ANN-BASED PROGRAM TEST METHOD AND TEST SYSTEM, AND APPLICATION

Title:

ANN-BASED PROGRAM TEST METHOD AND TEST SYSTEM, AND APPLICATION

Document Type and Number:

WIPO Patent Application WO/2021/115186

Kind Code:

A1

Abstract:

An ANN-based program test method, which tests a target program on which functions can be completed and codes can run normally, and relates to the field of software testing. Said method comprises: first constructing a test model by means of an artificial neural network, then inputting input values of a target program into the test model, and calculating test output values according to the test model; then, inputting the input values into the target program to obtain actual output values; according to deviation amounts between the test output values and the actual output values which correspond to the same input value, selecting input values corresponding to the top-ranked 50 to 150 deviation amounts; transmitting the selected input values to the target program for running; and comparing running results with actual functional requirements, and if one of the running results does not satisfy the actual functional requirements, determining that the target program has a defect. In cases where neither the specific code specification nor the actual requirements of the target program is known, the test method can be used to obtain an accurate test instance, reducing the resource consumption in the test process.

Inventors:

ZHANG HUACHEN (CN)

Application Number:

PCT/CN2020/133517

Publication Date:

June 17, 2021

Filing Date:

December 03, 2020

Export Citation:

Click for automatic bibliography generation Help

Assignee:

ZUNYI VOCATIONAL & TECHNICAL COLLEGE (CN)

International Classes:

G06N3/04; G06F11/36; G06N3/08

Foreign References:

CN111026664A	2020-04-17
CN105843743A	2016-08-10
CN108470000A	2018-08-31
CN108170594A	2018-06-15
CN110297764A	2019-10-01
CN103995775A	2014-08-20
US20190362222A1	2019-11-28

Other References: