Evolutionary technique with n-pool evolution

• US 9,304,895 B1
• Filed: 07/18/2013
• Issued: 04/05/2016
• Est. Priority Date: 07/15/2011
• Status: Active Grant

• Current Assignee: Cognizant Technology Solutions U.S. Corporation (Cognizant Technology Solutions Corp.)
• Original Assignee: Sentient Technologies (Barbados) Limited
• Inventors: Shahrzad, Hormoz, Hodjat, Babak, Fink, Daniel Edward, Kamali, Kaivan
• Primary Examiner(s): Le, Thu-Nguyet
• Application Number:
  US13/945,630
• Time in Patent Office: 992 Days
• Field of Search: 717/142
• US Class Current: 1/1
• CPC Class Codes: G06F 11/3668 Software testing software testing in telephone exchanges H04M3/242, testing of hardware G06F11/22
• CPC Class Codes: G06F 11/3692 for test results analysis
• CPC Class Codes: G06F 16/2455 Query execution
• CPC Class Codes: G06F 16/26 Visual data mining; Browsing structured data

Abstract

Roughly described, a training database contains N segments of data samples. Candidate individuals identify a testing experience level, a fitness estimate, a rule set, and a testing set TS_i of the data samples on which it is tested. The testing sets have fewer than all of the data segments and they are not all the same. Testing involves testing on only the individual'"'"'s assigned set of data segments, updating the fitness estimates and testing experience levels, and discarding candidates through competition. If an individual reaches a predetermined maturity level of testing experience, then validating involves further testing it on samples of the testing data from a testing data segment other than those in the individual'"'"'s testing set TS_i. Those individuals that satisfy validation criteria are considered for deployment.

Claims

1. A computer-implemented method, for use with a testing database containing a plurality of samples of testing data, the samples being distributed among N>

1 segments of the testing data, each of the segments including at least one of the samples and at least one of the segments including more than one of the samples,

for use further with a memory storing a candidate database having a pool of candidate individuals, each i'"'"'th one of the candidate individuals identifying a testing experience level, a fitness estimate, a rule set for applying to samples of data in the testing data, and a respective testing set TS_i of the testing data segments, the method comprising;

for each i'"'"'th one of the candidate individuals;

assigning to the i'"'"'th individual the respective testing set TS_i of the testing data segments,

a computer system testing the i'"'"'th individual on samples of the testing data from the i'"'"'th individual'"'"'s testing set TS_i of testing data segments,

updating the fitness estimate associated with the i'"'"'th individual in dependence upon results of the testing, and

updating the testing experience level associated with the i'"'"'th individual in dependence upon the number of samples on which the i'"'"'th individual is tested;

selecting individuals for discarding from the candidate pool in dependence upon their updated fitness estimates;

a computer system validating each individual whose testing experience level has reached a predetermined maturity level without being selected for discarding, including further testing each such individual on samples of the testing data from a testing data segment other than those in the individual'"'"'s testing set TS_i; and

providing for deployment selected ones of the individuals that satisfy predetermined deployment criteria after validation,

wherein each of the testing sets TS_i has fewer than all of the N testing data segments and at least one of the testing sets TS_i has different testing data segments than another of the initial testing sets TS_i.

2. The method of claim 1, wherein each of the segments of the testing data includes more than one of the samples.

3. The method of claim 1, wherein each of the testing sets TS_i contains exactly one of the testing segments.

4. The method of claim 1, wherein each of the samples in the testing data in each of the segments comprises a plurality of data entries, and wherein testing each individual on samples of the testing data comprises, for each particular sample:

classifying each of the entries in the particular sample in dependence upon the rules in the individual; and

scoring the individual'"'"'s performance on the particular sample in dependence upon all the classifications performed by the individual on data entries in the particular sample.

5. The method of claim 4, wherein each of the samples covers a predetermined time period, and each of the data entries comprises a price of a financial asset at a respective time within the time period covered by the sample,

and wherein the classifying step comprises recommending an action either to take a position, divest all or part of a position, or leave a position unchanged, with respect to the financial asset, in dependence upon the price of the financial asset as indicated in each entry.

6. The method of claim 1, wherein N=8.

7. The method of claim 1, wherein each s'"'"'th one of the segments has associated therewith a respective segment maturity experience level ML_s.

