Machine Learning and Data Mining in Pattern Recognition

Name: Machine Learning and Data Mining in Pattern Recognition | 7th International Conference, MLDM 2011, New York, NY, USA, August 30-September 3, 2011Proceedings
Brand: Springer
Price: 53.49 EUR
Availability: OnlineOnly

7th International Conference, MLDM 2011, New York, NY, USA, August 30-September 3, 2011Proceedings

Petra Perner(Editor)

Springer (Publisher)

Published on 12. August 2011

XII, 614 pages

E-Book

PDF with digital watermarking

System requirements

978-3-642-23199-5 (ISBN)

€53.49incl. 7% vat

System requirements

for PDF with digital watermarking

E-Book Single Licence

Available for download

Description

More details

Other editions

Content

Title
Preface
Organization
Table of Contents
Classification and Decision Theory
Quadratically Constrained Maximum a Posteriori Estimation for Binary Classifier
Introduction
Maximum a Posteriori-Based Classifier
Proposed Method
Linear Model and Training Method
A Unified Characterization of LSR and SVM
A New Classifier
Construction of GQCM Classifier
Experiments
Experiment Using Artificial Samples
Performance with UCI Data Sets
Discussion
Conclusions and Future Work
References
Hubness-Based Fuzzy Measures for High-Dimensional k-Nearest Neighbor Classification
Introduction
Related Work
Hubness-Weighted kNN
Fuzzy Nearest Neighbor Algorithm
Proposed Hubness-Based Fuzzy Measures
Experimental Evaluation
UCI Data Sets
ImageNet Data
Conclusions and Future Work
References
Decisions: Algebra and Implementation
Introduction
Decision Algebra
Decision Functions
Learning and Deciding
Auxiliary Decision Function Operations
Decision Lattices
The ``More Accurate'' Relations
Approximating Decision Functions
Experiments
Implementation Details
Decision Graph Sizes
k-Approximated Decision Graphs
Related Work
Conclusions and Future Work
References
Smoothing Multinomial Naïve Bayes in the Presence of Imbalance
Introduction
Related Work
Random OverSampling Expected Smoothing
ROSE Smoothing Background
ROSE Smoothing Approach
Experiments
Experiment 1: Standard Datasets
Experiment 2: Class Prior Controlled Data Sets
Conclusion
References
ACE-Cost: Acquisition Cost Efficient Classifier by Hybrid Decision Tree with Local SVM Leaves
Introduction
Preliminaries and Related Work
Computing Average Test Cost in Decision Tree and SVM
Cost Efficient SVM
Cost Efficient Decision Trees
Preprune and Postprune
ACE-Cost Approach: The Hybrid Decision Tree with Local SVM Leaves
Decision Tree Sketch
Postpruning with Local SVM
Feature Selection at Local SVM Leaves
Experimental Results
Performance Comparison on Standard Dataset
Synthetic Dataset
A Practical Application with Dependent Cost
Discussion and Future Work
Conclusion
References
Informative Variables Selection for Multi-relational Supervised Learning
Introduction
Approach Illustration
Evaluation Criterion
Grid Optimisation
Experiments
Protocol
Artificial Datasets
Stulong Dataset
Conclusion
References
Separability of Split Value Criterion with Weighted Separation Gains
Introduction
Split Criteria
Weighting Separability Gains
The Analysis
Comparison of the Split Criteria
Analysis of the $a$ Parameter
Conclusions
References
Granular Instances Selection for Fuzzy Modeling
Introduction
Background Studies
Instance Selection
Fuzzy Modeling
The Proposed Methodology
The Framework of the Proposed GIS Fuzzy Modeling Methodology
Granular Instances Selection
Evaluation
Experimental Studies
Conclusion
References
Parameter-Free Anomaly Detection for Categorical Data
Introduction
Related Work
Our Proposed Method
Problem Definition
Outlier Detection as an Optimization Problem
Outlier Factor
Update of Entropy and Weight
The ITB Methods and Approximate Optimization
Experimental Results
Effectiveness Test
Efficiency Test
Conclusion
References
Fuzzy Semi-supervised Support Vector Machines
Introduction
Fuzzy Semi-supervised Support Vector Machines Approach
Experimental Results
Datasets
Experimental Design
The Classification Tasks
Evaluation Procedure
Parameter Optimization
Experiments and Results
Conclusion
References
GENCCS: A Correlated Group Difference Approach to Contrast Set Mining
Introduction
Related Work
Problem Definition
