监督学习与CRF的应急预案命名实体识别     

刘彤  魏静  倪维健  陈思源 《教育技术导刊》2020,19(3):35-38

传统基于统计的命名实体识别方法存在需要大量人工标注的缺陷,导致识别准确率较低。为了提升识别效果,提出一种基于条件随机场的半监督学习方法（S-CRF）对命名实体进行识别。该方法将实体识别看作序列标注问题,对少量数据进行人工标注并构建实体集,通过K-means聚类算法选取有代表性的未标注数据文本进行自动标注,采用条件随机场对语料进行训练测试。选取中文应急预案文档进行实验,该方法在各个标签上的识别效果分别达到93.52%、93.04%、95.81%。实验结果表明,该方法优于传统规则方法,能有效提高应急预案命名实体的识别效果。  相似文献   

信息技术教育的现状与展望   总被引：1，自引：1，他引：1  

王国钧  吴孝燕  邵斌 《湖州师范学院学报》2003,25(3):112-116

在对浙江省湖州、嘉兴两市部分中学的信息技术学科教学进行抽样调查的基础上，从课程设置、教材使用、教学设备、师资力量等方面分析了中学信息技术学科教学的现状，并利用统计数字揭示了该学科教学目前存在的问题，最后对该学科教学的发展前景作了进一步的探讨。  相似文献