Neural Network Test Suite Generation for Faster System Verification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Test engineers face challenges in developing optimal test cases for computing systems due to resource-intensive manual efforts and potential biases in judgment, leading to inadequate verification of system functionality.

Innovation Solution

A method involving reinforcement learning through genetic algorithms to combine neural networks of different classes, mutating and breeding them to generate an optimal sequence of test cases for computing infrastructure testing, using a system with neural network units, fitness units, and breeding units to evaluate and improve neural network performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual test case development is performed by test engineers, then test cases can be created with human judgment and domain knowledge, but the process becomes resource-intensive and time-consuming

Engineering Contradiction:

Improveaccuracy of system verificationVSAvoidtime for developing test cases

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system enables self-service automated test case generation through neural networks that autonomously analyze computing system specifications and generate optimal test cases without requiring manual intervention from test engineers, thereby reducing time loss while maintaining verification accuracy

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical manual process of test case development with an automated neural network-based system that uses machine learning algorithms to generate test cases, substituting human manual effort with an intelligent automated system that operates faster and more efficiently

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Reliability

If manual test case development is performed by test engineers, then test cases can be customized according to specifications, but human judgment biases lead to inadequate verification

Engineering Contradiction:

Improveadequacy of system verificationVSAvoidcomplexity of test suite generation system

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent replaces human judgment with an automated neural network system that objectively analyzes specifications and generates test cases free from human biases, improving verification reliability by eliminating subjective judgment errors while using sophisticated machine learning models to handle the complexity of test generation

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The neural network acts as an intermediary between system specifications and test cases, translating requirements into optimized test sequences without human intervention, thereby eliminating bias while managing the complexity through intelligent algorithmic processing

Inventive Principle:

Principle #24Intermediary (Mediator)

3Productivity

If traditional test suite generation methods are used, then the process is simple and straightforward, but resource consumption is high

Engineering Contradiction:

Improveefficiency of test suite generationVSAvoidresource consumption for testing

Core Design Contradiction:

ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent implements dynamic test case selection where neural networks adaptively determine the optimal sequence and subset of test cases based on system characteristics and requirements, improving productivity by generating only necessary tests rather than executing exhaustive test suites, thereby reducing resource consumption

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes parameters such as test case selection criteria and sequencing based on neural network analysis of system specifications, optimizing the test suite to achieve maximum verification coverage with minimum resource consumption by dynamically adjusting test parameters rather than using fixed traditional approaches

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS12488234B2Reinforcement learning for testing suite generation

Publication Date: 2025.12.02 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

Aspects of the invention include mutating each neural network of a portion of a first array of neural networks, wherein each neural network of the first array of neural networks is configured to select a respective sequence of test cases for testing a computing infrastructure. Causing each neural network of a second array of neural networks to select a respective sequence of test cases for testing the computing infrastructure. Generating a child neural network by performing a crossover operation between a mutated neural network of the portion of the first array and a neural network of the second array of neural networks, the child neural network generating a new sequence of test cases for testing the computing infrastructure.