Discovery Science
14th International Conference, DS 2011, Espoo, Finland, October 5-7, Proceedings
Published on 30. September 2011
XIII, 381 pages
978-3-642-24477-3 (ISBN)
Description
This book constitutes the refereed proceedings of the 14th International Conference on Discovery Science, DS 2011, held in Espoo, Finland, in October 2011 - co-located with ALT 2011, the 22nd International Conference on Algorithmic Learning Theory. The 24 revised full papers presented together with 5 invited lectures were carefully revised and selected from 56 submissions. The papers cover a wide range including the development and analysis of methods for automatic scientific knowledge discovery, machine learning, intelligent data analysis, theory of learning, as well as their application to knowledge discovery.
More details
Other editions
Additional editions
Tapio Elomaa | Jaakko Hollmen | Heikki Mannila
Discovery Science
14th International Conference, DS 2011, Espoo, Finland, October 5-7, Proceedings
Book
09/2011
1st Edition
Springer
€53.49
Shipment within 7-9 days
Content
- Title Page
- Preface
- Organization
- Table of Contents
- On the Expressive Power of Deep Architectures
- Learning from Label Preferences
- Introduction
- Label Ranking
- Label Ranking by Pairwise Comparison
- Pairwise Classification
- Label Ranking by Pairwise Comparisons
- Combining Predicted Preferences into a Ranking
- LPC for Generalized Classification Problems
- Multilabel Classification
- Ordered and Hierarchical Classification
- Theoretical Foundations
- Classification
- Label Ranking
- Position Error
- Complexity
- Training Time
- Prediction Time
- Memory Requirements
- Conclusions and Outlook
- References
- Information Distance and Its Extensions
- Information in One Object
- Information Distance between Two Objects
- Informatin Distance among Many Objects
- Information Distance for Partial Matching
- Discussions
- References
- Models for Autonomously Motivated Exploration in Reinforcement Learning
- References
- Optimal Estimation
- Modeling Problem
- Models and Estimators
- Monotone Instance Ranking with mira
- Introduction
- Instance Ranking
- Monotone Instance Ranking
- A Monotone Scoring Function
- Example
- Experiments
- Artificial Data
- Real Data
- An Improved Monotone Ranking Function
- Additional Experiments
- Conclusion and Further Research
- References
- MOA-TweetReader: Real-Time Analysis in Twitter Streaming Data
- Introduction
- Real-Time Twitter Analysis Framework
- MOA-TweetReader
- MOA-TweetReader Feature Generation Filter
- Change Detection
- Adaptive Frequent Item Miner for Data Streams
- Applications
- Twitter Term Frequency Detection
- Twitter Sentiment Analysis
- Related Work
- Conclusions
- References
- Application of Semantic Kernels to Literature-Based Gene Function Annotation
- Introduction
- Related Work
- System Description
- Kernel Classifiers
- Regularized Linear Classifiers
- Kernelization
- Class Imbalance Handling
- Hyperparameter Tuning
- Latent Topic Kernels
- pLSA Background
- pLSA Kernels
- Experiments
- Classifier Comparison
- Kernel Comparison
- Comparison with Existing Methods
- Fast Learning, Prediction and Cross-Validation
- Conclusion and Future Work
- References
- "Tell Me More": Finding Related Items from User Provided Feedback
- Introduction
- Problem Statement
- Related Work
- Algorithm
- Graph Construction
- Random Walk with Restarts
- Tell Me More
- Experiments
- Text Documents
- Binary Data
- Numerical Data
- Discussion and Conclusion
- References
- MEI: Mutual Enhanced Infinite Generative Model for Simultaneous Community and Topic Detection
- Introduction
- Mutual Enhanced Infinite Generative Model
- Mutual Enhanced Generative Model
- Mutual Enhanced Infinite Generative Model
