Photo stimulated and controlled imaging neural network

Claim 2

A photo stimulated imaging neural network for processing an image which is directly imaged as an input onto the neural network, comprising: a direct imaging means utilizing light from a scene or an image of the scene for impressing an input image of light from the scene or image directly on said neural network as an input into said neural network; a photo receptor array means for converting said input image from said imaging means into a pattern of photo controlled electrical coupling to produce a plurality of electrical dendrite input signals; and a neuron amplifier array means having a plurality of individual electrical neuron amplifiers, said neuron amplifier array means providing a plurality of electrical axon output signals to a programmable gate array means in response to receipt of and stimulation by said plurality of electrical dendrite input signals from said photo receptor array means, said programmable gate array means comprising an array of electrical gates which is configured by programming to perform a particular logical operation.

Claim 3

The photostimulated imaging neural network according to claim 1 wherein said imaging means radiometrically forms an image directly on said photo receptor array means.

Claim 4

The photo stimulated imaging neural network according to claim 2 wherein said imaging means is a lens.

Claim 5

The photo stimulated imaging neural network according to claim 2 wherein said imaging means is a cathode ray tube.

Claim 6

The photo stimulated imaging neural network according to claim 2 wherein said imaging means is an intensifier.

Claim 7

The photo stimulated imaging neural network according to claim 2 wherein said imaging means is a laser scanner.

Claim 8

The photo stimulated imaging neural network according to claim 1 wherein said photoreceptor array means comprises: a plurality of synaptic photo controlled resistors which provide synapse-like interconnections between said plurality of individual neuron amplifiers, said synaptic photo controlled resistors emulating biological synaptic couplings in response to said stimulus data from said imaging means; a voltage supply means for providing an operating voltage to each of said individual neuron amplifiers; and a plurality of current limiting photo controlled resistors connected between said voltage supply means and said plurality of synaptic photo controlled resistors.

Claim 9

The photo stimulated imaging neural network according to claim 7 wherein said plurality of synaptic photo controlled resistors and said plurality of current limiting photocontrolled resistors are photo field effect transistors.

Claim 10

The photo stimulated imaging neural network according to claim 7 wherein said plurality of synaptic photo controlled resistors and said plurality of current limiting photo controlled resistors are bipolar photo transistors.

Claim 11

The photo stimulated imaging neural network according to claim 7 wherein said plurality of synaptic photo controlled resistors and said plurality of current limiting photo controlled resistors are photodiodes.

Claim 12

The photo stimulated imaging neural network according to claim 7 wherein said plurality of synaptic photo controlled resistors respond to incident radiation wave length and intensity with a specific value of resistance thus determining the feedback and coupling pattern between the plurality of individual neuron amplifiers.

Claim 13

The photo stimulated imaging neural network according to claim 11 wherein each synaptic photo controlled resistor of said plurality of synaptic photo controlled resistors provides equal resistance when stimulated by equal intensity light.

Claim 14

The photo stimulated imaging neural network according to claim 12 wherein said plurality of synaptic photo controlled resistors provides feedback and coupling from an output of one individual neuron amplifier to an input of all other individual neuron amplifiers in said plurality of individual neuron amplifiers.

Claim 15

The photo stimulated imaging neural network according to claim 13 wherein said plurality of individual neuron amplifiers each have an inhibitory negative feedback axon output and an enabling positive feedback axon output, said enabling positive feedback axon outputs are connected to said gate array.

Claim 16

The photo stimulated imaging neural network according to claim 14 wherein said inhibitory negative feedback axon outputs and said enabling positive feedback axon outputs of each individual neuron amplifier are coupled by said synaptic photo controlled resistors to all remaining individual neuron amplifiers resulting in a number of couplings one less then that of the number of said plurality of individual neuron amplifiers.

Claim 17

The photo stimulated imaging neural network according to claim 15 wherein said plurality of individual neuron amplifiers settle into a set of stable states based on the unique coupling and feedback which is a function of said stimulus data from said imaging means, said set of stable states constitutes a specific learning set.

