For Netscape 2 users, there is now a Framed version of this document

page1

TRITA FYS 9012
KTH-Frescati

REVIEW OF HARDWARE NEURAL NETWORKS: A USER'S PERSPECTIVE 

Clark S. Lindsey and Thomas Lindblad

Physics Dept. - Frescati, Royal Institute of Technology Frescativägen 24, 104 05 Stockholm, Sweden

lindsey@particle.kth.se,lindblad@vana.physto.se

Today one can choose from a wide range of neural network hardware. The most important benefit of such hardware is the great increase in speed over conventional sequential processors. The review here surveys a sample of neural network VLSI chips, accelerator boards, and multi-board neurocomputers. We look at the hardware from the potential users viewpoint and discuss some systems developed for high energy physics applications.
 

• Introduction
• Hardware Specifications
• Neural Networks in VLSI
• Digital
• Slice Architectures
• Multi-processor Chips
• Radial Basis Functions
• Other Digital Designs
• Analog
• Hybrid
• PC Accelerator Cards and Neurocomputers
• Neural Network Hardware in High Energy Physics
• Discussion
• References
• About this document ...