Learning device with continuous configuration capability

• US 9,747,554 B2
• Filed: 05/23/2014
• Issued: 08/29/2017
• Est. Priority Date: 05/24/2013
• Status: Active Grant

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: Canoy, Michael-David Nakayoshi
• Primary Examiner(s): Misir, Dave
• Application Number:
  US14/286,244
• Publication Number:
  US 20140351180A1
• Time in Patent Office: 1,194 Days
• Field of Search: 706 12
• CPC Class Codes: G06N 20/00 Machine learning

Abstract

An embodiment method for continuous configuration of learning devices includes operations for storing, by a learning device within a decentralized system of a plurality of learning devices, events obtained while in a monitoring mode, activating a triggered mode for a reflex when at least one of the stored events corresponds to a trigger pattern, determining whether the reflex has a trigger weight exceeding a trigger weight threshold, conducting the predetermined action associated with the reflex when the trigger weight exceeds the trigger weight threshold, obtaining at least one additional event while in the triggered mode, adjusting the trigger weight of the reflex when the at least one additional event corresponds to a correction pattern or a reward pattern occurring in response to conducting the predetermined action, and creating a second reflex when the at least one additional event does not correspond to a known pattern.

Claims

1. A method for continuous configuration of a first learning device, comprising:

storing, by the first learning device, events obtained while in a monitoring mode;

activating, by the first learning device, a first triggered mode for a first reflex when at least one of the stored events corresponds to a first trigger pattern associated with the first reflex, wherein the first reflex indicates a correlation between at least the first trigger pattern and a predetermined action;

determining whether the first reflex has a trigger weight exceeding a trigger weight threshold;

conducting the predetermined action associated with the first reflex in response to determining that the trigger weight associated with the first reflex exceeds the trigger weight threshold;

obtaining at least one additional event while in the first triggered mode;

determining whether the at least one additional event corresponds to a first correction pattern or a first reward pattern;

adjusting the trigger weight of the first reflex when the at least one additional event corresponds to at least one of the first correction pattern and the first reward pattern associated with the first reflex and occurring in response to conducting the predetermined action, wherein the adjustment to the trigger weight of the first reflex is different in response to determining that the at least one additional event corresponds to the first correction pattern, than the adjustment to the trigger weight in response to determining that the at least one additional event corresponds to the first reward pattern; and

creating a second reflex with a second trigger pattern, a second correction pattern, and a second reward pattern when the at least one additional event does not correspond to at least one of the first trigger pattern, the first correction pattern, and the first reward pattern associated with the first reflex.

2. The method of claim 1, wherein adjusting the trigger weight of the first reflex further comprises:

increasing the trigger weight of the first reflex when the at least one additional event corresponds to the first reward pattern of the first reflex; and

decreasing the trigger weight of the first reflex when the at least one additional event corresponds to the first correction pattern of the first reflex.

3. The method of claim 2, further comprising:

entering a critical learning period with the first reflex during which a high gain set is applied to the trigger weight.

4. The method of claim 2, further comprising:

entering a steady state learning period with the first reflex during which a low gain set is applied to the trigger weight.

5. The method of claim 1, wherein obtaining the at least one additional event while in the first triggered mode further comprises:

retrieving prior events from a memory.

6. The method of claim 1, wherein storing, by the first learning device, events obtained while in the monitoring mode comprises:

storing, in a memory, at least one event for a window of time such that the at least one event and associated patterns are removed from the memory after the window of time.

7. The method of claim 1, wherein at least one of the stored events is obtained based on occurrence data that is generated locally.

8. The method of claim 1, wherein at least one of the stored events is obtained based on occurrence data that is generated remotely.

9. The method of claim 8, wherein the at least one of the stored events that is obtained based on the occurrence data that is generated remotely is generated in response to the first learning device receiving an event report message from a second learning device, wherein the first learning device and the second learning device are part of a decentralized system comprised of a plurality of learning devices in a location that are connected via a wired or a wireless communication link.

10. The method of claim 8, wherein the at least one of the stored events that is obtained based on the occurrence data that is generated remotely corresponds to at least one of the first reward pattern and the first correction pattern.

11. The method of claim 1, wherein at least one of the stored events is obtained based on occurrence data that is received from non-learning devices via a wired or wireless communication link.

12. The method of claim 1, further comprising:

concurrently activating a second triggered mode when the at least one additional event corresponds to a different trigger pattern associated with a third reflex.

