Method and apparatus for generating test pattern in order to verify the lack of convolution operation and selecting optimizing test pattern from among the test patterns to improve the toughness tolerance and swinging property in extreme conditions

摘要

PROBLEM TO BE SOLVED: To provide a method for generating one or more test patterns to verify integrity of one or more parameters of a convolutional neural network (CNN) by using at least one test pattern so that an autonomous vehicle can be driven safely from hacker threats.;SOLUTION: A method includes steps of: generating test patterns by using a certain function so that saturation does not occur in a process to which convolution operations are applied after the test patterns are entered in at least one original CNN; generating each of evaluation scores of each of the test patterns by referring to each of the test patterns and one or more parameters of the original CNN; and selecting optimized test patterns from among the test patterns by referring to the respective evaluation scores.;SELECTED DRAWING: Figure 3;COPYRIGHT: (C)2020,JPO&INPIT