Model Verification Device Using Search Space Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing model verification techniques face inefficiencies in constrained search spaces due to excessive computational overhead when applying constraints, such as in hybrid systems with physical and digital components, where typical methods like falsification and stochastic hill climbing sample areas that do not satisfy constraints.
Innovation Solution
A model verification device that transforms a constrained search space into a less constrained space suitable for search algorithms like hill climbing, using proportional transformations to efficiently find counterexamples that violate specifications by sampling in an unconstrained space and then mapping back to the constrained space.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If constraint embedding (CE) method or lexicographic (LM) method is used to apply constraints to the search space, then the model verification can proceed with constraints, but computational overhead increases due to sampling in areas that do not satisfy the constraint
Solution Approach 1:
The patent inverts the traditional approach by first sampling from the unconstrained search space and then transforming the samples to satisfy constraints. Instead of constraining the search space before sampling (which causes wasted computations), the method samples freely and applies constraint satisfaction as a post-processing transformation step, thereby eliminating computational overhead while maintaining constraint reliability
Solution Approach 2:
The patent introduces a transformation function as an intermediary between the unconstrained search space and the constrained search space. This intermediary transforms samples that violate constraints into valid samples that satisfy constraints, allowing the verification process to operate in the simpler unconstrained space while still producing results that satisfy the original constraints
Data Source
AI summary
A model verification device includes a memory, and a processor coupled to the memory and configured to extract a sample from a search space, transform the extracted sample into an input on a constrained search space to which a constraint with respect to a model is applied, according to a predetermined transform rule; and determine whether an output of the model for the input satisfies a specification, and determine the input as a counterexample when the output does not satisfy the specification.