13. The method of claim 1, wherein conducting the predetermined action associated with the first reflex in response to determining that the trigger weight associated with the first reflex exceeds the trigger weight threshold comprises:

generating a resulting event based on conducting the predetermined action; and

broadcasting an event report message that indicates the resulting event and that is capable of being used by another learning device.

14. The method of claim 1, further comprising:

receiving, at the first learning device, an event report message from another learning device, wherein the received event report message indicates another event that was generated by the another learning device in response to the another learning device obtaining another event that corresponds to the first trigger pattern.

15. The method of claim 1, wherein activating the first triggered mode for the first reflex associated with the predetermined action when at least one of the stored events corresponds to the first trigger pattern associated with the first reflex comprises:

applying a pattern filter to a pattern based on the at least one of the stored events; and

activating the first triggered mode for the first reflex associated with the predetermined action when the at least one of the stored events corresponds to the first trigger pattern associated with the first reflex and when the pattern is not discarded based on the applied pattern filter.

16. The method of claim 1, wherein the first reflex and the second reflex are only removable from the first learning device based on a garbage collection policy.

17. The method of claim 1, wherein conducting the predetermined action associated with the first reflex in response to determining that the trigger weight associated with the first reflex exceeds the trigger weight threshold comprises generating a pattern of events that are used by a motor driver to drive an actuator.

18. The method of claim 1, wherein one or more of the first correction pattern and the first reward pattern are order-dependent.

19. The method of claim 1, wherein one or more of the first correction pattern and the first reward pattern are order-independent.

20. The method of claim 1, wherein obtaining the at least one additional event while in the first triggered mode comprises obtaining the at least one additional event based on one or more periodic timing mechanisms.

21. The method of claim 20, further comprising polling one or more sensors based on the one or more periodic timing mechanisms.

22. The method of claim 20, wherein the one or more periodic timing mechanisms are shared by multiple learning devices.

23. A learning device, comprising:

a processor configured with processor-executable instructions for performing operations comprising;

storing, by the processor of the learning device, events obtained while in a monitoring mode;

activating, by the processor of the learning device, a first triggered mode for a first reflex when at least one of the stored events corresponds to a first trigger pattern associated with the first reflex, wherein the first reflex indicates a correlation between at least the first trigger pattern and a predetermined action;

determining whether the first reflex has a trigger weight exceeding a trigger weight threshold;

conducting the predetermined action associated with the first reflex in response to determining that the trigger weight associated with the first reflex exceeds the trigger weight threshold;

obtaining at least one additional event while in the first triggered mode;

determining whether the at least one additional event corresponds to a first correction pattern or a first reward pattern;

adjusting the trigger weight of the first reflex when the at least one additional event corresponds to at least one of the first correction pattern and the first reward pattern associated with the first reflex and occurring in response to conducting the predetermined action, wherein the adjustment to the trigger weight of the first reflex is different in response to determining that the at least one additional event corresponds to the first correction pattern, than the adjustment to the trigger weight in response to determining that the at least one additional event corresponds to the first reward pattern; and

creating a second reflex with a second trigger pattern, a second correction pattern, and a second reward pattern when the at least one additional event does not correspond to at least one of the first trigger pattern, the first correction pattern, and the first reward pattern associated with the first reflex.

24. The learning device of claim 23, wherein the processor is configured with processor-executable instructions for performing operations such that adjusting the trigger weight of the first reflex further comprises:

increasing the trigger weight of the first reflex when the at least one additional event corresponds to the first reward pattern of the first reflex; and

decreasing the trigger weight of the first reflex when the at least one additional event corresponds to the first correction pattern of the first reflex.

25. The learning device of claim 24, wherein the processor is configured with processor-executable instructions for performing operations further comprising:

entering a critical learning period with the first reflex during which a high gain set is applied to the trigger weight.

26. The learning device of claim 24, wherein the processor is configured with processor-executable instructions for performing operations further comprising:

entering a steady state learning period with the first reflex during which a low gain set is applied to the trigger weight.

27. The learning device of claim 23, wherein the processor is configured with processor-executable instructions for performing operations such that obtaining the at least one additional event while in the first triggered mode further comprises:

retrieving prior events from a memory.

28. The learning device of claim 23, wherein the processor is configured with processor-executable instructions for performing operations such that storing events obtained while in the monitoring mode comprises:

storing, in a memory, at least one event for a window of time such that the at least one event and associated patterns are removed from the memory after the window of time.

