Big Data Analytics: Cluster Analysis And Pattern Recognition. Examples With Matlab

BIG DATA AND ANALYTICS. NEURAL NETWORKS TOOLS
1.1 INTRODUCTION TO BIG DATA AND ANALYTICS
1.2 MATLAB AND BIG DATA
1.2.1 Access Data
1.2.2 Explore, Process, and Analyze Data
1.2.3 Develop Predictive Models
1.3 NEURAL NETWORKS WITH MATLAB
Cluster ANALYSIS with NEURAL NETWORKS
2.1 INTRODUCTION
2.2 Using the Neural Network Clustering Tool
2.3 Using Command-Line Functions
Cluster ANALYSIS with NEURAL NETWORKS. Self-Organizing Map
3.7.1 One-Dimensional Self-Organizing Map
3.7.2 Two-Dimensional Self-Organizing Map
3.7.3 Training with the Batch Algorithm
CLUSTER ANALYSIS WITH NEURAL NETWORKS. Self-Organizing Maps FUNCTIONS
4.1 FUNCTIONS
4.2 nnstart
4.3 view
4.4 selforgmap
4.5 train
4.6 plotsomhits
4.7 plotsomnc
4.8 plotsomnd
4.9 plotsomplanes
4.10 plotsompos
4.11 plotsomtop
4.12 genFunction
COMPETITIVE NEURAL NETWORKS
5.1 Create a Competitive Neural Network
5.1.1 Kohonen Learning Rule (learnk)
5.1.2 Bias Learning Rule (learncon)
5.1.3 Training
5.1.4 Graphical Example
5.2 Competitive Layers
5.2.1 Functions
5.3 nnstart
5.4 view
5.5 selforgmap
5.6 train
5.7 plotsomhits
5.8 plotsomnc
5.9 plotsomnd
5.10 plotsomplanes
5.11 plotsompos
5.12 plotsomtop
5.13 genFunction
Competitive Layer FUNCTIONS
6.1 FUNCTIONS
6.2 COMPETLAYER
6.3 view
6.4 trainru
6.5 learnk
6.6 learncon
BIG DATA CLASSIFICATION. Classify Patterns with a Neural Network
7.1 INTRODUCTION
7.2 Using the Neural Network Pattern Recognition Tool
7.3 Using Command-Line Functions
BIG DATA CLASSIFICATION. WORK FLOW FOR NEURAL NETWORK DESIGN
8.1 INTRODUCTION
8.2 Four Levels of Neural Network Design
8.3 Multilayer Neural Networks and Backpropagation Training
8.4 Multilayer Neural Network Architecture
8.4.1 Neuron Model (logsig, tansig, purelin)
8.4.2 Feedforward Neural Network
8.5 Understanding Neural Network Toolbox Data Structures
8.5.1 Simulation with Concurrent Inputs in a Static Network
8.5.2 Simulation with Sequential Inputs in a Dynamic Network
8.5.3 Simulation with Concurrent Inputs in a Dynamic Network
8.6 Neural Network Object Properties
8.6.1 General
8.6.2 Architecture
8.6.3 Subobject Structures
8.6.4 Functions
8.6.5 Weight and Bias Values
8.7 Neural Network Subobject Properties
8.7.1 Inputs
8.7.2 Layers
8.7.3 Outputs
8.7.4 Biases
8.7.5 Input Weights
8.7.6 Layer Weights
BIG DATA TOOLS. FunctionS FOR PATTERN RECOGNITION AND CLASSIFICATION
9.1 INTRODUCTION
9.2 view NEURAL NETWORK
9.3 Pattern Recognition and Learning Vector Quantization
9.3.1 Pattern recognition network: patternnet
9.3.2 Learning vector quantization neural network: lvqnet
9.4 Training Options and Network Performance
9.4.1 Receiver operating characteristic: roc
9.4.2 Plot receiver operating characteristic: plotroc
9.4.3 Plot classification confusion matrix: plotconfusion
9.4.4 Neural network performance: crossentropy
9.4.5 Construct and Train a Function Fitting Network
9.4.6 Create and train Feedforward Neural Network
9.4.7 Create and Train a Cascade Network
9.5 Network performance
9.5.1 Description
9.5.2 Examples
