Computationefficiency.doc

Boris Mirkin School of Computer Science and Information Systems, Birkbeck College, University of London UK Computational Intelligence: Association, Aggregation and Visualization Contents: 0. Introduction: . . . . . . . . 3 01. Computational intelligence. 02. Data visualization. 03. Case study problems. 1. Aggregating and visualizing a single feature: 1D analysis 18 1.1. Distribution 1.2. Centre and spread: definitions, properties, integral perspectives Project 1.1 Analysis of a multimodal distribution 1.3. Confidence and computational experiment Project 1.2 Data mining with a confidence interval: Bootstrap Project 1.3 K-fold cross-validation 1.4. Modelling uncertainty: Interval and fuzzy values. 2 Associating and visualizing two features: 2D analysis 35 2.1. Both quantitative: Linear regression and residual variance Correlation coefficient Project 2.1. 2D analysis, linear regression and bootstrapping Estimating non-linear regression Project 2.2 Non-linear regression versus its linearized version: evolutionary algorithm for estimation 2.2. Nominal and quantitative: table regression and correlation ratio 2.3. Both categorical: Contingency table Quetélet indices Chi-squared and its visualization Aggregation of categories 3. Association with decision rules 53 3.1. Linear regression 3.2 . Discriminant function and SVM 3.3. Decision trees 4. Learning neural networks for prediction 4.1. Artificial neuron and perceptron 4.2. Multilayer network 4.3. Back-propagation algorithm 5. Aggregation with Principal Component Analysis 5.1. One component 5.2. Principal Components and SVD 5.3. Popular applications 5.4. Correspondence analysis 5.5. Self-associative neural nets. 6. K-Means clustering 6.1. Batch and incremental K-Means 6.2. Anomalous pattern and iK-Means 6.3. Experimentally determining K 6.4. Evolutionary approaches to K-Means Genetic algorithms Evolutionary algorithms Particle swarm optimization 6.5. Extensions of K-Means Fuzzy Kohon