29. The learning device of claim 23, wherein at least one of the stored events is obtained based on occurrence data that is generated locally.

30. The learning device of claim 23, wherein at least one of the stored events is obtained based on occurrence data that is generated remotely.

31. The learning device of claim 30, wherein the at least one of the stored events that is obtained based on the occurrence data that is generated remotely is generated in response to the learning device receiving an event report message from another learning device, wherein the learning device and the another learning device are part of a decentralized system comprised of a plurality of learning devices in a location that are connected via a wired or a wireless communication link.

32. The learning device of claim 30, wherein the at least one of the stored events that is obtained based on the occurrence data that is generated remotely corresponds to at least one of the first reward pattern and the first correction pattern.

33. The learning device of claim 23, wherein at least one of the stored events is obtained based on occurrence data that is received from non-learning devices via a wired or wireless communication link.

34. The learning device of claim 23, wherein the processor is configured with processor-executable instructions for performing operations further comprising:

concurrently activating a second triggered mode when the at least one additional event corresponds to a different trigger pattern associated with a third reflex.

35. The learning device of claim 23, wherein the processor is configured with processor-executable instructions for performing operations such that conducting the predetermined action associated with the first reflex in response to determining that the trigger weight associated with the first reflex exceeds the trigger weight threshold comprises:

generating a resulting event based on conducting the predetermined action; and

broadcasting an event report message that indicates the resulting event and that is capable of being used by another learning device.

36. The learning device of claim 23, wherein the processor is configured with processor-executable instructions for performing operations further comprising:

receiving an event report message from another learning device, wherein the received event report message indicates another event that was generated by the another learning device in response to the another learning device obtaining another event that corresponds to the first trigger pattern.

37. The learning device of claim 23, wherein the processor is configured with processor-executable instructions for performing operations such that activating the first triggered mode for the first reflex associated with the predetermined action when at least one of the stored events corresponds to the first trigger pattern associated with the first reflex comprises:

applying a pattern filter to a pattern based on the at least one of the stored events; and

activating the first triggered mode for the first reflex associated with the predetermined action when the at least one of the stored events corresponds to the first trigger pattern associated with the first reflex and when the pattern is not discarded based on the applied pattern filter.

38. The learning device of claim 23, wherein the first reflex and the second reflex are only removable from the learning device based on a garbage collection policy.

39. The learning device of claim 23, wherein the processor is configured with processor-executable instructions for performing operations such that conducting the predetermined action associated with the first reflex in response to determining that the trigger weight associated with the first reflex exceeds the trigger weight threshold comprises generating a pattern of events that are used by a motor driver to drive an actuator.

40. The learning device of claim 23, wherein one or more of the first correction pattern and the first reward pattern are order-dependent.

41. The learning device of claim 23, wherein one or more of the first correction pattern and the first reward pattern are order-independent.

42. The learning device of claim 23, wherein the processor is configured with processor-executable instructions for performing operations such that obtaining the at least one additional event while in the first triggered mode comprises obtaining the at least one additional event based on one or more periodic timing mechanisms.

43. The learning device of claim 42, wherein the processor is configured with processor-executable instructions for performing operations further comprising polling one or more sensors based on the one or more periodic timing mechanisms.

44. The learning device of claim 42, wherein the one or more periodic timing mechanisms are shared by multiple learning devices.

45. A non-transitory processor-readable storage medium having stored thereon processor-executable instructions configured to cause a processor of a learning device to perform operations comprising:

storing, by the processor of the learning device, events obtained while in a monitoring mode;

activating, by the processor of the learning device, a first triggered mode for a first reflex when at least one of the stored events corresponds to a first trigger pattern associated with the first reflex, wherein the first reflex indicates a correlation between at least the first trigger pattern and a predetermined action;

determining whether the first reflex has a trigger weight exceeding a trigger weight threshold;

conducting the predetermined action associated with the first reflex in response to determining that the trigger weight associated with the first reflex exceeds the trigger weight threshold;

obtaining at least one additional event while in the first triggered mode;

determining whether the at least one additional event corresponds to a first correction pattern or a first reward pattern;

adjusting the trigger weight of the first reflex when the at least one additional event corresponds to at least one of the first correction pattern and the first reward pattern associated with the first reflex and occurring in response to conducting the predetermined action, wherein the adjustment to the trigger weight of the first reflex is different in response to determining that the at least one additional event corresponds to the first correction pattern, than the adjustment to the trigger weight in response to determining that the at least one additional event corresponds to the first reward pattern; and

creating a second reflex with a second trigger pattern, a second correction pattern, and a second reward pattern when the at least one additional event does not correspond to at least one of the first trigger pattern, the first correction pattern, and the first reward pattern associated with the first reflex.

