Abstract
The fault-tolerance of a new combinatorial neural network model, called the summation network, is investigated. The summation network uses processors which have n inputs, each labelled with a positive or negative integer weight of O(n log n) bits, and O(n log n) outputs. The outputs encode, in binary, the sum of the weights of the 'on' inputs to the processor. The processors are interconnected in a network with unlimited fan-in and fan-out. It is shown that fault-tolerance of the summation network is directly related to the product of error-probability and fan-in. Intuitively, the larger the fan-in, the more likely it is that an error in another part of the network will affect an otherwise correct processor.
Original language | English (US) |
---|---|
Number of pages | 1 |
Journal | Neural Networks |
Volume | 1 |
Issue number | 1 SUPPL |
DOIs | |
State | Published - Jan 1 1988 |
Event | International Neural Network Society 1988 First Annual Meeting - Boston, MA, USA Duration: Sep 6 1988 → Sep 10 1988 |
All Science Journal Classification (ASJC) codes
- Cognitive Neuroscience
- Artificial Intelligence