9.6 Fit Regression Model and Plot Fitted Values versus Targets
9.6.1 Description
9.6.2 Examples
9.7 Plot Output and Target Values
9.7.1 Description
9.7.2 Examples
9.8 Plot Training State Values
9.9 Plot Performances
9.10 Plot Histogram of Error Values
9.10.1 Syntax
9.10.2 Description
9.10.3 Examples
9.11 Generate MATLAB function for simulating neural network
9.11.1 Create Functions from Static Neural Network
9.11.2 Create Functions from Dynamic Neural Network
9.12 A COMPLETE EXAMPLE: House Price Estimation
9.12.1 The Problem: Estimate House Values
9.12.2 Why Neural Networks?
9.12.3 Preparing the Data
9.12.4 Fitting a Function with a Neural Network
9.12.5 Testing the Neural Network
9.13 Autoencoder class
9.13.1 trainAutoencoder
9.13.2 Construct Deep Network Using Autoencoders
9.13.3 decode
9.13.4 encode
9.13.5 predict
9.13.6 stack
BIG DATA TOOLS. MULTILAYER Neural Network
10.1 Create, Configure, and Initialize Multilayer Neural Networks
10.1.1 Other Related Architectures
10.2 FUNCTIONS FOR Create, Configure, and Initialize Multilayer Neural Networks
10.2.1 Initializing Weights (init)
10.2.2 feedforwardnet
Syntax
feedforwardnet(hiddenSizes,trainFcn)
Description
Feedforward networks consist of a series of layers. The first layer has a connection from the network input. Each subsequent layer has a connection from the previous layer. The final layer produces the network’s output.
Examples
Feedforward Neural Network
10.2.3 configure
10.2.4 init
To Get Help
Algorithms
10.2.5 train
10.2.6 trainlm
10.2.7 tansig
10.2.8 purelin
10.2.9 cascadeforwardnet
10.2.10 patternnet
10.3 Train and Apply Multilayer Neural Networks
10.3.1 Training Algorithms
10.3.2 Training Example
10.3.3 Use the Network
10.4 Train ALGORITMS IN Multilayer Neural Networks
10.4.1 trainbr:Bayesian Regularization
10.4.2 trainscg: Scaled conjugate gradient backpropagation
10.4.3 trainrp: Resilient backpropagation
10.4.4 trainbfg: BFGS quasi-Newton backpropagation
10.4.5 traincgb: Conjugate gradient backpropagation with Powell-Beale restarts
10.4.6 traincgf: Conjugate gradient backpropagation with Fletcher-Reeves updates
10.4.7 traincgp: Conjugate gradient backpropagation with Polak-Ribiére updates
10.4.8 trainoss: One-step secant backpropagation
10.4.9 traingdx: Gradient descent with momentum and adaptive learning rate backpropagation
10.4.10 traingdm: Gradient descent with momentum backpropagation
10.4.11 traingd: Gradient descent backpropagation
ANALYZE AND DEPLOY TRAINED NEURAL NETWORK
11.1 ANALYZE NEURAL NETWORK PERFORMANCE
11.2 Improving Results
11.3 Deployment Functions and Tools for Trained Networks
11.4 Generate Neural Network Functions for Application Deployment
11.5 Deploy Neural Network Simulink Diagrams
11.5.1 Example
11.5.2 Suggested Exercises
11.6 Deploy Training of Neural Networks
BIG DATA PARALLEL COMPUTING. TRAINING SCALABILITY AND EFICIENCE
12.1 Neural Networks with Parallel and GPU Computing
12.1.1 Modes of Parallelism
12.1.2 Distributed Computing
12.1.3 Single GPU Computing
12.1.4 Distributed GPU Computing
12.1.5 Deep Learning
