Learning and Intelligent Optimization

Name: Learning and Intelligent Optimization | 5th International Conference, LION 5, Rome, Italy, January 17-21, 2011, Selected Papers
Brand: Springer
Price: 53.49 EUR
Availability: OnlineOnly

5th International Conference, LION 5, Rome, Italy, January 17-21, 2011, Selected Papers

Carlos A. Coello-Coello(Editor)

Springer (Publisher)

Published on 15. November 2011

XVII, 636 pages

E-Book

PDF with digital watermarking

System requirements

978-3-642-25566-3 (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

Intro
Title
Preface
Organization
Table of Contents
Main Track (Regular Papers)
Multivariate Statistical Tests for Comparing Classification Algorithms
Introduction
Pairwise Comparison
Univariate Case
Multivariate Case
Analysis of Variance
Univariate Case
Multivariate Case
Experiments
Setup
Results
Conclusions
References
Using Hyperheuristics under a GP Framework for Financial Forecasting
Introduction
Presentation of EDDIE 8
Hyperheuristics Framework
Heuristics and Operators
The Framework
Experimental Setup
Results
Conclusion
References
On the Effect of Connectedness for Biobjective Multiple and Long Path Problems
Introduction
Background
Multiobjective Combinatorial Optimization
Local Search and Connectedness
The Single-Objective Long k-Path Problem
The Biobjective Long k-Path Problem
Definition
Experimental Analysis
The Biobjective Multiple k-Path Problem
Definition
Experimental Analysis
Conclusions and Future Works
References
Improving Parallel Local Search for SAT
Introduction
Background
The Propositional Satisfiability Problem
Local Search for SAT
Refinements
Previous Work
Complete Methods for Parallel SAT
Incomplete Methods for Parallel SAT
Cooperative Algorithms
Knowledge Sharing in Parallel Local Search for SAT
Using Best Known Configurations
Weighting Best Known Configurations
Restart Policy
Experiments
Experimental Settings
Practical Performances with 4 Cores
Practical Performances with 8 Cores
Hardware Impact
Conclusions and Future Work
References
Variable Neighborhood Search for the Time-Dependent Vehicle Routing Problem with Soft Time Windows
Introduction
Problem Description
Solution Method
Initial Solution
Shaking
Local Search
Acceptance Decision
Computational Results
Conclusion
References
Solving the Two-Dimensional Bin Packing Problem with a Probabilistic Multi-start Heuristic
Introduction
Organization of the Paper
Related Work
A New ILP Model
The Proposed Algorithm
Probabilistic LGFi
Multi-start Algorithm
Experimental Evaluation
Problem Instances
Parameter Setting
Computational Results
Conclusions
References
Genetic Diversity and Effective Crossover in Evolutionary Many-objective Optimization
Introduction
Analysis of Pareto Optimal Solutions in Many-objective 0/1 Knapsack Problem
Mating Based on Proximity in Objective Space
Related Works
Local Recombination
Controlling Crossed Genes for Crossover
Problem of Local Recombination in MaOPs
CCG for Two-Point Crossover (CCGTX)
CCG for Uniform Crossover (CCGUX)
Preparation
Algorithms and Selection Methods
Problems, Parameters and Metrics
Experimental Results and Discussion
Diversity of Genes in the Population Obtained by Conventional Crossover
Effects of Local Recombination in MaOPs
Effects of CCGTX in MaOPs
Effects of CCGUX in MaOPs
Conclusions
References
An Optimal Stopping Strategy for Online Calibration in Local Search
Introduction
The Bruss Algorithm
The Estimation of the Probability of Success in Local Exploration
Illustration of the Approach
Experiments
Conclusion and Future Work
References
Analyzing the Effect of Objective Correlation on the Efficient Set of MNK-Landscapes
Introduction
Background
Multiobjective Combinatorial Optimization
Metaheuristics for Multiobjective Combinatorial Optimization
NK- and MNK-Landscapes
MNK-Landscapes: Multiobjective NK-Landscapes with Correlation
Definition
Correlation between Objective Functions
Analysis of the Efficient Set Properties
Cardinality of the Efficient Set
Number of Supported Efficient Solutions
Connectedness of the Efficient Set
Discussion
References
Instance-Based Parameter Tuning via Search Trajectory Similarity Clustering
Introduction
Preliminaries
Automated Parameter Configuration Problem
One-Size-Fits-All Configurator
Instance-Based Configurator
Performance Metric
Solution Approach
Search Trajectory Similarity
Search Trajectory Representation
Similarity Calculation
Clustering Method
Training and Testing Phases
Experimental Design
Experiment Settings
Validity and Statistical Significant Measurement