- Chinese Restaurant Process Metaphor
- Model Learning via Gibbs Sampling
- Sampling Equations
- Parameter Estimation Algorithm
- Hyper-parameter Setting
- Experiments
- Evaluation Criterion
- Baseline Models
- Dataset
- Performance Study
- Select the Number of Communities and Topics
- Case Study
- Conclusion and Future Work
- References
- A Methodology for Mining Document-Enriched Heterogeneous Information Networks
- Introduction
- Related Work
- Proposed Methodology
- Constructing Feature Vectors from Text Documents
- Constructing Structural-Context Feature Vectors with Personalized PageRank
- Combining Feature Vectors
- Efficient Classification with PageRank-Based Centroid Classifier
- VideoLectures.net Categorization Case Study
- Dataset
- Results of Text Mining and Diffusion Kernels
- Results of the Proposed Methodology
- Notes on Time and Space Complexity
- Conclusions and Future Work
- References
- Multiple Hypothesis Testing in Pattern Discovery
- Introduction
- Statistical Significance Testing in Data Mining
- Multiple Hypothesis Testing with Randomization
- Proof of Theorem 1
- Empirical p-Values
- Marginal Probabilities as Test Statistic
- Related Work
- Experiments
- Frequent Itemsets
- Frequent Subgraphs
- Discussion and Conclusions
- References
- A Parameter-Free Method for Discovering Generalized Clusters in a Network
- Introduction
- Related Work
- Problem Definition
- AutoPart: An Existing MDL-Based Clustering Method
- Proposed Method
- Cost Functions
- Search Algorithm
- Experiments
- Experimental Setting
- Synthetic Datasets
- The Enron Dataset
- Conclusions and Future Work
- References
- Detecting Anti-majority Opinionists Using Value-Weighted Mixture Voter Model
- Introduction
- Opinion Dynamics Models
- Voter and Anti-voter Models
- Value-Weighted Mixture Voter Model
- Learning Problem and Behavior Analysis
- Case of uniform opinion values:
- Learning Method
- Experimental Evaluation
- Experimental Settings
- Comparison Methods
- Experimental Results
- Conclusion
- References
- Using Ontologies in Semantic Data Mining with SEGS and g-SEGS
- Introduction
- Motivation
- Related Work
- Semantic Data Mining with g-SEGS
- Semantic Data Mining
- Hypothesis Language
- Input
- Rule Construction
- Rule Filtering and Evaluation
- g-SEGS Implementation
- An Illustrative Example
- Functional Genomics Use Cases
- Conclusions
- References
- Mining Classification Rules without Support: an Anti-monotone Property of Jaccard Measure
- Introduction
- Related Works
- An Anti-monotone Property of Jaccard Measure
- Framework
- Jaccard Measure
- Jaccard's Anti-monotone Property
- The Algorithm
- Experimental Efficiency
- Study on $Mushroom$ Database
- Study on 10 Databases
- Discussion on Nuggets
- Conclusion
- References
- Bootstrapping Parameter Estimation in Dynamic Systems
- Introduction
- Least Square Parameter Estimation and Bootstrap Methods
- Bootstrapped Parameter Estimator for Dynamic Systems
- Applications: Biochemical Systems
- Experiments and Results
- Conclusion
- References
- Network Effects on Tweeting
- Introduction
- Data
- Tweeting and the Social Network
- The Retweet Network
- URL Retweeting
- The Social and Retweet Networks
- Conclusion and Future Work
- References
- Context-Aware Personal Route Recognition
- Introduction
- Route Recognition Using Instance-Based Learning
- The Setting of the Route Recognition Task
- Baseline Route Recognition Approach
- Using Context Information for Route Recognition
- Performance Evaluation
- Experimental Goals
- Data
- Data Preprocessing and Route Labeling
- Evaluation Criteria
- Experimental Protocol
- Setting the Parameters
- Context in the Dataset
- Effects of Contextual Correction to the Route Recognition Accuracy