8. The method of claim 7, wherein all of the ML_s are equal.

9. The method of claim 1, further comprising forming new individuals in the candidate pool in dependence upon a respective set of at least one parent individual from the candidate pool.

10. The method of claim 1, for use in a server infrastructure with respect to a collection of at least one client device,

further comprising assigning each of the client devices to one of the testing sets TS_i, not all of the client devices being assigned to any single one of the testing sets TS_i,

wherein testing each individual from the candidate pool on samples of the testing data from the individual'"'"'s testing set TS_i of testing data segments, comprises sending the individual to a client device assigned to the testing subset TS_i.

11. The method of claim 10, wherein further testing each such individual on samples of the testing data from a testing data segment other than those in the individual'"'"'s testing set TS_i, comprises sending the individual to a client device assigned to a testing subset that does not overlap with any testing subset on which the individual was tested.

12. A computer-implemented system, for use with a testing database containing a plurality of samples of testing data, the samples being distributed among N>

1 segments of the testing data, at least one of the segments including more than one of the samples, comprising;

a memory storing a candidate database having a pool of candidate individuals, each i'"'"'th one of the candidate individuals identifying a testing experience level, a fitness estimate, a rule set for applying to samples of data in the testing data, and a respective testing set TS_i of the testing data segments; and

a training system which;

for each i'"'"'th one of the candidate individuals;

assigns to the i'"'"'th individual the respective testing set TS_i of the testing data segments,

tests the i'"'"'th individual on samples of the testing data from the i'"'"'th individual'"'"'s testing set TS_i of testing data segments,

updates the fitness estimate associated with the i'"'"'th individual in dependence upon results of the testing, and

updates the testing experience level associated with the i'"'"'th individual in dependence upon the number of samples on which the i'"'"'th individual is tested;

selects individuals for discarding from the candidate pool in dependence upon their updated fitness estimates;

validates each individual whose testing experience level has reached a predetermined maturity level without being selected for discarding, including further testing each such individual on samples of the testing data from a testing data segment other than those in the individual'"'"'s testing set TS_i; and

providing for deployment selected ones of the individuals that satisfy predetermined deployment criteria after validation,

wherein each of the testing sets TS_i has fewer than all of the N testing data segments and at least one of the testing sets TS_i has different testing data segments than another of the initial testing sets TS_i.

13. The computer-implemented system of claim 12, wherein each of the segments of the testing data includes more than one of the samples.

14. The computer-implemented system of claim 12, wherein each of the testing sets TS, contains exactly one of the testing data segments.

15. The computer-implemented system of claim 12, wherein each of the samples in the testing data in each of the segments comprises a plurality of data entries, and wherein in testing each individual on samples of the testing data comprises, for each particular sample the training system:

classifies each of the entries in the particular sample in dependence upon the rules in the individual; and

scores the individual'"'"'s performance on the particular sample in dependence upon all the classifications performed by the individual on data entries in the particular sample.

16. The computer-implemented system of claim 15, wherein each of the samples covers a predetermined time period, and each of the data entries comprises a price of a financial asset at a respective time within the time period covered by the sample,

and wherein classifying comprises recommending an action either to take a position, divest all or part of a position, or leave a position unchanged, with respect to the financial asset, in dependence upon the price of the financial asset as indicated in each entry.

17. The computer-implemented system of claim 12, wherein each s′

th one of the segments has associated therewith a respective segment maturity experience level ML_s.

18. The computer-implemented system of claim 17, wherein all of the ML_s, are equal.

19. The computer-implemented system of claim 12, wherein the training system further forms new individuals in the candidate pool in dependence upon a respective set of at least one parent individual from the candidate pool.

20. The computer-implemented system of claim 12, for use in a server infrastructure with respect to a collection of at least one client device,

wherein the training system further assigns each of the client devices to one of the testing sets TS_i, not all of the client devices being assigned to any single one of the testing sets TS_i,

wherein in testing each individual from the candidate pool on samples of the testing data from the individual'"'"'s testing set TS_i of testing data segments, the server infrastructure sends the individual to a client device assigned to the testing subset TS_i.

21. The computer-implemented system of claim 20, wherein in further testing each such individual on samples of the testing data from a testing data segment other than those in the individual'"'"'s testing set TS_i, the server infrastructure sends the individual to a client device assigned to a testing subset that does not overlap with any testing subset on which the individual was tested.