48 results about "Mutation testing" patented technology

The invention discloses a software test case evolution generation method based on search domain reduction, and aims to dynamically determine the correlation between an input variable and a variant, remove an irrelevant input variable equivalent to reduction of a search domain, and efficiently generate a variation test case by adopting a genetic algorithm. The method comprises the steps that firstly, a mutation test case generation optimization model is established, a fitness function is designed, and then the correlation between a variant and an input variable is dynamically determined based on fitness change; then, a mutation test case generation optimization model is improved, the decision variable is a related input variable, and the mutation test case generation optimization model based on the related input variable is established; and finally, for the new model, a test case is effectively generated by adopting a genetic algorithm in a search domain formed by related input variables.

The present disclosure discloses a method for judging a stubborn variant of a parallel program, which is characterized in that it includes: dividing the execution sequence of the sending nodes of the variant statement into equivalence classes according to the execution status of the variant statement; determining the factors affecting the stubbornness of the variant ;According to the factors that affect the recalcitrance of the variant, combined with the influence of the execution order of the sending node after the equivalence class division on the factors, calculate the recalcitrant composite index of the variant; set the threshold and compare it with the recalcitrant composite index to complete the recalcitrant mutation Judgment of variants; this disclosure combines the concept of parallel programs with the principle of mutation testing, and screens out stubborn variants of parallel programs based on the relationship between the execution order of sending nodes and factors affecting the stubbornness of variants, and focuses on testing the stubborn variants , thereby improving the test efficiency.

The invention provides a redundant variant identification method and identification device, capable of effectively reducing the time expenditure of a variant test execution process under the conditionof not reducing the effectiveness of a variant test. The redundant variant identification method comprises the following steps: dividing variants into different variant groups according to variationpositions; carrying out code preprocessing on the original program and the variant group, so that variants in the original program and the variant group can be selectively symbolized and executed; performing selection symbolic execution on the preprocessed original program and variant, and recording symbolic execution states of the original program and variant after execution; comparing the symbolic execution states of the original program and the variants to generate a state change sequence; and in each variant group, identifying the variants to which the repeatedly occurring state change sequences belong as redundant variants. The invention relates to the technical field of software.

The invention provides a program state-oriented fault classification method, a mutation testing method and a device. The method comprises the following steps of: constructing a fault node set and a state mutation point set for each sentence of a source program to be tested based on a mutation operator corresponding to the syntax of the source program to be tested; constructing a variation flow diagram of the source program to be tested based on the fault node set, the state variation point set, the triggering relationship and the propagation relationship between the state variations of each statement; and according to the direct backward dominance relationship of the variant flow diagram, constructing a fault class dominance tree to perform fault classification on the source program to betested. The invention classifies and sorts the fault programs according to the general fault test principle, and supports multiple types of mutation operators, has good scalability, thereby reducing the test times, optimizing the fault set, simplifying the fault analysis cost, and improving the mutation test efficiency.

Improved methods of assessing status of a solid tumor cancer in a subject involving detection of tumor-associated mutations in the subject's blood.

The invention relates to a method for predicting demand for variation test intensity in an evolutionary environment. Based on feature extraction and error detection numbers, this method constructs feature matrix and intensity vector for model training through data cleaning and feature selection for each version of the software; on this basis, the BP neural network method is used to build a prediction model, and the During construction, the weight of the model is continuously adjusted through the results of signal forward propagation and error backward propagation, so as to learn and generate a variation test intensity demand prediction model BP-Model; finally, input the feature vector of the current software version into BP-Model, and finally generate Mutation testing intensity requirements for the current software version. The purpose of the present invention is to solve the problem that the intensity of mutation testing of new version software is unknown at present, and then help testers formulate reasonable mutation testing requirements, and help testers build effective test case sets within a limited time.

The invention discloses an interpretable model-based cycle neural network backdoor attack detection method, which abstracts the RNN model and performs backdoor detection on the text in three steps: firstly, a machine learning algorithm is used to cluster the RNN hidden layer vectors, and construct an uncertain Finite automaton; secondly, according to the constructed uncertain finite automata, the state transition path of the text is obtained, so as to calculate the weight of each word in the text; finally, based on the idea of ​​mutation testing, the backdoor in the text is detected. Through the above method, it is possible to accurately explain the decision of RNN on the text and accurately detect the backdoor text.