12.1.6 Parallel Time Series
12.1.7 Parallel Availability, Fallbacks, and Feedback
12.2 Automatically Save Checkpoints During Neural Network Training
12.3 Optimize Neural Network Training Speed and Memory
12.3.1 Memory Reduction
12.3.2 Fast Elliot Sigmoid
OPTIMAL SOLUTIONS
13.1 Representing Unknown or Don’t-Care Targets
13.1.1 Choose Neural Network Input-Output Processing Functions
13.1.2 Representing Unknown or Don’t-Care Targets
13.2 Configure Neural Network Inputs and Outputs
13.3 Divide Data for Optimal Neural Network Training
13.4 Choose a Multilayer Neural Network Training Function
13.4.1 SIN Data Set
13.4.2 PARITY Data Set
13.4.3 ENGINE Data Set
13.4.4 CANCER Data Set
13.4.5 CHOLESTEROL Data Set
13.4.6 DIABETES Data Set
13.4.7 Summary
13.5 Improve Neural Network Generalization and Avoid Overfitting
13.5.1 Retraining Neural Networks
13.5.2 Multiple Neural Networks
13.5.3 Early Stopping
13.5.4 Index Data Division (divideind)
13.5.5 Random Data Division (dividerand)
13.5.6 Block Data Division (divideblock)
13.5.7 Interleaved Data Division (divideint)
13.5.8 Regularization
13.5.9 Modified Performance Function
13.5.10 Automated Regularization (trainbr)
13.5.11 Summary and Discussion of Early Stopping and Regularization
13.5.12 Posttraining Analysis (regression)
13.6 Train Neural Networks with Error Weights
13.7 Normalize Errors of Multiple Outputs
CLASSIFICATION WITH NEURAL NETWORKS. EXAMPLES
14.1 Crab Classification
14.1.1 Why Neural Networks?
14.1.2 Preparing the Data
14.1.3 Building the Neural Network Classifier
14.1.4 Testing the Classifier
14.2 Wine Classification
14.2.1 The Problem: Classify Wines
14.2.2 Why Neural Networks?
14.2.3 Preparing the Data
14.2.4 Pattern Recognition with a Neural Network
14.2.5 Testing the Neural Network
14.3 Cancer Detection
14.3.1 Formatting the Data
14.3.2 Ranking Key Features
14.3.3 Classification Using a Feed Forward Neural Network
14.4 Character Recognition
14.4.1 Creating the First Neural Network
14.4.2 Training the first Neural Network
14.4.3 Training the Second Neural Network
14.4.4 Testing Both Neural Networks
AUTOENCODERS AND CLUSTERING WITH NEURAL NETWORKS. EXAMPLES
15.1 Train Stacked Autoencoders for Image Classification
15.1.1 Data set
15.1.2 Training the first autoencoder
15.1.3 Visualizing the weights of the first autoencoder
15.1.4 Training the second autoencoder
15.1.5 Training the final softmax layer
15.1.6 Forming a stacked neural network
15.1.7 Fine tuning the deep neural network
15.1.8 Summary
15.2 Transfer Learning Using Convolutional Neural Networks
15.3 Iris Clustering
15.3.1 Why Self-Organizing Map Neural Networks?
15.3.2 Preparing the Data
15.3.3 Clustering with a Neural Network
15.4 Gene Expression Analysis
15.4.1 The Problem: Analyzing Gene Expressions in Baker’s Yeast (Saccharomyces Cerevisiae)
15.4.2 The Data
15.4.3 Filtering the Genes
15.4.4 Principal Component Analysis
15.4.5 Cluster Analysis: Self-Organizing Maps
Self-organizing Networks. EXAMPLES
16.1 Competitive Learning
16.2 One-Dimensional Self-organizing Map
16.3 Two-Dimensional Self-organizing Map