位置：成果数据库 > 期刊 > 期刊详情页
大数据下的PAC-Bayesian学习理论综述
  • ISSN号：0253-2395
  • 期刊名称：山西大学学报(自然科学版)
  • 时间：2015.3.5
  • 页码：-
  • 分类：TP181[自动化与计算机技术—控制科学与工程;自动化与计算机技术—控制理论与控制工程]
  • 作者机构：[1]中国科学院自动化研究所,北京100190
  • 相关基金：国家自然科学基金重点项目（U1135005）
  • 相关项目：跨媒体协同处理与服务的理论和应用研究
作者： 杨雪冰|张文生|杨阳|
关键词： PAC-Bayesian界, 学习理论, 贝叶斯学习, 大数据, PAC-Bayesian bound, learning theory, Bayes learning, big data
中文摘要：

概率近似正确（PAC）是研究＂可学习＂的理论框架。近年来,研究人员融合贝叶斯方法与不依赖分布的PAC性能度量提出了所谓的PAC-Bayesian学习理论。该理论因其对于任意概念空间任意测度的先验均能给出泛化误差界而在人工智能不同领域的相关算法分析中得到广泛应用。文章综述了PAC-Bayesian学习理论的由来及其核心思想,进而结合大数据的特点,论述了PAC-Bayesian适合于大数据相关算法的理论分析。

英文摘要：

The theory of probably approximately correct（PAC）is a framework for the study of learnable.In recent years,researchers combined Bayesian method with distribution-free PAC guarantees and proposed so-called PAC-Bayesian learning theory.This theory can give generalization error bounds for an arbitrany prior measure on an arbitrary concept space,so it has been widely used in different fields of artificial intelligence to analyze related algorithms.This paper surveys the derivation of PAC-Bayesian learning theory and its core ideas.Further,considering the characteristics of big data,this paper discusses why PAC-Bayesian is useful for theoretical analysis of the related algorithms for big data.

同期刊论文项目
跨媒体协同处理与服务的理论和应用研究
期刊论文 88 会议论文 28
同项目期刊论文
Adaptive iterative learning control of non-linear MIMO continuous systems with iteration - varying i
Particle swam optimization with an aging leader and challengers
Multiple Populations for Multiple Objectives : A Coevolutionary Technique for Solving Multiobjective
Single Image Super-Resolution Using Combined Total Variation Regularization by Split Bregman Iterati
On Improving Aggregate Recommendation Diversity and Novelty in Folksonomy-based Systems
Evolutionay Strategy based on Mixture Gaussian Models
A Novel Shape Matching Method Based On Feature Points
具有迭代初始误差的高相对度线性离散系统的迭代学习控制理论
基于支持向量机的不平衡数据分类的改进欠采样方法
Weighted attentional blocks for probalilistic object tracking
Discriminant Graph Based Linear Embedding
Social Sensing Enanced Time Ruler for Real-Time Bus Service
WTrack: HMM-based Walk Pattern Recognition and Indoor Pedestrian Tracking Using Phone Inertial Senso
Differential Evolution With Two-Level Parameter Adaption
Image Magnification Based on Classified Training Set
A particle swarm optimization using local stochastic search and enhancing diversity for continuous o
Robust iterative learning control with rectifying action for nonlinear discrete time-delayed systems
Improved heuristic equivalent search algorithm based on Maximal Information Coefficient for Bayesian
Differential evolution with two-level parameter adaptation.
基于贝叶斯服务依赖图的错误定位
基于随机谱梯度的在线学习
Localization and Recognition of Dynamic DigitGestures for Smart TV Systems
EnergyConsumption Prediction based on Time-Series Models for CPU-Intensive Activitiesin the Cloud
AMapReduce Reinforced Distributed Sequential Pattern Mining Algorithm
A novel pheromone-based evolutionary algorithmfor solving degree-constrained minimum spanning tree p
Space-based initialization strategy for particle swarm optimization
MICHAC:Defect Prediction via Feature Selection based on Maximal InformationCoefficient with Hierarch
基于Dropout深度网络的两步图像标注算法
A set-based locally informed discrete particleswarm optimization
Virtual Power Meter Supported Power Consumption Prediction of Web Services
GreenOCR: An Energy-efficient Optimal Clustering Routing Protocol
Adaptive ILC algorithms of nonlinear continuous systems with non-parametric uncertainties for non-re
Queuing Theory based Efficiency Optimization of Business Process for Academic Community Cloud
LOTMAP: Learning to Maximize Top-N Recommendation with Mean Average Precision
Global Coupled Learning and Local Consistencies Ensuring for Sparse-based Tracking
A TV-L1 Based Nonrigid Image Registration by Coupling Parametric and Non-Parametric Transformation
改进的OWL-QN方法解稀疏logistic回归问题
基于高阶奇异值分解和均方差迭代的图像去噪
Multiple Populations for Multiple Objectives: A Coevolutionary Technique for Solving Multiobjec
Discriminative subspace learning with sparse representation view-based model for robust visual track
Robust iterative learning control with rectifying action for nonlinear discrete time - delayed syste
Adaptive ILC for tracking non-repetitive reference trajectoryof 2-D FMM under random boundary condit
Iterative learning control for lineardiscrete-time systems with high relative degree under initial s
A unified adaptive control framework ofnon-parameterized nonlinear continuous systems for repetitive
Discriminative Object Tracking via Sparse Representation and Online Dictionary Learning
AParticle Swarm Optimization Using Local Stochastic Search for ContinuousOptimization
AutomaticallyConstructing Course Dependence Graph based on Association Semantic Link Model
ANovel Source-Location Anonymity Protocol in Surveillance Systems
Image automatic annotation via multi-view deep representation
Discriminative Reverse Sparse Tracking Via Weighted Multi-task Learning
Structure Learning for Weighted Networks Based on Bayesian Nonparametric Models
FastService Process Fragment Indexing and Ranking
基于深度学习框架的隐藏主题变量图模型
ARG-based segmentation of radioactive ray image in contraband check
A total variation based nonrigid image registration by combining parametric and non-parametric trans
Iterative Learning Control for Two Dimensional Discrete Systems with Fornasini–MarchesiniModel
Learning a Coupled Linearized Method in Online Setting.
PowerConsumption Prediction of Web Services for Energy-efficient Service Selection
基于热度矩阵的微博热点话题发现
引入测评机制的综合学习粒子群优化算法
融合加权动态权威度和兴趣度的专家推荐方法
基于退火过渡采样的无向主题模型学习方法
一种基于k近邻图的稀有类检测算法？
期刊信息
  • 《山西大学学报：自然科学版》
  • 北大核心期刊（2011版）
  • 主管单位:山西省教育厅
  • 主办单位:山西大学
  • 主编：杨斌盛
  • 地址：太原市坞城路92号
  • 邮编：030006
  • 邮箱：xbbjb@sxu.edu.cn
  • 电话：0351-7010455
  • 国际标准刊号：ISSN：0253-2395
  • 国内统一刊号：ISSN：14-1105/N
  • 邮发代号:22-42
  • 获奖情况:
  • 边疆七年获山西省一级期刊荣誉（1993-1999）
  • 国内外数据库收录:
  • 俄罗斯文摘杂志,美国化学文摘（网络版）,英国动物学记录,中国中国科技核心期刊,中国北大核心期刊（2004版）,中国北大核心期刊（2008版）,中国北大核心期刊（2011版）,中国北大核心期刊（2014版）
  • 被引量:5651