Mutual Information and All Confidence
Correlated Group Difference
Background
Data Format
Search for Quantitative Contrast Sets
Distribution Difference
Our Proposed Approach
Tests for Significance
Comparison of Contrasting Groups
Discretization
Mining Correlated Group Differences
Experimental Results
Performance of GENCCS
Effect of Mutual Information and All Confidence
Conclusion
References
Collective Classification Using Heterogeneous Classifiers
Introduction
Background
Notation
Collective Classification
Collective Classification Using Heterogeneous Classifiers
Related Work
Experimental Setup
Datasets
Sampling
Classification Methods
Experimental Results
Analysis of Average Local Accuracy Values
Performance of Different Classifiers
Performance of Classifier Combination
Discussion
References
Spherical Nearest Neighbor Classification: Application to Hyperspectral Data
Introduction
Mapping of Images to a Hypersphere
Tangent Space and Manifolds
Exponential and Log Maps
Spherical Metrics
Spherical Geodesic and Mahalanobis Metrics
Spherical Discriminant Adaptive Nearest-Neighbor Classifier
Experiments
Data
Results
Conclusions
References
Adaptive Kernel Diverse Density Estimate for Multiple Instance Learning
Introduction
Maximum Diverse Density
Kernel Density Estimate
Kernel Density Estimate for MIL
Kernelized Diverse Density Estimate of Positive Bags
Kernel Density Estimate of Negative Bags
Objective Function
Optimization and Implementation
Experiments
Conclusions
References
Boosting Inspired Process for Improving AUC
Introduction
Boosting Inspired Process AUCBoost
Weighted AUC (WAUC)
Experiments
Base Learning Algorithm: C4.4
Base Learning Algorithm: Naive Bayes
Conclusions and Future Work
References
Theory of Learning
Investigation in Transfer Learning: Better Way to Apply Transfer Learning between Agents
Introduction
Reinforcement Learning
The Q-Learning Algorithm
Accelerating Reinforcement Learning through Heuristics
Case Based Reasoning and Transfer Learning
Combination of the Techniques: Transfer Learning with Case Based Reasoning and Reinforcement Learning
The Transfer Learning Experience
Conclusion
References
Exploration Strategies for Learned Probabilities in Smart Terrain
Probabilistic Smart Terrain
Learned Probabilities
Object Categories and Prior Knowledge
Bayesian Parameter Learning
Exploration Strategies and Benchmarks
Defining Information Gain
Defining a Simple Two-Object Case
Estimating Distance Traveled and Error
Normalizing Error over Categories Based on Prevalence
Estimating Information Gain
A Simple Example of Information Gain
Creating the Influence Map
Inverse Falloff
Cumulative Effect of Objects
Updating the Influence Map When Information Is Learned
Benchmark Testing
Different Levels of Prior Knowledge
Different Category Prevalence
Significantly Closer Objects
Aggregate Influence
Empirical Demonstration of Learning
Evaluating Ability to Move Towards the Best Objects
Evaluating the Value of Learning
Conclusions and Ongoing Research
Ongoing Research
References
Sensitivity Analysis for Weak Constraint Generation
Introduction
Abstract Problem Statement
Formulation of the Multi-constraint Planning Problem
Examples
Solution Approach
Plan Changes
Sensitivity Analysis, Simulation, and Clustering
Strategic Release Planning: A Multi-constraint Planning Problem
Simulation Overview
Data Generation and Pre-processing
Sensitivity Analysis and Clustering
Simulation and Clustering Experiment
Use Case Scenario
Discussion
Limitations
Conclusions and Future Work
References
Dictionary Learning Based on Laplacian Score in Sparse Coding
Introduction
Related Work
Laplacian Score for Dictionary Learning
Experiments
Experiment Datasets
Experiment Setup
Experiment Result
Conclusion
References
Clustering
A Practical Approach for Clustering Transaction Data
Introduction
Clustering Algorithm
Objective Function
Clustering Procedure
Empirical Evaluation
Comparing Algorithms
Quality Measure
Experiments on Synthetic Data
Experiments on Real-World Data
Conclusion
References
Hierarchical Clustering with High Order Dissimilarities
Introduction
Dissimilarity Increments
Dissimilarity Increments Distribution
Hierarchical Clustering Algorithm
Algorithm
Minimum Description Length Criterion
