The book introduces the reader to Deep Learning, an advanced machine learning method to analyze data and find patterns by means of self-adapting, self-improving neural networks. After an overview of the fundamentals, the book explains the different network architectures used in Deep Learning and the different learning methods such as energy-driven, reductionist and success target learning. The last part deals with the advanced concepts of Deep Learning such as weak and unsupervised training and hybrid network architectures.
Table of Contents
PART I: BASIC UNDERSTANDINGRelevance of Machine Learning
Basic Idea of Deep Learning
Neural Networks as Multivariate, Multidimensional Models
Optimization of Network Parameters -
Quality of Modelling
PART II: NETWORK ARCHITECTURES
Basic Architecture and Extensions
Analysis of Image Data
Analysis of Point Clouds
Time Series and Variable Input Data
Learning with Success Targets
Energy-Driven Learning Methods
Reduction to Essential Information
Cooperation of Several Networks
PART III: NETWORK INSIGHTS AND ADVANCED CONCEPTS
Understanding of Trained Networks
Systematic Uncertainties
Weak and Unsupervised Training
Hybrid Architectures
