Neural processor

• US 7,174,325 B1
• Filed: 05/29/2003
• Issued: 02/06/2007
• Est. Priority Date: 06/07/2002
• Status: Expired due to Fees

• Current Assignee: George Mason Intellectual Properties, Inc.
• Original Assignee: George Mason Intellectual Properties, Inc.
• Inventors: Ascoli, Giorgio A.
• Primary Examiner(s): Knight; Anthony
• Assistant Examiner(s): Holmes; Michael B.
• Application Number:
  US10/446,840
• Time in Patent Office: 1,349 Days
• Field of Search: 706/43, 706/34
• US Class Current: 706/43
• CPC Class Codes: G06N 3/049 Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
• CPC Class Codes: G06N 3/063 using electronic means

Abstract

Disclosed is a digital neural processor comprising at least one neural processing element. The neural processing elements including at least one simulated dendrite and a simulated axon. Each of the simulated dendrites may include: a dendrite input capable of receiving at least one dendrite input signal and a dendrite signal propagation function capable of calculating a dendrite output signal in discrete time steps from each dendrite input signal. The signal propagation function may further include a delay parameter; a duration parameter; and an amplitude parameter. The simulated axon includes an axon input capable of receiving dendrite output signals, an axon function, capable of calculating an axon output signal from dendrite output signal(s) and an axon output capable of outputting the axon output signal.

Claims

1. A digital neural processor comprising at least one neural processing element, said neural processing element including:

a. at least one simulated dendrite, each of said simulated dendrites including;

i. a dendrite input capable of receiving at least one dendrite input signal;

ii. a dendrite signal propagation function for calculating a square pulse dendrite output signal in discrete time steps from each dendrite input signal, said square pulse dendrite output signal defined by;

1. a delay parameter;

2. a duration parameter; and

3. an amplitude parameter; and

b. a simulated axon including;

i. an axon input capable of receiving each said square pulse dendrite output signal from said at least one simulated dendrite;

ii. an axon function, capable of calculating an axon output signal from at least one square pulse dendrite output signal and from at least one previous axon output signals; and

iii. an axon output capable of outputting said axon output signal.

2. A neural processor according to claim 1, wherein each of said at least one dendrite input further includes a timestamp.

3. A neural processor according to claim 1, wherein said delay parameter is represented by a relative timestamp.

4. A neural processor according to claim 3, wherein said timestamp is an integer timestamp.

5. A neural processor according to claim 1, wherein said amplitude parameter is constant during the duration of a stimulus period.

6. A neural processor according to claim 1, wherein said duration parameter is represented by a relative timestamp.

7. A neural processor according to claim 6, wherein said timestamp is an integer timestamp.

8. A neural processor according to claim 1, wherein said axonal function includes a refractory period.

9. A neural processor according to claim 1, wherein said axon function includes a summation function.

10. A neural processor according to claim 1, wherein said axon function includes a threshold.

11. A neural processor according to claim 10, wherein said axon output signal is generated by said axon function when said threshold is exceeded and when said axon output was not generated during the previous said refractory period.

12. A neural processor according to claim 10, wherein a desired activity level is considered in selecting said threshold.

13. A neural processor according to claim 1, wherein said axon output signal is a square wave.

14. A neural processor according to claim 1, further including a neural topology, said neural topology defined by connecting at least one of said axon output of at least one of said at least one neural processing element with at least one of said dendrite input of at least one of another of said at least one neural processing element.