Introduction
Motivation: Data, Models and Molecular Sciences
Optimization
Calculus
Iterative Optimization
Iterative Local Optimization
Iterative Global Optimization
Constrained Iterative Optimization
Model Functions
Linear Model Functions with One Argument
Non-linear Model Functions with One Argument
Linear Model Functions with Multiple Arguments
Non-linear Model Functions with Multiple Arguments
Multiple Model Functions
Summary
Data Structures
Data for Curve Fitting
Data for Machine Learning
Inputs for Clustering
Inspection of Data Sets and Inputs
Scaling of Data
Data Errors
Regression versus Classification Tasks
The Structure of CIP Calculations
Curve Fitting
Basics
Fitting Data
Useful Quantities
Smoothing Data
Evaluating the Goodness of Fit
How to Guess a Model Function
Problems and Pitfalls
Parameters’ Start Values
How to Search for Parameters’ Start Values
More Difficult Curve Fitting Problems
Inappropriate Model Functions
Parameters’ Errors
Correction of Parameters’ Errors
Confidence Levels of Parameters’ Errors
Estimating the Necessary Number of Data
Large Parameters’ Errors and Educated Cheating
Experimental Errors and Data Transformation
Empirical Enhancement of Theoretical Model Functions
Data Smoothing with Cubic Splines
Cookbook Recipes for Curve Fitting
Clustering
Basics
Intuitive Clustering
Clustering with a Fixed Number of Clusters
Getting Representatives
Cluster Occupancies and the Iris Flower Example
White-Spot Analysis
Alternative Clustering with ART-2a
Clustering and Class Predictions
Cookbook Recipes for Clustering
Machine Learning
Basics
Machine Learning Methods
Multiple Linear Regression (MLR)
Three-Layer Perceptron-Type Neural Networks
Support Vector Machines (SVM)
Evaluating the Goodness of Regression
Evaluating the Goodness of Classification
Regression: Entering Non-linearity
Classification: Non-linear Decision Surfaces
Ambiguous Classification
Training and Test Set Partitioning
Cluster Representatives Based Selection
Iris Flower Classification Revisited
Adhesive Kinetics Regression Revisited
Design of Experiment
Concluding Remarks
Comparative Machine Learning
Relevance of Input Components
Pattern Recognition
Cookbook Recipes for Machine Learning
Appendix - Collecting the Pieces
Discussion
Computers Are about Speed
Isn’t It Just ...?
... Optimization?
... Data Smoothing?
Computational Intelligence
Final Remark
Cover
Front Matter
Introduction
Motivation: Data, Models and Molecular Sciences
Optimization
Calculus
Iterative Optimization
Iterative Local Optimization
Iterative Global Optimization
Constrained Iterative Optimization
Model Functions
Linear Model Functions with One Argument
Non-linear Model Functions with One Argument
Linear Model Functions with Multiple Arguments
Non-linear Model Functions with Multiple Arguments
Summary
Multiple Model Functions
Data Structures
Data for Machine Learning
Data for Curve Fitting
Inspection of Data Sets and Inputs
Inputs for Clustering
Scaling of Data
Data Errors
Regression versus Classification Tasks
The Structure of CIP Calculations
Curve Fitting
Basics
Fitting Data
Useful Quantities
Smoothing Data
Evaluating the Goodness of Fit
How to Guess a Model Function
Problems and Pitfalls
Parameters’ Start Values
How to Search for Parameters’ Start Values
More Difficult Curve Fitting Problems
Inappropriate Model Functions
Parameters’ Errors
Correction of Parameters’ Errors
Confidence Levels of Parameters’ Errors
Estimating the Necessary Number of Data
Large Parameters’ Errors and Educated Cheating
Experimental Errors and Data Transformation
Empirical Enhancement of Theoretical Model Functions
Data Smoothing with Cubic Splines
Cookbook Recipes for Curve Fitting
Clustering
Basics
Intuitive Clustering
Clustering with a Fixed Number of Clusters
Getting Representatives
Cluster Occupancies and the Iris Flower Example
White-Spot Analysis
Alternative Clustering with ART-2a
Clustering and Class Predictions
Cookbook Recipes for Clustering
Machine Learning
Basics
Machine Learning Methods
Multiple Linear Regression (MLR)
Three-Layer Perceptron-Type Neural Networks
Support Vector Machines (SVM)
Evaluating the Goodness of Regression
Evaluating the Goodness of Classification
Regression: Entering Non-linearity
Classification: Non-linear Decision Surfaces
Ambiguous Classification
Training and Test Set Partitioning
Cluster Representatives Based Selection
Iris Flower Classification Revisited
Adhesive Kinetics Regression Revisited
Design of Experiment
Comparative Machine Learning
Concluding Remarks
Relevance of Input Components
Pattern Recognition
Cookbook Recipes for Machine Learning
Appendix - Collecting the Pieces
Discussion
Computers Are about Speed
Isn’t It Just ...?
... Data Smoothing?
... Optimization?
Computational Intelligence
Final Remark
Back Matter