Featuring a concise five-part presentation of the concepts essential to neural network construction and application, this video focuses on software-oriented applications rather than hardware-based construction. It helps students and professionals effectively:
* learn state-of-the-art neural networks technology,
* understand the computing elements that make up simple associative networks,
* explore what functions neural networks can do best, and
* determine the future capabilities and limitations of neural network technology for their own fields.
Acting as a valuable companion to the video presentation, the manual includes:
* detailed text to back up material and concepts presented in the video,
* numerous figures, drawings, and extensive demonstrations,
* an extensive list of further resources and reading recommended by the author, and
* key words in neural networks vocabulary in boldface throughout the text.
Sprache
Zielgruppe
ISBN-13
978-1-56321-001-3 (9781563210013)
Copyright in bibliographic data is held by Nielsen Book Services Limited or its licensors: all rights reserved.
Schweitzer Klassifikation
Contents: Part I:Introduction To Neural Networks. Brain Like Computers. Hardware and Software for Them. Computing Elements. Massive Parallelism. Memory Based Computation. Rules in Language and Their Peculiarities. Practical Applications Are Cognitive Applications. Part II:Introduction To The Models: Linear Associator. Representation Assumption. State Vectors. The Model Neuron: The Synaptic Inner Product. The Output Non-Linearity. System Dynamics. The Hebb Rule and Outer Product Matrix. Simple Association. Part III:Concepts. Capacity in Neural Networks. Coping With Complexity. Psychological Concepts. Prototypes. Psychological Dot Pattern Experiments. Representation of a Dot Pattern. Part IV:Energy Minimizing Models: The BSB Model. Error Correction. Simple Layered Networks. Auto-association. Two Layered Networks. Hetero-association. Multilayer Systems. Energy Minimizing Auto-associators. Part V:Demonstrations. A Neural Net Knowledge Base. Distributed Codes vs. Grandmother Cells. Problem: No Audit Trail. Combining Information From Many Areas. "Intuitive" Systems. Learning the Functional Dependencies in Ohm's Law. Ambiguity in Language. Demonstration of Disambiguation With a Neural Net. Part VI:Conclusions. Ten Conclusions Concerning the Potentials and Limitations of Neural Networks.
