Hierarchical computing modules for performing recognition using spatial distance and temporal sequences

• US 7,613,675 B2
• Filed: 01/11/2007
• Issued: 11/03/2009
• Est. Priority Date: 02/10/2006
• Status: Active Grant

• Current Assignee: Numenta, Inc.
• Original Assignee: Numenta, Inc.
• Inventors: Hawkins, Jeffrey, George, Dileep
• Primary Examiner(s): Vincent; David R
• Assistant Examiner(s): Kennedy; Adrian L
• Application Number:
  US11/622,457
• Publication Number:
  US 20070192270A1
• Time in Patent Office: 1,027 Days
• Field of Search: 706/52, 706/55, 706/16, 706/23, 700/44, 700/47
• US Class Current: 706/55
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 3/04 Architecture, e.g. interconnection topology
• CPC Class Codes: G06N 3/045 Combinations of networks
• CPC Class Codes: G06N 3/047 Probabilistic or stochastic networks
• CPC Class Codes: G06N 3/049 Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

A hierarchy of computing modules is configured to learn a cause of input data sensed over space and time, and is further configured to determine a cause of novel sensed input data dependent on the learned cause. The hierarchy is further configured to associate a first pattern in the input data and a second pattern in the input data to a same possible cause of the input data.

Claims

1. A computer-implemented system for determining an identity of a sensed object or a state of a sensed object associated with spatial patterns and temporal sequences in input data, the input data representing at least one of image, video, audio, text, weather conditions, tactile data or data associated with operation of a machine, comprising:

a hierarchy of computing modules configured to receive first input data to learn spatial patterns and temporal sequences in the first input data associated with a sensed object or a state of a sensed object in a learning stage, the hierarchy in an inference stage subsequent to the learning stage further configured to receive second input data and generate identity information representing probabilities that spatial patterns and temporal sequences in the second input data correspond to the spatial patterns and temporal sequences learned by the hierarchy in the learning stage, the hierarchy of computing modules comprising;

a first module configured to output first information about patterns responsive to receiving a first spatial pattern in the second input data, the first information about patterns representing information about spatial patterns and temporal sequences in the second input data corresponding to the learned spatial patterns and temporal sequences, the first module configured to output the same first information about patterns responsive to receiving a second spatial pattern in the second input data within a spatial distance from the first spatial pattern; and

a second module associated with the first module to receive the first information, the second module configured to generate the identity information, wherein the identity information is one of displayed, output for controlling a physical object or stored in a storage medium, for identification of a sensed object or a state of a sensed object.

2. The system of claim 1, wherein the first module receives a control signal from the second module directing the first module to output the same first information about patterns for the first spatial pattern and the second spatial pattern.

3. The system of claim 2, wherein the second module has an input range greater than an input range of the first module.

4. The system of claim 1, wherein at least part of the hierarchy is implemented in hardware.

9. The system of claim 1, wherein the first module is configured to determine that the second spatial pattern is within the spatial distance from the first spatial pattern responsive to a predetermined number of bits in the first spatial pattern coinciding with bits in the second spatial pattern.

10. The system of claim 1, wherein the second module is configured to output the same identity information responsive to receiving second information, the second information within a spatial distance from the first information.

5. A computer-implemented method for determining an identity of a sensed object or a state of a sensed object associated with spatial patterns and temporal sequences in input data, the input data representing at least one of image, video, audio, text, weather conditions, tactile data or data associated with operation of a machine, comprising:

learning patterns and sequences in first input data associated with a sensed object or a state of a sensed object responsive to receiving the first input data in a learning stage at a hierarchy of computing modules comprising at least a first module and a second module;

the first module in an inference stage subsequent to the learning stage outputting first information about patterns responsive to receiving a first spatial pattern in second input data, the first information about patterns representing information about spatial patterns and temporal sequences in the second input data corresponding to the learned spatial patterns and temporal sequences;

the first module in the inference stage outputting the same first information about patterns responsive to receiving a second spatial pattern in the second input data within a spatial distance from the first spatial pattern; and

the second module generating identity information responsive to receiving the first information about patterns, wherein the identity information is one of displayed, output for controlling a physical object or stored in a storage medium, for identification of a sensed object or a state of a sensed object.

6. The computer-implemented method of claim 5, further comprising:

providing a control signal generated from the second module to the first module, the control signal directing the first module to generate the same first information about patterns for the first spatial pattern and the second spatial pattern.

11. The computer-implemented method of claim 5, further comprising determining that the second spatial pattern is within the spatial distance from the first spatial pattern by comparing bits in the first spatial pattern with bits in the second spatial pattern.

12. The computer-implemented method of claim 5, wherein generating the identity information comprises generating the same identity information responsive to receiving second information within a spatial distance from the first information.

7. A computer-readable storage medium having instructions therein that are executable by a processor for determining an identity of a sensed object or a state of a sensed object associated with spatial patterns and temporal sequences in input data, the input data representing at least one of image, video, audio, text, weather conditions, tactile data or data associated with operation of a machine, the instructions comprising instructions to:

learn patterns and sequences in first input data associated with a sensed object or a state of a sensed object responsive to receiving the first input data in a learning stage at a hierarchy of computing modules comprising at least a first module and a second module;

output, in an inference stage subsequent to the learning stage, first information about patterns from the first module responsive to receiving a first spatial pattern in second input data, the first information about patterns representing information about spatial patterns and temporal sequences in the second input data corresponding to the learned spatial patterns and temporal sequences;

output, in the inference state, the same first information about patterns from the first module responsive to receiving a second spatial pattern in the second input data within a spatial distance from the first spatial pattern; and

generate at the second module identity information responsive to receiving the first information about patterns, wherein the identity information is one of displayed, output for controlling a physical object or stored in a storage medium, for identification of a sensed object or a state of a sensed object.

8. The computer-readable medium of claim 7, providing a control signal generated by the second module to the first module, the control signal directing the first module to generate the same first information about patterns for the first spatial pattern and the second spatial pattern.

13. The computer-readable medium of claim 7, further comprising instructions to determine whether the second spatial pattern is within the spatial distance by comparing bits in the first spatial pattern with bits in the second spatial pattern.

14. The computer-readable medium of claim 7, wherein instructions to generate the output information comprises instructions to generate the same output information responsive to receiving the second information within a spatial distance from the first information.