Neural Networks: History and Applications

✍ Scribed by Doug Alexander

• Publisher: Nova Science Publishers, Incorporated
• Year: 2020
• Tongue: English
• Leaves: 234
• Category: Library

✦ Synopsis

"With respect to the ever-increasing developments in artificial intelligence and artificial neural network applications in different scopes such as medicine, industry, biology, history, military industries, recognition science, space, machine learning and etc., Neural Networks: History and Applications first discusses a comprehensive investigation of artificial neural networks. Next, the authors focus on studies carried out with the artificial neural network approach on the emotion recognition from 2D facial expressions between 2009 and 2019. The major objective of this study is to review, identify, evaluate and analyze the performance of artificial neural network models in emotion recognition applications. This compilation also proposes a simple nonlinear approach for dipole mode index prediction where past values of dipole mode index were used as inputs, and future values were predicted by artificial neural networks. The study was also conducted for seasonal dipole mode index prediction because the dipole mode index is more prominent in the Sep-Oct-Nov season. A subsequent study focuses on how mammography has a high false negative and false positive rate. As such, computer-aided diagnosis systems have been commercialized to help in micro-calcification detection and malignancy differentiation. Yet, little has been explored in differentiating breast cancers with artificial neural networks, one example of computer-aided diagnosis systems. The authors aim to bridge this gap in research. The penultimate chapter reviews the general conditions under which synaptic plasticity most effectively takes place to support the supervised learning of a precise temporal code. Then, the accuracy of each plasticity rule with respect to its temporal encoding precision is examined, and the maximum number of input patterns it can memorize using the precise timings of individual spikes as an indicator of storage capacity in different control and recognition tasks is explored. In closing, a case study is presented centered on an intelligent decision support system that is built on a neural network model based on the Encog machine learning framework to predict cryptocurrency close prices"--

✦ Table of Contents

Contents
Preface
Chapter 1
Artificial Neural Networks, Concept, Application and Types
Abstract
Introduction
Biological Neural Networks
What Is a Neural Network?
Artificial Neurons
What Is an ANN?
Neural Network Structure
Training Methods
Application of ANNs
ANN Types
Perceptron Neural Network
Single Layer Perceptron
Learning a Perceptron
Multilayer Perceptron Neural Networks
Radial Basis Function Neural Networks
Hopfield Neural Network
Hamming Neural Network
Kohonen Self-Organized Map Neural Network
Time Delay Neural Network
Deep Feed Forward Neural Networks
Recurrent Neural Networks
Long-Short Term Memory Neural Networks
Auto Encoders Neural Networks
Markov Chains Networks
Conclusion, Advantages and Disadvantages of AANs
References
Chapter 2
Emotion Recognition from Facial Expressions Using Artificial Neural Networks: A Review
Abstract
Introduction
Literature Review
Results
Conclusion
References
Chapter 3
Dipole Mode Index Prediction with Artificial Neural Networks
Abstract
1. Introduction
2. Data
3. Methodology
4. Results
4.1. Monthly DMI Prediction
4.2. Seasonal DMI Prediction
4.3. Important IOD Events
5. Discussion
Acknowledgment
References
Biographical Sketches
Chapter 4
Efficacy of Artificial Neural Networks in Differentiating Breast Cancers in Digital Mammography
Abstract
1. Introduction
1.1. ANN Structure
2. Materials and Methods
2.1. Feature Extraction
2.2. ANN Training and Testing
2.3. Ethical Consideration
2.4. Data Analysis
3. Results
4. Discussion
Conclusion
Appendix A
References
Chapter 5
Supervised Adjustment of Synaptic Plasticity in Spiking Neural Networks
Abstract
1. Supervised Learning in SNN via Reward-Modulated Spike-Timing-Dependent Plasticity
1.1. Modeling of Spiking Neural Network and the Supervised R-STDP Learning Rule
1.2. Reference Dataset
1.2.1. Obstacle Avoiding Dataset
1.2.2. Goal-Approaching Dataset
1.2.3. Calculating the Speed of the Robot Motor
1.3. Controller
1.4. Testing Environments and the Performance of the Controller to Achieve Target-Reaching Tasks in Different Scenarios
1.4.1. Testing Environment
1.4.2. Goal-Approaching Sub-Controller
1.4.3. Obstacle-Avoiding Sub-Controller
1.4.4. Overall Performance
1.5. Perspectives and Limitations of SNN-Based Supervised Controllers for Performing Target-Reaching Tasks
2. Applying Symmetric STDP Rules to Define Biologically Plausible Supervised Learning Method for SNNs
2.1. Network Architecture and Neuronal Dynamics
2.2. Performance of the Recognition Task
2.3. Clustering Ability of the Model
2.4. Comparison to Other SNN Models
2.5. Comparison to STDP-Based Models with or without Backpropagation
2.6. SNN-Based Supervised Learning and Handwritten Digit Recognition Tasks
3. Learning Structure of Sensory Inputs with Synaptic Plasticity Leads to Synaptic Interference
3.1. Simulation Procedure, Network Structure and Plasticity Model
3.2. Data Preparation
3.2.1. Synthetic Signal Data
3.2.2. Speaker Recognition Data
3.2.3. Pre-Processing of the Human Motion Data
3.3. Training Recurrent Networks with Plasticity
3.4. Analysis of Synaptic Adaptation
3.5. Learning Input-Specific Adaptations
3.6. Classification Performance with Plasticity and Synaptic Interference
3.7. Evolution of Synaptic Weights
3.8. Synaptic Interference and Its Impact on Learning Performance
References
Chapter 6
A Review on Intelligent Decision Support Systems and a Case Study: Prediction of Cryptocurrency Prices with Neural Networks
Abstract
Introduction
Related Works
Medical Diagnosis IDSS
Business/Financial IDSS
Environment/Energy IDSS
Computational Intelligence and Data Mining Methods for IDDS
A Case Study to Predict Cryptocurrency Prices with Neural Networks
Model Building and Training
Experimental Results
Conclusion
References
Index
Blank Page