No label defined (Q3843142)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 1194484
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 1194484 |
Statements
25 August 1998
0 references
Neural network analysis
0 references
Neural network architectures
0 references
Neural network applications
0 references
neural networks
0 references
neural network architectures
0 references
neural network analysis
0 references
neural network applications
0 references
The book is an edited collection of high quality chapters on various spots in neural computing. Part one (Chapters 1-3) is on understanding and simplifying neural networks. Chapter 1 looks at six mathematical methods for analyzing a biologically plausible model of a neural network (NN). The aim is to understand how a massive hidden layer operates, thereby getting insight into brain functioning. Chapter 2 sets up a different task: to simplify the NN. The author proposes an information maximization method as a more rigorous alternative to the heuristic pruning schemes developed so far. Chapter 3 gives an extensive account on rule extraction methods from a trained NN. These are discussed in the light of the long-term developments in AI (artificial intelligence) about explanation capability and quality of rule based systems. Part two (Chapters 4-7) considers nonstandard NN architectures and efficient learning methods. Chapter 4 guides the reader through pulse-based designs. Using a few basic hardware schemes the authors show how their model of a neuron operates on pulse-encoded signals. Cellular neural networks (CNNs) are introduced in Chapter 5 and their use for parallel image processing is explained. The author presents a sound (personal) view on CNNs as locally connected neural network dynamical systems. A number of gradient descent and gradient conjugate algorithms for efficient NN training are summarized in Chapter 6, ending with a discussion on on-line versus off-line training. Chapter 7 advocates training of a committee of NNs by finding an individual training set for each member through a Kohonen feature map. Part three (Chapters 8-10) contains 3 NN applications. Chapter 8 introduces traditional and fuzzy control, and describes how NNs are used in the design and operation of control systems. Chapter 9 contains methods for image compression using classical feed-forward NN models trained with backpropagation, random NNs, variants of vector quantization, etc. A case study on oil spill detection by a NN (Chapter 10) completes the book.
0 references