Experimental Setup
Empirical Evaluation
Performance Comparison
Comparison on Feature Selection
Sensitivity Analysis on Different Initial Sequence Configurations
Computational Results
Discussion
Conclusion and Future Works
References
Effective Probabilistic Stopping Rules for Randomized Metaheuristics: GRASP Implementations
Introduction and Motivation
GRASP and Experimental Environment
Normal Approximation for GRASP Iterations
Probabilistic Stopping Rule
Concluding Remarks
References
A Classifier-Assisted Framework for Expensive Optimization Problems: A Knowledge-Mining Approach
Introduction
Background
Expensive Optimization Problems
Simulator Infeasible Vectors
Proposed Framework
The Model
The Classifier
The Framework
Performance Analysis
Test Problem and Benchmarks
Knowledge-Mining the Classifier
Summary
References
Robust Gaussian Process-Based Global Optimization Using a Fully Bayesian Expected Improvement Criterion
Introduction
Efficient Global Optimization
The Expected Improvement Sampling Criterion for a Gaussian Process
Classical Parametrized Covariance Functions
The EGO Algorithm
The Case of Deceptive Functions
Fully Bayesian One-Step Lookahead Optimization
Student EI
Numerical Experiments
Optimization of a Deceptive Function
Comparison on Sample Paths of a Gaussian Process
References
Hierarchical Hidden Conditional Random Fields for Information Extraction
Introduction
Hierarchical Hidden Conditional Random Fields
Information Extraction
Paper Organization
HHMMs
Representing an HHMM as a DBN
HHCRFs
Model
Parameter Estimation
Sentence Representation
Hierarchical Models for Information Extraction
Upper and Lower Levels
Model Learning
Inference
Experiments
Data
Retrieved Results
Performance Evaluation
Results
Conclusion
References
Solving Extremely Difficult MINLP Problems Using Adaptive Resolution Micro-GA with Tabu Search
Introduction
Related Work
GAs for Solving MINLP Problems
The Proposed Algorithm
Variables Encoding and Genetic Operators
Constraint Handling
Micro GA
Adaptive Resolution Approach
Local Search
Avoiding Redundancy
Results
Environment and Parameters
Results and Discussion
Conclusions and Future Work
References
Adaptive Abnormality Detection on ECG Signal by Utilizing FLAC Features
Introduction
Architecture of the Proposed Framework
Preprocessing in Frequency Domain
Local Auto-correlation on Complex Fourier Values (FLAC) for ECG
Complex Subspace Method
Experiments
Conclusions
References
Gravitational Interactions Optimization
Introduction
Review GSA GIO and CSS
Newton's Law of Universal Gravitation
Gravitational Interactions Optimization
Gravitational Interactions for Unimodal Optimization
Gravitational Interactions for Multimodal Optimization
Experiments
Test Functions
Results
Conclusions
References
On the Neutrality of Flowshop Scheduling Fitness Landscapes
Motivations
Background
Definition of the Permutation Flowshop Scheduling Problem
Neighborhood and Local Search
Fitness Landscape
Neutral Networks Analysis for the Permutation Flowshop Scheduling Problem
Experimental Design
Neutral Degree
Typology of Neutral Networks
Exploiting Neutrality to Solve the FSP
Reaching Portals
How to Guide the Search?
Discussion
References
A Reinforcement Learning Approach for the Flexible Job Shop Scheduling Problem
Introduction
Flexible Job Shop Scheduling Problem
Problem Formulation
Previous Approaches
Dispatching Rules
Reinforcement Learning
Q-Learning
The Proposed Approach: Learning / Optimization
Pseudo-code of the Algorithm
Example
Experimental Results
Instances
Parameters
Comparative Study
Conclusions and Future Work
References
Supervised Learning Linear Priority Dispatch Rules for Job-Shop Scheduling
Introduction
Priority Dispatch Rules for Job-Shop Scheduling
Logistic Regression
Experimental Study
Data Generation
Training Size and Accuracy
Comparison with Single Priority Dispatching Rules
Robustness towards Data Distributions
Fixed Weights
Summary and Conclusion
References
Fine-Tuning Algorithm Parameters Using the Design of Experiments Approach
Introduction
Automated Tuning Framework
Screening Phase
Exploration Phase
Exploitation Phase
Experimental Results
Traveling Salesman Problem (TSP)
Quadratic Assignment Problem (QAP)
Conclusion
References
MetaHybrid: Combining Metamodels and Gradient-Based Techniques in a Hybrid Multi-Objective Genetic Algorithm
GA Elements: Focus on Elitism
SQP Elements: Focus on Constraints
Metamodels Derivatives
Hybridization in a Parallel Environment
Tests
ZDT4
CEC '09 UP2
Rotated OSY
CTP2