Claim 18

The photo stimulated imaging neural network according to claim 16 wherein said learning set consists of binary states which insofar as position or level are equal contributors to said learning set, said learning set is stored and utilized by said gate array which is controlled by a control means, said gate array outputs a plurality of output gates to a summing means wherein said output gates are summed and passed through a threshold device for comparison or storage.

Claim 19

The photo stimulated imaging neural network according to claim 16 wherein said plurality of individual neuron amplifiers settle into said set of stable states in a non-linear fashion as a function of a predetermined firing rate time constant and sigmoid response of each individual neuron amplifier.

Claim 20

The photo stimulated imaging neural network according to claim 18 wherein said sigmoid response of each individual neuron amplifier provides each amplifier with remote cutoff and saturation transfer characteristics.

Claim 21

The photo stimulated imaging neural network according to claim 19 wherein said sigmoid response is provided by a local parallel feedback network comprised of a first diode/zener diode branch which conducts from said input to said individual neuron amplifier to said output of said individual neuron amplifier and a second diode/zener diode branch which conducts in an opposite direction as that from said first branch.

Claim 22

The photo stimulated imaging neural network according to claim 18 wherein said predetermined firing rate time constant of each individual neuron amplifier determines the duration of time in which it takes said individual neuron amplifier to settle into a stable state after being excited.

Claim 23

The photo stimulated imaging neural network according to claim 21 wherein said predetermined firing rate time constant is provided by a parallel network consisting of a resistor and capacitor connected across each said individual neuron amplifier.

Claim 24

The photo stimulated imaging neural network according to claim 22 wherein said resistor and capacitor of said parallel network are photo controlled.

Claim 25

The photo stimulated imaging neural network according to claim 23 wherein said photo controlled capacitor is a photo controlled depletion layer capacitance of a pn junction.

Claim 26

The photo stimulated imaging neural network according to claim 21 wherein said photo stimulated imaging neural network further comprises: a photo interconnect and response array for providing connectivity and temporal control for said network in response to a plurality of external stimulus signals; and a photo stimulus means for providing a light pattern which inpinges on said photo interconnect and response array to form an operational network, said light pattern constitutes said plurality of external stimulus signals.

Claim 27

The photo stimulated imaging neural network according to claim 25 wherein said photo interconnect and response array comprises a plurality of elements each comprising: at least one neuron amplifier disconnect means for controlling the input and output connection of each of said individual neuron amplifiers; and a parallel network consisting of a resistor and capacitor which is connected across each of said individual neuron amplifiers.

Claim 28

The photo stimulated imaging neural network according to claim 26 wherein said neuron amplifier disconnect means comprises two photo controlled resistors, wherein one photo controlled resistor is connected across the input of each of said individual neuron amplifiers and the second is connected across said output of each of said individual neuron amplifiers.

Claim 29

The photo stimulated imaging neural network according to claim 27 wherein said predetermined firing rate time constant is provided by said parallel network of said photo interconnect and response array.

Claim 30

The photo stimulated imaging neural network according to claim 28 wherein said resistor and capacitor are photo controlled.

Claim 31

The photo stimulated imaging neural network according to claim 27 wherein said neuron amplifier disconnect means is operable to disconnect any or all of said individual neuron amplifiers thereby determining how many of said individual neuron amplifiers contribute to said learning set.

Claim 32

A photostimulated imaging neural network for processing an image which is directly imaged as an input onto the neural network, comprising: a direct imaging means utilizing light from a scene or an image of the scene for impressing an input image of light from the scene or image directly on said neural network as an input into said neural network; a combined photo receptor and photo interconnect and response array means for converting said input image from said imaging means into a pattern of photo controlled electrical couplings to produce a plurality of electrical dendrite input signals, and for providing connectivity and temporal control for said network in response to a plurality of photo stimulus signals provided by a photo stimulus means; a neuron amplifier array means having a plurality of individual electrical neuron amplifiers, said neuron amplifier array means providing a plurality of electrical axon output signals to a programmable gate array means in response to receipt of and stimulation by said plurality of electrical dendrite input signals from said combined photo receptor and photo interconnect and response array means, said programmable gate array means comprising an array of electrical gates which is configured by programming to perform a particular logical operation; and a photo stimulus means for providing a light pattern which impinges on said combined photo receptor and photo interconnect and response array to form an operational network, said light pattern constituting said plurality of photo stimulus signals.

