System including inductive learning arrangement for adaptive management of behavior of complex entity

• US 5,034,898 A
• Filed: 05/02/1989
• Issued: 07/23/1991
• Est. Priority Date: 05/02/1989
• Status: Expired due to Term

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Digital Equipment Corporation (HP Inc.)
• Inventors: Tseng, Mitchell M., Lu, Stephen C., Teng, Henry S.
• Primary Examiner(s): MacDonald, Allen R.
• Application Number:
  US07/346,133
• Time in Patent Office: 812 Days
• Field of Search: 364/513
• US Class Current: 706/62
• CPC Class Codes: G06N 20/00 Machine learning

Abstract

A system and method for adaptively managing the behavior of, for example, a digital data processing system, using inductive learning techniques. The new system includes a data acquistion portion that acquires event data in response to selected events. An inductive engine generates profile knowledge in response to acquired event data during a learning phase, for storage in a profile knowledge store. A profile comparator, during an operational phase, subsequent to the learning phase, performs a verification operation in connection with comparing event data to said profile knowledge stored in said profile knowledge store. A control portion enables event data acquired by said data acquisition portion to be directed to the inductive engine during a learning phase and to the profile comparator during an operational phase to thereby permit the comparator to perform a verification operation in connection with said event data acquired during the operational phase.

Claims

1. A computer system comprising:

A. a data acquisition portion for acquiring event data in response to selected events;

B. an inductive engine for generating profile knowledge in response to event data received by said inductive engine;

C. a profile knowledge store for storing profile knowledge generated by said inductive engine;

D. a profile comparator for performing a verification operation in connection with comparing event data to said profile knowledge stored in said profile knowledge store; and

E. a control portion for enabling event data acquired by said data acquisition portion to be directed to said inductive engine during a learning phase and to said profile comparator during an operational phase to thereby permit the comparator to perform a verification operation in connection with said event data acquired during said operational phase.

2. A system as defined in claim 1 wherein said data acquisition portion divides acquired event data into a plurality of classes, said control portion enabling said inductive engine to separately generate profile knowledge in connection with event data in each class, and to enable said profile comparator to separately perform said verification operation in connection with event data and profile knowledge in each class.

3. A system as defined in claim 1 in which a result of a verification operation is a verification failure, said profile knowledge store including a verification knowledge store for storing verification knowledge and said system further comprising a verification portion that performs a verification confirmation operation in connection with event data and the verification knowledge contained in said verification knowledge store, said control portion enabling said verification portion in response to a verification failure to confirm the result.

4. A system as defined in claim 3 wherein said verification portion further generates profile knowledge for storage in said profile knowledge store for use during a subsequent verification operation.

5. A system as defined in claim 1 in which the profile knowledge generated by said inductive engine is in the form of one or more rules stored in said profile knowledge store, said profile comparator using the event data to identify applicable rules and process them.

6. A computer implemented method comprising the steps of:

A. acquiring, during a learning phase, training event data in response to selected events;

B. enabling, during said learning phase, an inductive engine to receive acquired training event data and generating in response thereto profile knowledge;

C. maintaining profile knowledge generated by said inductive engine for use during a later operational phase;

D. acquiring, during an operational phase, operational event data in response to selected events

E. using the profile knowledge and said operational event data to perform a verification operation in connection with said operational event data.

7. A method as defined in claim 6 wherein, during data acquisition during said learning phase, training event data is divided into a plurality of classes, said inductive engine separately generating profile knowledge in connection with training event data in each class, the operational event data acquired during said operational phase similarly being divided into a plurality of classes and the verification operation being performed in connection with operational event data and profile knowledge in each class.

8. A method as defined in claim 6 in which one result of a verification operation is a verification failure, said profile knowledge store including a verification knowledge store for storing verification knowledge, said method further comprising the step of a verification confirmation operation in connection with event data and the verification knowledge contained in said verification knowledge store in response to a verification failure to confirm the result.

9. A method as defined in claim 8 wherein said verification confirmation operation further includes the generation of profile knowledge for storage in said profile knowledge store for use during a subsequent verification operation.

10. A method as defined in claim 6 in which the profile knowledge generated by said inductive engine is in the form of one or more rules stored in said profile knowledge store and used during a verification operation.