- Aggregated Results
- Related Work
- Conclusions and Future Work
- References
- Scalable Detection of Frequent Substrings by Grammar-Based Compression
- Introduction
- Preliminaries
- Pattern Detection Algorithm
- Computational Experiments
- Conclusion
- References
- A Statistical Model for Topically Segmented Documents
- Introduction
- Related Work
- Model Definition
- Perplexity and Cluster Analysis
- Evaluation and Results
- Perplexity Evaluation
- Clustering Evaluation
- Conclusions
- References
- Predicting Structured Outputs $k$-Nearest Neighbours Method
- Predicting Structured Outputs
- Three Tasks of Predicting Structured Outputs
- Methods for Predicting Structured Outputs
- $k$-Nearest Neighbors for Structured Prediction
- $k$-Nearest Neighbors
- Prototype Calculation in $k$-NN for Structured Prediction
- Implementing $k$-NN for Structured Prediction ($k$NN-SP)
- Experimental Evaluation
- Datasets
- Evaluation Metrics
- Estimating and Comparing Predictive Performance
- Results
- Multi-Target Prediction
- Hierarchical Multi-label Classification
- Predicting Short Time Series
- Conclusions and Further Work
- References
- The Augmented Itemset Tree: A Data Structure for Online Maximum Frequent Pattern Mining
- Introduction
- Related Work
- Algorithm
- Problem Statement
- Main Idea of the Used Data Structure
- Definition of the AIST
- InsertPattern
- SetNextNode.
- UpdateMoreGeneralCounts.
- Experiments
- Data Sets
- Empirical Evaluation
- Conclusion
- References
- Word Clouds for Efficient Document Labeling
- Introduction
- Related Work
- Methodology
- Keyword and Key Sentence Extraction
- Keyword Layout
- User Evaluation
- Design
- Procedure
- Test Material
- Participants
- Environment
- Results
- Influence on Labeling Accuracy
- Influence on Labeling Time
- Influence on Classifier Accuracy
- Discussion
- Conclusion and Future Work
- References
- Global and Local Spatial Autocorrelation in Predictive Clustering Trees
- Introduction
- Related Work
- Spatial Autocorrelation
- Building Predictive Clustering Trees
- Learning Spatial PCTS
- The Algorithm
- Estimating the Bandwidth
- Time Complexity
- Empirical Evaluation
- Datasets
- Experimental Setup
- Results and Discussion
- Conclusions
- References
- Rule Stacking: An Approach for Compressing an Ensemble of Rule Sets into a Single Classifier
- Introduction
- Rule Learning
- Rule Stacking
- Stacking
- Motivation for Rule Stacking
- Generating the Meta Data
- Re-transforming the Meta Classifier
- Experimental Setup
- Experiments
- Conclusions
- References
- Graph Clustering Based on Optimization of a Macroscopic Structure of Clusters
- Introduction
- Related Work
- Previous Work
- HITS as a Classification Method
- Our Algorithm
- The Problem
- Macroscopic Structures
- Optimization
- Experiment
- Evaluation
- Environment
- Exp. 1: Exhaustive Study on Variously Noisy Graphs
- Exp. 2: Application to Vector Clustering
- Conclusion
- References
- Modeling the Temperature of Hot Rolled Steel Plate with Semi-supervised Learning Methods
- Introduction
- Temperature Modeling in the Hot Plate Rolling Process
- COREG
- Learning Methods
- Neural Network Model
- MARS
- Stochastic Gradient Boosting Machine
- Results
- Prediction Accuracy
- Analysis of the Features
- Discussion and Conclusion
- References
- Controlled Permutations for Testing Adaptive Classifiers
- Introduction
- The Risk of Order-Dependence Bias in Evaluation
- Proposed Permutations
- Setting
- The Time Permutation
- The Speed Permutation
- The Shape Permutation
- Controlling the Permutations
- Measuring the Extent of Permutations
- The Theoretical Extent of Our Permutations
- Preserving the Distributions
- Experiments
- Related Work
- Conclusion
- References
- Author Index