Claim 33

A photostimulated imaging neural network for processing an image which is directly imaged as an input onto the neural network, comprising: a direct imaging means utilizing light from a scene or an image of the scene for impressing an input image of light from the scene or image directly on said neural network as an input into said neural network; a combined photo receptor and photo interconnect and response array means for converting said input image from said imaging means into a pattern of photo controlled electrical couplings to produce a plurality of electrical dendrite input signals, and for providing connectivity and temporal control for said network in response to a plurality of photo stimulus signals provided by a photo stimulus means; a neuron amplifier array means having a plurality of individual electrical neuron amplifiers, said neuron amplifier array means providing a plurality of electrical axon output signals to a programmable gate array means in response to receipt of and stimulation by said plurality of electrical dendrite input signals from said combined photo receptor and photo interconnect and response array means, said programmable gate array means comprising an array of electrical gates which is configured by programming to perform a particular logical operation; and a photo stimulus means for providing a light pattern which impinges on said combined photo receptor and photo interconnect and response array to form an operational network, said light pattern constituting said plurality of photo stimulus signals.

Claim 34

A photo stimulated imaging neural network according to claim 31 wherein said imaging means radiometrically images said data onto said photo receptor and photo interconnect and response array means.

Claim 35

The photo stimulated imaging neural network according to claim 32 wherein said imaging means is a lens.

Claim 36

The photo stimulated imaging neural network according to claim 32 wherein said imaging means is a cathode ray tube.

Claim 37

The photo stimulated imaging neural network according to claim 32 wherein said imaging means is an intensifier.

Claim 38

The photo stimulated imaging neural network according to claim 32 wherein said imaging means is a laser scanner.

Claim 39

The photo stimulated imaging neural network according to claim 31 wherein said photo receptor and photo interconnect and response array means comprises: a plurality of synaptic photo controlled resistors which provide synapse-like interconnections between said plurality of individual neuron amplifiers, said synaptic photo controlled resistors emulating biological synaptic couplings in response to said stimulus data from said imaging means; a voltage supply means for providing an operating voltage to each of said individual neuron amplifiers; a plurality of a current limiting photo controlled resistors connected between said voltage supply means and said plurality of synaptic photo controlled resistors; a neuron amplifier disconnect means for controlling the input and output connection of each individual neuron amplifier; and a parallel network consisting of a resistor and capacitor which is connected across each of said individual neuron amplifiers.

Claim 40

The photo stimulated imaging neural network according to claim 37 wherein said predetermined firing rate time constant is provided by said parallel network of said photo receptor and photo interconnect and response array means.

Claim 41

The photo stimulated imaging neural network according to claim 37 wherein said plurality of synaptic photo controlled resistors and said plurality of current limiting photo controlled resistors are photo field effect transistors.

Claim 42

The photo stimulated imaging neural network according to claim 37 wherein said plurality of synaptic photo controlled resistors and said plurality of current limiting photo controlled resistors are bipolar photo transistors.

Claim 43

The photo stimulated imaging neural network according to claim 37 wherein said plurality of synaptic photo controlled resistors and said plurality of current limiting photo controlled resistors are photo diodes.

Claim 44

The photo stimulated imaging neural network according to claim 37 wherein said plurality of synaptic photo controlled resistors respond to incident radiation wave length and intensity with a specific value of resistance thus determining the feedback and coupling pattern between the plurality of individual neuron amplifiers.

Claim 45

The photo stimulated imaging neural network according to claim 42 wherein each synaptic photo controlled resistor of said plurality of synaptic photo controlled resistors provides equal resistance when stimulated by equal intensity light.