Sym-Part
Lennard-Jones
Conclusions
References
Designing Stream Cipher Systems Using Genetic Programming
Introduction
Stream Cipher Systems
Genetic Programming and Simulated Annealing
Simple Genetic Programming Method
Function Library
Representation Scheme
Fitness Function
Algorithm Parameters
The Design Algorithm
Simulated Annealing Programming Method
Adaptive Genetic Programming Method
Results
Conclusion
References
GPU-Based Multi-start Local Search Algorithms
Introduction
Parallel Local Search Algorithms and GPU Computing
Parallel Models of LS Algorithms
GPU Computing
Design and Implementation of Multi-start Local Search Algorithms on GPU
Multi-start Local Search Algorithms Based on the Iteration-Level
Design of Multi-start Local Search Algorithms Based on the Algorithmic-Level
Memory Management of Multi-start Local Search Algorithms on the Algorithmic-Level
Experiments
Measures of the Efficiency of Multi-start Algorithms Based on the Algorithmic-Level
Measures of the Efficiency of Large GPU-Based Implementations
Discussion and Conclusion
References
Active Learning of Combinatorial Features for Interactive Optimization
Introduction
Overview of Our Approach
Satisfiability Modulo Theory
Satisfiability Modulo Theory Solvers
Weighted MAX-SMT
Related Works
Experimental Results
Weighted MAX-SAT
Weighted MAX-SMT
Discussion
References
A Genetic Algorithm Hybridized with the Discrete Lagrangian Method for Trap Escaping
Introduction
The Hybrid Method
The Hybrid Method Applied to WDP
Winner Determination
The Discrete Lagrangian Method for WDP
The Scheme of the Hybrid Algorithm for WDP
Experiments
Experimental Settings
Results
Conclusion
References
Greedy Local Improvement of SPEA2 Algorithm to Solve the Multiobjective Capacitated Transshipment Problem
Introduction
Model
Problem Description
Modeling Assumptions
Model Formulation
Objective Functions Estimation
Evolutionary Multiobjective Optimization
SPEA2: Brief Description
SPEA2 with a Greedy Local Search
Optimization Results
Cost vs. Fill Rate Problem
Cost vs. Lead Time
Conclusions
References
Hybrid Population-Based Incremental Learning Using Real Codes
Introduction
Hybrid Algorithm
Population-Based Incremental Learning
Evolutionary Direction Recombination
Approximate Gradient
Hybrid Algorithm
Testing Functions
Comparison Results
Conclusions and Discussion
References
Pareto Autonomous Local Search
Introduction
Neighborhood, Selectors and Operators
General Definitions
Permutations
Operator Control for Local Search
Metrics
Operator Selection
Experiments
Experimental Protocol
Results and Discussion
Conclusion
References
Transforming Mathematical Models Using Declarative Reformulation Rules
Introduction
Structures
Compositions
Creating a Model
Reformulations
ProdBC to MILP
SAbs to Composition
VAbs to LP
SemiContinuous to MILP
ProdCC to MILP
SemiAssign to MILP
Constraint to MILP
OFMin to MILP
IndComposition to MILP
Composition to MILP
Applying the ARRs to HCP
Discussion
References
Learning Heuristic Policies - A Reinforcement Learning Problem
Introduction
Learning Heuristics - A Reinforcement Learning Problem
Bin Packing
Illustrative Example Using Bin-Packing
On-Line Bin Packing
Off-Line Bin Packing
Summary and Conclusions
References
Continuous Upper Confidence Trees
Introduction
Progressive Widening for Upper Confidence Trees
Progressive Widening
Why It Does Not Work as Is for Randomized Transitions in Continuous Domains
Proposed Solution: Double Progressive Widening
Experiments
Trap Problem
The Power Management Problem
Conclusion
References
Main Track (Short Papers)
Towards an Intelligent Non-stationary Performance Prediction of Engineering Systems
Introduction
Partial Non-stationary Kriging
Engine Casing Temperature Response Prediction
Conclusions
References
Local Search for Constrained Financial Portfolio Selection Problems with Short Sellings
Introduction
The Portfolio Selection Problem with Short Sellings
Local Search
Experiments
References
Clustering of Local Optima in Combinatorial Fitness Landscapes
Introduction
Methodology
Results and Discussion
References
Special Session: IMON
Multi-Objective Optimization with an Adaptive Resonance Theory-Based Estimation of Distribution Algorithm: A Comparative Study
Introduction
The Model-Building Issue
Model Building with Adaptive Resonance Theory
Gaussian ART for Model-Building
Multi-Objective ART-Based EDA
Experimental Study
Final Remarks
References
Multi-Objective Differential Evolution with Adaptive Control of Parameters and Operators
Introduction
Related Work
