Automated case generation using a genetic algorithm
|Title||Automated case generation using a genetic algorithm|
|Publication Type||Conference Paper|
|Year of Publication||2017|
|Authors||Borck H, Boddy M|
|Conference Name||Proceedings of the Genetic and Evolutionary Computation Conference Companion|
Case-Based Reasoning is a learn-by-experience approach in which past problem solving instances, called cases, are used to solve novel input problems. Authoring these cases is often a manual process requiring the assistance of a domain expert. To alleviate this problem, we have developed CBGen, a Genetic Algorithm-based approach for case creation. CBGen uses individuals that represent the initial parameters of a low fidelity simulator and a target task that must be simulated. The encoding of each individual is used to run the simulations and store how the task was completed. Individuals are then evaluated based on the expected benefit of the case they generated being retained by the system. A preliminary proof-of-concept in an augmented reality domain validate the feasibility of using CBGen to automatically create a case base.