Neural network system and circuit for use therein
First Claim
1. A neuronal circuit for a neural network system which circuit enables digital determination of new states of M neurons by determining products of states of N input neurons and M input vectors of N synaptic coefficients which are associated with each interconnection between neurons and which are stored in a programmable memory,characterized in that the neuronal circuit comprises an integrated circuit for performing for a given neuron chosen from the M neurons, parallel processing of all synaptic coefficients of the given neuron and which comprises:
- (a) the programmable digital memory which is integrated in the circuit for storing the N×
M synaptic coefficients of all M neurons encoded on p bits, and for simultaneously supplying all synaptic coefficients of the given neuron,(b) N digital multipliers for determining the products of the N synaptic coefficients of the given neuron and the states of the N input neurons,(c) an adder tree for determining the sum S of all said products for the given neuron, and(d) a digital state memory for receiving the states of the N input neurons which may be encoded from 1 to n bits in a reconfigurable mode, the circuit being controlled by control means,where N, M, p, and n represent positive integers and S represents a number.
1 Assignment
0 Petitions
Accused Products
Abstract
A neural network system comprises a memory for storing in binary code the synaptic coefficients indicative of the interconnections among the neurons. Means are provided for simultaneously supplying all the synaptic coefficients associated with a given neuron. Digital multipliers are provided for determining the product of the supplied synaptic coefficients and the relevant neuron states of the neurons connected to said given neuron. The multipliers deliver their results into an adder tree for determining the sum of the products. As a result of the parallel architecture of the system high operating speeds are attainable. The modular architecture enables extension of the system.
98 Citations
29 Claims
-
1. A neuronal circuit for a neural network system which circuit enables digital determination of new states of M neurons by determining products of states of N input neurons and M input vectors of N synaptic coefficients which are associated with each interconnection between neurons and which are stored in a programmable memory,
characterized in that the neuronal circuit comprises an integrated circuit for performing for a given neuron chosen from the M neurons, parallel processing of all synaptic coefficients of the given neuron and which comprises: -
(a) the programmable digital memory which is integrated in the circuit for storing the N×
M synaptic coefficients of all M neurons encoded on p bits, and for simultaneously supplying all synaptic coefficients of the given neuron,(b) N digital multipliers for determining the products of the N synaptic coefficients of the given neuron and the states of the N input neurons, (c) an adder tree for determining the sum S of all said products for the given neuron, and (d) a digital state memory for receiving the states of the N input neurons which may be encoded from 1 to n bits in a reconfigurable mode, the circuit being controlled by control means, where N, M, p, and n represent positive integers and S represents a number. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29)
-
Specification