Parameter Setting in Evolutionary Algorithms
Adaptive Operator Selection
Adaptive Multi-Objective DE
Fitness Evaluation
Replacement Mechanism
Adaptive Operator Selection
Adaptive Parameter Control of CR and F
Performance Comparison
Experimental Settings
Experimental Results
Conclusion
References
Distribution of Computational Effort in Parallel MOEA/D
Introduction
Sequential MOEA/D
A Parallel Model of MOEA/D
Extending the Proposed Parallel Model
pMOEA/Dv1
pMOEA/Dv2
Experimentation
Configuration
Benchmark
Analysis of the Results
Conclusions and Future Work
References
Multi Objective Genetic Programming for Feature Construction in Classification Problems
Introduction
Methods
Conclusions
References
Special Session: LION-PP
Sequential Model-Based Optimization for General Algorithm Configuration
Introduction
Existing Work on Sequential Model-Based Optimization (SMBO)
Random Online Aggressive Racing (ROAR)
Generalization I: An Intensification Mechanism for Multiple Instances
Defining ROAR
Sequential Model-Based Algorithm Configuration (SMAC)
Generalization II: Models for Categorical Parameters
Generalization III: Models for Sets of Problem Instances
Generalization IV: Using the Model to Select Promising Configurations in Large Mixed Numerical/Categorical Configuration Spaces
Experimental Evaluation
Experimental Setup
Experimental Results for Single Instance Scenarios
Experimental Results for General Multi-instance Configuration Scenarios
Conclusion
References
Generalising Algorithm Performance in Instance Space: A Timetabling Case Study
Introduction
Course Timetabling
Visualising Instance Space
Self-Organising Feature Maps
Visualising the Instance Space
Visualising the Footprints of Algorithm Performance
Partitioning the Instance Space via Decision Trees
Conclusions
References
Special Session: Self* EAs
A Hybrid Fish Swarm Optimisation Algorithm for Solving Examination Timetabling Problems
Introduction
Uncapacitated Examination Timetabling Problem
The Fish Swarm Optimisation Algorithm
The Hybrid Approach
Nelder-Mead Simplex Algorithm
Multi Decay Rate Great Deluge Algorithm
Simulation Results
Conclusion and Future Work
References
The Sandpile Mutation Operator for Genetic Algorithms
Introduction
SOC in Evolutionary Computation
The Sandpile Model and the Sandpile Mutation
Test Set and Results
Functions
Methodology
Results
Mutation Rate Analysis
Conclusions and Future Work
References
Self-adaptation Techniques Applied to Multi-Objective Evolutionary Algorithms
Introduction
Previous Related Work
Our Proposed Approach
Phase 1: Sensitivity Analysis
Self-adaptation of Parameters
The Individual
Crossover Operator
Mutation Operator
Inheritance-Fertilization Operator
Stopping Criterion and a Varying Population Size
Experimental Results
Experimental Setup
Discussion of Results
Conclusions and Future Research
References
Analysing the Performance of Different Population Structures for an Agent-Based Evolutionary Algorithm
Introduction
Experiments and Results
Conclusions
References
Special Session: LION-SWAP
EDACC - An Advanced Platform for the Experiment Design, Administration and Analysis of Empirical Algorithms
Introduction
EDACC - Overview of the Main Components
Information Extraction
Instance Properties
Result Properties
Analysis and Statistical Evaluation
EDACC - Competition Mode
Implementation Details
Related Work
Conclusion and Future Work
References
HAL: A Framework for the Automated Analysis and Design of High-Performance Algorithms
Introduction
HAL: A Framework for Meta-algorithmics
Meta-algorithmic Problems
The High-Performance Algorithm Laboratory
The HAL 1.0 Core Infrastructure
Experiment Modelling
Execution and Data Management
User Interface
Case Study: Analysis and Design with HAL 1.0
The Single-Algorithm Analysis Problem
The Pairwise Comparison Problem
The Algorithm Configuration Problem
Conclusions and Future Work
References
Hyperion - A Recursive Hyper-Heuristic Framework
Introduction
Domain Analysis
TheHyperion Hyper-Heuristic Framework
Design-Space of Hyper-Heuristics
Application to SAT
Conclusion and Future Work
References
The Cross-Domain Heuristic Search Challenge - An International Research Competition
Introduction
The HyFlex Framework
Challenge Description and Scoring System
Final Remarks
References
Author Index

Content (PDF)

System requirements

Save as PDF Copy link into clipboard

Schweitzer Fachinformationen

Learning and Intelligent Optimization

Description

More details

Other editions

Additional editions

Content

System requirements