Algorithm Analysis
Experimental Results
Datasets
Parameter Selection
Exprimental Results and Discussion
Conclusions
References
Clust-XPaths: Clustering of XML Paths
Introduction
Related Work
XML Structure Clustering
XML Content Clustering
Clust-XPaths: Our Approach
Thesaurus
Paths Matrix
Clustering
Experiments
Evaluation Prototype
Test Collection and Results
Conclusion
References
Comparing Clustering and Metaclustering Algorithms
Introduction
Related Work
Meta-clustering
Bagged Clustering
Majority Voting
Graph Partitioning
Cluster Validation Techniques
Experimental Results
Conclusions
References
Applications in Medicine
Detection of Phenotypes in Microarray Data Using Force-Directed Placement Transforms
Introduction
Force-Directed Placement Strategies
Modified Force-Directed Placement Transform
Application to Model Data
Feature Selection Algorithm
Results on Model Data
Application to Real Data
Discussion
References
On the Temporal Behavior of EEG Recorded during Real Finger Movement
Introduction
Exploiting the Temporal Information
Hidden Markov Models for BCI
Conditional Random Fields for BCI
Methods for Temporal Classification of Self-Paced EEG
Method I
Method II
Method III
Experiments
Results
Discussion and Conclusion
References
A Machine Learning and Data Mining Framework to Enable Evolutionary Improvement in Trauma Triage
Problem Description
A Model for Intelligent Triage Support
Data
Realtime Decision Support through Machine Learning
Experiments: Methodology and Results
Discussion
Mining for Deeper Understanding
Knowledge Frontiers
Related Work
Algorithm Description
Experimental Methodology
Results and Discussion
Conclusion and Future Work
References
A Decision Support System Based on the Semantic Analysis of Melanoma Images Using Multi-elitist PSO and SVM
Introduction
Formulation of the Problem of the Semantic Analysis of the Melanoma Malignum
The MEPSO Algorithm
The Basic Concept of the SVM Classifier
Experimental Results
Conclusion
References
WebMining/Information Mining
Authorship Similarity Detection from Email Messages
Introduction
Stylistic Features
Authorship Similarity Detection
Frequent Pattern Matching
Style Differentiation
Detection Algorithm
Baseline Comparison Methods
The Enron Email Corpus
Experiment Results
Conclusion
References
An Investigation Concerning the Generation of Text Summarisation Classifiers Using Secondary Data
Introduction
Related Work
Problem Definition
Classifier Generation Using Secondary Data
The SAVSNET Application
Evaluation
Conclusion
References
Comparing the One-vs-One and One-vs-All Methods in Benthic Macroinvertebrate Image Classification
Introduction
Method
Linearly Separable Case
Linearly Non-separable Case
Nonlinear Support Vector Machines
One-vs-All
One-vs-One
Experimental Tests
Data Description and Test Arrangements
Results
Discussion
References
Incremental Web-Site Boundary Detection Using Random Walks
Introduction
Preliminaries
k-Means Clustering
Clustering Based on Random Crawling
Experiments
Data Set
Evaluation Criteria
Results
Discussion
References
Discovering Text Patterns by a New Graphic Model
Introduction
The Method
Defining the Task
The Model
Properties of the Model
Finding &c1*, ., cN*&
Complexities
Three Tasks
Descriptions
Definition of Task 1
Definition of Task 2
Definition of Task 3
Empirical Results
Experiments Set Up
Results on the First Task
Results on the Second Task
Results on the Third Task
Related Researches and More Comparisons
Conclusions
References
Topic Sentiment Change Analysis
Introduction
Related Works
Solution Overview
Topic-Level Sentiment Analysis
Topic Content Division
Topic Sentiment Evaluation
Sentiment Change Analysis
Time Period Partition
Cause Identification
Experiments
Experiment Setup
Topic Identification Results
Sentiment Change Analysis Results
Conclusions
References
Adaptive Context Modeling for Deception Detection in Emails
Introduction
Related Work
Proposed Deception Detector
Prediction by Partial Matching
Generalized Suffix Tree Data Structure
Adaptive Context Modeling
Experimental Results
Conclusions
References
Contrasting Correlations by an Efficient Double-Clique Condition
Introduction
Preliminaries
Correlation Based on k-Way Mutual Information