46. The non-transitory processor-readable storage medium of claim 45, wherein the stored processor-executable instructions are configured to cause the processor to perform operations such that adjusting the trigger weight of the first reflex further comprises:

increasing the trigger weight of the first reflex when the at least one additional event corresponds to the first reward pattern of the first reflex; and

decreasing the trigger weight of the first reflex when the at least one additional event corresponds to the first correction pattern of the first reflex.

47. The non-transitory processor-readable storage medium of claim 46, wherein the stored processor-executable instructions are configured to cause the processor to perform operations further comprising:

entering a critical learning period with the first reflex during which a high gain set is applied to the trigger weight.

48. The non-transitory processor-readable storage medium of claim 46, wherein the stored processor-executable instructions are configured to cause the processor to perform operations further comprising:

entering a steady state learning period with the first reflex during which a low gain set is applied to the trigger weight.

49. The non-transitory processor-readable storage medium of claim 45, wherein the stored processor-executable instructions are configured to cause the processor to perform operations such that obtaining the at least one additional event while in the first triggered mode further comprises:

retrieving prior events from a memory.

50. The non-transitory processor-readable storage medium of claim 45, wherein the stored processor-executable instructions are configured to cause the processor to perform operations such that storing events obtained while in the monitoring mode comprises storing, in a memory, at least one event for a window of time such that the at least one event and associated patterns are removed from the memory after the window of time.

51. The non-transitory processor-readable storage medium of claim 45, wherein at least one of the stored events is obtained based on occurrence data that is generated locally.

52. The non-transitory processor-readable storage medium of claim 45, wherein at least one of the stored events is obtained based on occurrence data that is generated remotely.

53. The non-transitory processor-readable storage medium of claim 52, wherein the at least one of the stored events that is obtained based on the occurrence data that is generated remotely is generated in response to the learning device receiving an event report message from another learning device, wherein the learning device and the another learning device are part of a decentralized system comprised of a plurality of learning devices in a location that are connected via a wired or a wireless communication link.

54. The non-transitory processor-readable storage medium of claim 52, wherein the at least one of the stored events that is obtained based on the occurrence data that is generated remotely corresponds to at least one of the first reward pattern and the first correction pattern.

55. The non-transitory processor-readable storage medium of claim 45, wherein at least one of the stored events is obtained based on occurrence data that is received from non-learning devices via a wired or wireless communication link.

56. The non-transitory processor-readable storage medium of claim 45, wherein the stored processor-executable instructions are configured to cause the processor to perform operations further comprising:

concurrently activating a second triggered mode when the at least one additional event corresponds to a different trigger pattern associated with a third reflex.

57. The non-transitory processor-readable storage medium of claim 45, wherein the stored processor-executable instructions are configured to cause the processor to perform operations such that conducting the predetermined action associated with the first reflex in response to determining that the trigger weight associated with the first reflex exceeds the trigger weight threshold comprises:

generating a resulting event based on conducting the predetermined action; and

broadcasting an event report message that indicates the resulting event and that is capable of being used by another learning device.

58. The non-transitory processor-readable storage medium of claim 45, wherein the stored processor-executable instructions are configured to cause the processor to perform operations further comprising:

receiving an event report message from another learning device, wherein the received event report message indicates another event that was generated by the another learning device in response to the another learning device obtaining another event that corresponds to the first trigger pattern.

59. The non-transitory processor-readable storage medium of claim 45, wherein the stored processor-executable instructions are configured to cause the processor to perform operations such that activating the first triggered mode for the first reflex associated with the predetermined action when at least one of the stored events corresponds to the first trigger pattern associated with the first reflex comprises:

applying a pattern filter to a pattern based on the at least one of the stored events; and

activating the first triggered mode for the first reflex associated with the predetermined action when the at least one of the stored events corresponds to the first trigger pattern associated with the first reflex and when the pattern is not discarded based on the applied pattern filter.

