Action based learning

• US 8,175,984 B2
• Filed: 12/05/2008
• Issued: 05/08/2012
• Est. Priority Date: 12/05/2007
• Status: Active Grant

• Current Assignee: Numenta, Inc.
• Original Assignee: Numenta, Inc.
• Inventors: George, Dileep
• Primary Examiner(s): Gaffin, Jeffrey A
• Assistant Examiner(s): Afolabi, Ola Olude
• Application Number:
  US12/315,957
• Publication Number:
  US 20090150311A1
• Time in Patent Office: 1,250 Days
• Field of Search: 706/12, 706/16, 706/52
• US Class Current: 706/12
• 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

Abstract

A set of sequences of sensed input patterns associated with a set of actions is generated by performing at least a first action on data derived from a real-world system. A subset of the sequences of sensed input patterns that form a group associated with the first action is determined. A new sequence of sensed input patterns is received. A first value which indicates the probability that the new sequence of sensed input patterns is associated with the first action based on the subset of sequences of sensed input patterns is determined and stored in a memory associated with the computer system.

Claims

1. A computer system comprising:

a memory;

a processor;

a child node module stored in the memory and executable by the processor to;

receive a set of sequences of sensed input patterns associated with a set of actions, wherein the set of actions are associated with a hierarchy of actions that defines a set of subsumptive relationships between the actions;

determine that a subset of the sequences of sensed input patterns form a group associated with a first action, wherein the first action is associated with a hierarchy of actions;

receive a new sequence of sensed input patterns;

determine a first value which indicates the probability that the new sequence of sensed input patterns is associated with the first action based on the group associated with the first action;

label the new sequence of sensed input patterns with the first action and at least a second action responsive to the first value exceeding a threshold value, wherein the at least a second action is determined according to the set of subsumptive relationships defined by the hierarchy of actions; and

store the first value in a memory associated with the child node.

2. The system of claim 1, wherein the child node is further adapted to transmit the first value to a parent node.

3. The system of claim 1, further comprising labeling the set of sequences of sensed input patterns with the set of actions.

4. The system of claim 1, wherein the set of sequences of sensed input patterns associated with the set of actions are derived from computationally performing the set of actions on real-world data.

5. The system of claim 1, wherein the set of sequences of sensed input patterns associated with the set of actions are derived from physically performing the set of actions on one or more real-world systems.

6. A computer-readable storage medium encoded with executable computer program code, the program code comprising program code for:

receiving, at a child node, a set of sequences of sensed input patterns associated, with a set of actions, wherein the set of actions are associated with a hierarchy of actions that defines a set of subsumptive relationships between the actions;

determining, at the child node, that a subset of the sequences of sensed input patterns form a group associated with a first action, wherein the first action is associated with a hierarchy of actions;

receiving, at the child node, a new sequence of sensed input patterns;

determining, at the child node, a first value which indicates the probability that the new sequence of sensed input patterns is associated with the first action based on the group associated with the first action;

responsive to the first value exceeding and at least second action; and

storing the first value in a memory associated with the child node.

7. The medium of claim 6, further comprising program code for a parent node and the child node comprise a hierarchical network of nodes.

8. The medium of claim 6, further comprising labeling the set of sequences of sensed input patterns with the set of actions.

9. The medium of claim 6, wherein the set of sequences of sensed input patterns associated with the set of actions are derived from computationally performing the set of actions on real-world data.

10. The medium of claim 6, wherein the set of sequences of sensed input patterns associated with the set of actions are derived from computationally performing the set of actions on real-world data.