Classification under Data Contamination with Applications

来源 :上海交通大学 | 被引量 : 0次 | 上传用户：jack332904910

【摘要】

：

【作者】

：

Donghui Yan Aiyou Chen Peng Gong

【机构】

：

University of Massachusetts Dartmouth,MA 02747,USA

【出处】

：

上海交通大学

【发表日期】

：

2016年4期

【关键词】

：

Data contamination data quality distribution drifts classification machine learn

下载到本地 , 更方便阅读

下载此文赞助VIP

声明 : 本文档内容版权归属内容提供方 , 如果您对本文有版权争议 , 可与客服联系进行内容授权或下架

论文部分内容阅读

　　Data contamination refers to the phenomenon where part of the data is randomly replaced by data generated from an unknown distribution.It applies to a wide range of real world problems related to data quality,such as label noise,drifts in data distribution,and random errors in data entry etc.The impact of data contamination to the accuracy of pattern classification is studied and an asymptotic error bound is established.Several applications will be discussed for which the model gives insights.

其他文献

电池原料Co3O4合成大颗粒单晶LiCoO2关键技术研究

随着国际钴价持续走低，电池正极材料钴酸锂产品行业内利润逐步降低，竞争加剧，客户在要求常规钴酸锂产品加工成本不断降低的同时，对于电池原料的性能要求不断提高，对其充放电容量、放电平台、循环性能及安全性能的要求越来越苛刻。本文从高温助熔剂的原理出发,选择合适的助剂,采用高温固相法合成出了大颗粒、高振实密度的单晶LiCoO2.以充放电电流为0.1C(15mA/g)在3.0～4.3V电压范围内,其首次充电容

会议

电池正极材料钴酸锂高温固相法颗粒粒度充放电容量

ADDRESSING UNMEASURED CONFOUNDING IN COMPARATIVE EFFECTIVENESS RESEARCH

While the use of real world/observational/big data for comparative effectiveness analyses has grown in recent years,causal inference from such data typically relies on the unprovable assumption of no

会议

Unmeasured ConfoundingComparative EffectivenessBayesian Twin RegressionCausal

An Optimal Method for Covariate Balancing and Its Properties

This article introduces a new randomization procedure to improve the covariate balance across treatment groups.Covariate balance is one of the most important concerns for successful comparative studie

会议

Causal inferencecovariate-adaptive randomizationclinical trialMahalanobis dis

Sequential Multiple Testing for Variable Selection in High Dimensional Linear Model

Covariance test(Lockhart et al.2014)provided p-values for all variables that enter into a linear model sequentially along a lasso solution path.Using these p-values to select a model with inferential

会议

Covariance testFalse Discovery RateLASSOSequential Multiple TestingOrdered H

Optimal control and additive perturbations help in estimating ill-posed and uncertain dynamical syst

Ordinary Differential Equations(ODE)are routinely calibrated on real data for estimating unknown parameters or for reverse-engineering.Nevertheless,standard statistical technics can give disappointing

会议

Ordinary differential equationoptimal controlparametric estimationsemi-parame

PHASE TRANSITIONS FOR CLUSTERING,SIGNAL RECOVERY AND HYPOTHESIS TESTING

We consider a two-class clustering problem,where we observe a large number of features but only a small fraction of them contribute to the class labels.There are three problems here: 1)Test whether th

会议

High dimensional dataclusteringhypothesis testingsignal recoveryphase transi

Testing the Capital Asset Pricing Model with Economic Regime Shifts

This paper develops a stochastic model for asset returns to test the conventional Capital Asset Pricing Model.Assuming the economic regime over time shifts according to a Markov chain embedded in econ

会议

Hidden Markov ModelEconomic IndicatorsThe Capital Asset Pricing ModelBayesian

Joint Models for Mixed Types of Longitudinal and Survival Data

Joint models for longitudinal and survival data have received much attention in recent years.Here we consider joint models for HIV vaccine studies.In these studies,joint models are complicated by the

会议

censoringHIV vaccinejoint modelslongitudinal datasurvival data

Semiparametric monitoring test based on clustered data

Due to factors such as climate change,forest fire,plague of insects on lumber quality,it is important to update procedures in American Society for Testing and Materials(ASTM)Standard D1990(adopted in

会议

BootstrapClusterComposite likelihoodMonitoring testMultiple sample

Gradient-induced Model-free Variable Selection

Variable selection is central to sparse modeling,and many methods have been proposed under various model assumptions.In this talk,we will present a model-free variable selection method that allows for

会议

Classification under Data Contamination with Applications

与本文相关的学术论文