60. The non-transitory processor-readable storage medium of claim 45, wherein the first reflex and the second reflex are only removable from the learning device based on a garbage collection policy.

61. The non-transitory processor-readable storage medium of claim 45, wherein the stored processor-executable instructions are configured to cause the processor to perform operations such that conducting the predetermined action associated with the first reflex in response to determining that the trigger weight associated with the first reflex exceeds the trigger weight threshold comprises generating a pattern of events that are used by a motor driver to drive an actuator.

62. The non-transitory processor-readable storage medium of claim 45, wherein one or more of the first correction pattern and the first reward pattern are order-dependent.

63. The non-transitory processor-readable storage medium of claim 45, wherein one or more of the first correction pattern and the first reward pattern are order-independent.

64. The non-transitory processor-readable storage medium of claim 45, wherein the stored processor-executable instructions are configured to cause the processor to perform operations such that obtaining the at least one additional event while in the first triggered mode comprises obtaining the at least one additional event based on one or more periodic timing mechanisms.

65. The non-transitory processor-readable storage medium of claim 64, wherein the stored processor-executable instructions are configured to cause the processor to perform operations further comprising polling one or more sensors based on the one or more periodic timing mechanisms.

66. The non-transitory processor-readable storage medium of claim 64, wherein the one or more periodic timing mechanisms are shared by multiple learning devices.

67. A system, comprising:

a first learning device within a plurality of learning devices, comprising;

a first processor; and

a first memory coupled to the first processor,

wherein the first processor is configured with processor-executable instructions for performing operations comprising;

storing, in the first memory, events obtained while in a monitoring mode;

activating a first triggered mode for a first reflex when at least one of the stored events corresponds to a first trigger pattern associated with the first reflex, wherein the first reflex indicates a correlation between at least the first trigger pattern and a predetermined action;

determining whether the first reflex has a trigger weight exceeding a trigger weight threshold;

conducting the predetermined action associated with the first reflex in response to determining that the trigger weight associated with the first reflex exceeds the trigger weight threshold;

obtaining at least one additional event while in the first triggered mode;

determining whether the at least one additional event corresponds to a first correction pattern or a first reward pattern;

adjusting the trigger weight of the first reflex when the at least one additional event corresponds to at least one of the first correction pattern and the first reward pattern associated with the first reflex and occurring in response to conducting the predetermined action, wherein the adjustment to the trigger weight of the first reflex is different in response to determining that the at least one additional event corresponds to the first correction pattern, than the adjustment to the trigger weight in response to determining that the at least one additional event corresponds to the first reward pattern; and

creating a second reflex with a second trigger pattern, a second correction pattern, and a second reward pattern when the at least one additional event does not correspond to at least one of the first trigger pattern, the first correction pattern, and the first reward pattern associated with the first reflex; and

a second learning device within the plurality of learning devices, comprising;

a second processor; and

a second signal transmitter coupled to the second processor, and wherein the second processor is configured with processor-executable instructions for performing operations comprising transmitting, via the second signal transmitter, a first event report message, and

wherein the first learning device further comprises;

a first signal receiver coupled to the first processor, and

wherein the first processor is configured with processor-executable instructions for performing operations further comprising receiving, via the first signal receiver, the first event report message transmitted by the second learning device.

68. The system of claim 67, wherein the first processor is configured with processor-executable instructions for performing operations such that adjusting the trigger weight of the first reflex further comprises:

increasing the trigger weight of the first reflex when the at least one additional event corresponds to the first reward pattern of the first reflex; and

decreasing the trigger weight of the first reflex when the at least one additional event corresponds to the first correction pattern of the first reflex.

69. The system of claim 68, wherein the first processor is configured with processor-executable instructions for performing operations further comprising:

entering a critical learning period with the first reflex during which a high gain set is applied to the trigger weight.

70. The system of claim 68, wherein the first processor is configured with processor-executable instructions for performing operations further comprising:

entering a steady state learning period with the first reflex during which a low gain set is applied to the trigger weight.

71. The system of claim 67, wherein the first processor is configured with processor-executable instructions for performing operations such that obtaining the at least one additional event while in the first triggered mode further comprises:

retrieving prior events from the first memory.

72. The system of claim 67, wherein the first processor is configured with processor-executable instructions for performing operations such that storing events obtained while in the monitoring mode comprises storing, in the first memory, at least one event for a window of time such that the at least one event and associated patterns are removed from the first memory after the window of time.