Problem of Mining Correlation Contrast Sets
Detecting Correlation Contrast Sets with Double-Clique Methods
Excluding Useless Itemsets with Anti-correlation Graphs
Enumerating Cliques in Undirected Graph
Additional Support Constraint
Algorithm for Extracting Correlation Contrast Sets
Experimental Results
Contrasted Databases
Extracted Correlation Contrast Sets
Computational Performance
Conclusion and Further Research
References
Machine Learning and Image Mining
Estimating Image Segmentation Difficulty
Introduction
Mathematical Background
Details of the Approach
Transformation of Difficulty Measures
Feature Extraction
Modeling Process
Experiments and Results
Building the Model with Labeled Data
Applying the Model to Additional Data
Conclusion and Future Work
References
Mining Spatial Trajectories Using Non-parametric Density Functions
Introduction
Related Work
Trajectory Mining with Density Functions
Trajectory Density Estimation
The DENTRAC Trajectory Clustering Algorithm
Hill Climbing Procedure
Complexity of DENTRAC
Post Analysis for Trajectory Clusters
Experimental Evaluations
Datasets
Results for the Oldenburg Traffic Data
Post Analysis by Cluster Average Density and the Density of Density Attractors
Results of Atlantic Hurricane Tracks Data
Conclusion and Future Works
References
Exploring Synergetic Effects of Dimensionality Reduction and Resampling Tools on Hyperspectral Imagery Data Classification
Introduction
Methodology
Data Preprocessing
Classification
Experimental Set-Up
Classification Performance Measures
Results and Discussion
Conclusions and Further Extensions
References
A Comparison between Haralick´s Texture Descriptor and the Texture Descriptor Based on Random Sets for Biological Images
Introduction
Texture Descriptors
Haralick´s Texture Descriptor
Texture Descriptor Based on Random Sets
Material and Application
Results
Discussion
Conclusion
References
Time Series and Frequent Item Set Mining
Unsupervised Discovery of Motifs under Amplitude Scaling and Shifting in Time Series Databases
Introduction
Background
Related Work
Algorithm
Thresholding
Experimental Results
Synthetic Data
Real-World Data with Known Motifs
Discord Discovery in Synthetic Data
Empirical Results and Discussion
Spurious Motifs
Conclusion and Future Work
References
Static Load Balancing of Parallel Mining of Frequent Itemsets Using Reservoir Sampling
Introduction
Notation
The Lattice of All Itemsets
Sampling Methods
Database Sample
The Reservoir Sampling Algorithm
Error of the Estimation of the Size of a Union of PBECs
Summary of the Previous Two Methods
Proposal of a New DM Parallel Method
Detailed Description of Phase 1
Detailed Description of Phase 2
Detailed Description of Phase 3
Detailed Description of Phase 4
The Parallel-FIMI-Reservoir Method
Experimental Evaluation
Evaluation of the Speedup
Conclusion and Future Work
References
GA-TVRC: A Novel Relational Time Varying Classifier to Extract Temporal Information Using Genetic Algorithms
Introduction
Background
Classification Method
Genetic Algorithms
Related Work
Genetic Algorithm Enhanced Time Varying Relational Classifier (GA-TVRC)
Training Phase
Validation Phase Using Evolutionary Strategies
Test Phase
Experimental Results
Datasets
Methodology
Results and Analysis
Conclusions and Future Work
References
Aspects of Machine Learning and Data Mining
Detection of Communities and Bridges in Weighted Networks
Introduction
Background and Motivation
Methodology
Fuzzy Clustering of Weighted Graphs
Fuzzy Clustering of Unweighted Graphs
Bridgeness Measure
Experimental Setup
Datasets
Edge Weights
Label Correspondence
Evaluation Metrics
Software
Results
Accuracy
Sensitivity Analysis
Bridgeness Analysis
Conclusion and Future Work
References
Techniques for Improving Filters in Power Grid Contingency Analysis
Introduction
Context
A Metric for Evaluating Filters
Resource-Aware Filter Combination
Multi-criteria Optimization
Related Work
Future Work
Conclusion
References
Author Index

System requirements

Save as PDF Copy link into clipboard

Schweitzer Fachinformationen

Machine Learning and Data Mining in Pattern Recognition

Description

More details

Other editions

Additional editions

Content

System requirements