Dynamically stable associative learning neural system with one fixed weight
First Claim
1. A neural network comprising:
- at least one conditioned signal input neuron and at least one unconditioned signal input neuron;
an output neuron connected to each said unconditioned signal input neuron along an unconditioned signal pathway having a fixed excitory weight associated therewith and said conditioned signal input neuron connected to each said output neuron along an conditioned signal pathway have a variable weight associated therewith;
means for defining a matrix of multi-bit digital weight values associated with each conditioned signal path, the weights having a normalized range of -1 to +1; and
means for determining the degree of similarity between a matrices representative of an input signal pattern applied to at least each said unconditioned signal input neuron and said matrices or multi-bit digital weight values associated with each conditioned signal path and for increasing the weight along those conditioned signal pathways having a degree of similarity above a predetermined similarity threshold.
5 Assignments
0 Petitions
Accused Products
Abstract
A dynamically stable associative learning neural network system include a plurality of synapses (122,22-28), a non-linear function circuit (30) and an adaptive weight circuit (150) for adjusting the weight of each synapse based upon the present signal and the prior history of signals applied to the input of the particular synapse and the present signal and the prior history of signals applied to the input of a predetermined set of other collateral synapses. A flow-through neuron circuit (1110) embodiment includes a flow-through synapse (122) having a predetermined fixed weight. A neural network is formed by a set of flow-through neuron circuits connected by flow-through synapses to form separate paths between each input (215) and a corresponding output (245). In one embodiment (200), the neuron network is initialized by setting the adjustable synapses at some value near the minimum weight and setting the flow-through neuron circuits at some arbitrarily high weight. The neural network embodiments are taught by successively application of sets of inputs signals to the input terminals until a dynamic equilibrium is reached.
-
Citations
3 Claims
-
1. A neural network comprising:
-
at least one conditioned signal input neuron and at least one unconditioned signal input neuron; an output neuron connected to each said unconditioned signal input neuron along an unconditioned signal pathway having a fixed excitory weight associated therewith and said conditioned signal input neuron connected to each said output neuron along an conditioned signal pathway have a variable weight associated therewith; means for defining a matrix of multi-bit digital weight values associated with each conditioned signal path, the weights having a normalized range of -1 to +1; and means for determining the degree of similarity between a matrices representative of an input signal pattern applied to at least each said unconditioned signal input neuron and said matrices or multi-bit digital weight values associated with each conditioned signal path and for increasing the weight along those conditioned signal pathways having a degree of similarity above a predetermined similarity threshold. - View Dependent Claims (2, 3)
-
Specification