73. The system of claim 67, wherein at least one of the stored events is obtained based on occurrence data that is generated locally.

74. The system of claim 67, wherein at least one of the stored events is obtained based on occurrence data that is generated remotely.

75. The system of claim 74, wherein the at least one of the stored events that is obtained based on the occurrence data that is generated remotely corresponds to at least one of the first reward pattern and the first correction pattern.

76. The system of claim 67, further comprising:

one or more non-learning devices, and

wherein at least one of the stored events is obtained by the first learning device based on occurrence data that is received from the one or more non-learning devices via a wired or wireless communication link.

77. The system of claim 67, wherein the first processor is configured with processor-executable instructions for performing operations further comprising:

concurrently activating a second triggered mode when the at least one additional event corresponds to a different trigger pattern associated with a third reflex.

78. The system of claim 67, wherein the first learning device further comprises a first signal transmitter coupled to the first processor, and

wherein the first processor is configured with processor-executable instructions for performing operations such that conducting the predetermined action associated with the first reflex in response to determining that the trigger weight associated with the first reflex exceeds the trigger weight threshold comprises:

generating a resulting event based on conducting the predetermined action; and

broadcasting, via the first signal transmitter, a second event report message that indicates the resulting event and that is capable of being used by another learning device.

79. The system of claim 67, wherein the second processor is configured with processor-executable instructions for performing operations further comprising:

obtaining a first event that corresponds to the first trigger pattern; and

generating a second event in response to obtaining the first event that corresponds to the first trigger pattern, and

wherein the first event report message indicates the second event.

80. The system of claim 67, wherein the first processor is configured with processor-executable instructions for performing operations such that activating the first triggered mode for the first reflex associated with the predetermined action when at least one of the stored events corresponds to the first trigger pattern associated with the first reflex comprises:

applying a pattern filter to a pattern based on the at least one of the stored events; and

activating the first triggered mode for the first reflex associated with the predetermined action when the at least one of the stored events corresponds to the first trigger pattern associated with the first reflex and when the pattern is not discarded based on the applied pattern filter.

81. The system of claim 67, wherein the first reflex and the second reflex are only removable from the first memory of the first learning device based on a garbage collection policy.

82. The system of claim 67, wherein the first processor is configured with processor-executable instructions for performing operations such that conducting the predetermined action associated with the first reflex in response to determining that the trigger weight associated with the first reflex exceeds the trigger weight threshold comprises generating a pattern of events that are used by a motor driver to drive an actuator.

83. The system of claim 67, wherein one or more of the first correction pattern and the first reward pattern are order-dependent.

84. The system of claim 67, wherein one or more of the first correction pattern and the first reward pattern are order-independent.

85. The system of claim 67, wherein the first processor is configured with processor-executable instructions for performing operations such that obtaining the at least one additional event while in the first triggered mode comprises obtaining the at least one additional event based on one or more periodic timing mechanisms.

86. The system of claim 85, wherein the first processor is configured with processor-executable instructions for performing operations further comprising polling one or more sensors based on the one or more periodic timing mechanisms.

87. The system of claim 85, wherein the one or more periodic timing mechanisms are shared by the second learning device.

88. A method for decentralized training of a first learning device, comprising:

receiving one or more events at the first learning device, the first learning device having one or more associated reflexes, each reflex having at least one associated trigger pattern and at least one associated action;

determining whether the received one or more events match a first trigger pattern associated with a first reflex associated with the first learning device, wherein the first reflex indicates a correlation between at least the first trigger pattern and a first action;

determining whether the first reflex has a trigger weight exceeding a trigger weight threshold;

conducting the first action associated with the first reflex in response to determining that the trigger weight associated with the first reflex exceeds the trigger weight threshold;

obtaining at least one additional event;

determining whether the at least one additional event corresponds to a first correction pattern or a first reward pattern;

adjusting the trigger weight of the first reflex when the at least one additional event corresponds to at least one of the first correction pattern and the first reward pattern associated with the first reflex and occurring in response to conducting the first action, wherein the adjustment to the trigger weight of the first reflex is different in response to determining that the at least one additional event corresponds to the first correction pattern, than the adjustment to the trigger weight in response to determining that the at least one additional event corresponds to the first reward pattern; and

creating a second reflex associated with the first learning device with a second trigger pattern and a second action in response to determining that the received one or more events do not match any pattern associated with the first learning device.