Claim 46

The photo stimulated imaging neural network according to claim 43 wherein said plurality of synaptic photo controlled resistors provide feedback and coupling from an output of one individual neuron amplifier to an input of all other individual neuron amplifiers in said plurality of individual neuron amplifiers.

Claim 47

The photo stimulated imaging neural network according to claim 37 wherein said neuron amplifier disconnect means comprises two photo controlled resistors, wherein one photo controlled resistor is connected across said input of each of said individual neuron amplifier and the second is connected across said output of each of said individual neuron amplifiers.

Claim 48

The photo stimulated imaging neural network according to claim 45 wherein said plurality of individual neuron amplifiers each have an inhibitory negative feedback axon output and an enabling positive feedback axon output, said enabling positive feedback axon outputs are connected to said gate array.

Claim 49

The photo stimulated imaging neural network according to claim 46 wherein said inhibitory negative feedback axon outputs and said enabling positive feedback axon outputs of each individual neuron amplifier are coupled by said synaptic photo resistors to all remaining individual neuron amplifiers resulting in a number of couplings one less than that of the number of said plurality of individual neuron amplifiers.

Claim 50

The photo stimulated imaging neural network according to claim 47 wherein said plurality of individual neuron amplifiers settle into a set of stable states based on the unique coupling and feedback which is a function of said stimulus data from said imaging means, said stable states constitute a specific learning set.

Claim 51

The photo stimulated imaging neural network according to claim 48 wherein said learning set consists of binary states which insofar as position or level are equal contributors to said learning set, said learning set is stored and utilized by said gate array which is controlled by a control means, said gate array outputs a plurality of output gates to a summing means wherein said output gates are summed and passed through a threshold device for comparison or storage.

Claim 52

The photo stimulated imaging neural network according to claim 49 wherein said plurality of individual neuron amplifiers settle into said set of stable states in a non-linear fashion as a function of a predetermined firing rate time constant and sigmoid response of each individual neuron amplifier.

Claim 53

The photo stimulated imaging neural network according to claim 50 wherein said sigmoid response of each individual neuron amplifier provides each amplifier with remote cutoff and saturation transfer characteristics.

Claim 54

The photo stimulated imaging neural network according to claim 50 wherein said predetermined firing rate time constant of each individual neuron amplifier determines the duration of time in which it takes said individual neuron amplifier to settle into a stable state after being excited.

Claim 57

The method according to claim 53 wherein said image forming step includes radiometrically imaging said stimulus data onto said photo receptor means.

Claim 58

A method according to claim 54 wherein said converting step comprises the steps of: passing said stimulus data through a plurality of synaptic photo controlled resistors that comprise said photo receptor array and which provide synapse-like interconnections between a plurality of individual neuron amplifiers, said plurality of synaptic photo controlled resistors emulating biological synaptic couplings in response to said stimulus data; providing feedback between said plurality of individual neuron amplifiers; and supplying an initialization voltage to said plurality of individual neuron amplifiers.

Claim 59

The method according to claim 55 wherein said step of providing a plurality of axon output signals step comprises the steps of: converting said plurality of electrical dendrite input signals into said plurality of axon output signals by passing said input signals through said plurality of individual neuron amplifiers wherein through said synaptic couplings and feedback, said plurality of neuron amplifiers settle into a set of stable states; and providing said set of stable states to said gate array means wherein they are electronically summed for further learning functions and comparisons.

Claim 60

The method according to claim 56 wherein said method further comprises the steps of: controlling the input and output connections of each neuron amplifier of said plurality of individual neuron amplifiers; and controlling the temporal response of said photo stimulated imaging neural network.

Claim 61

The method according to claim 57 wherein said controlling the input and output connections step comprises switching any of said individual neuron amplifiers from said plurality of individual neuron amplifiers in and out of said photo stimulated imaging neural network.

Claim 62

The method according to claim 57 wherein said controlling the temporal response step comprises introducing a firing rate time constant into said photo stimulated imaging neural network.