Coevolutionary Computation for Synthesis of Recognition Systems

摘要: This paper introduces a novel visual learning method that involves cooperative coevolution and linear genetic programming. Given exclusively training images, the evolutionary algorithm induces set of sophisticated feature extraction agents represented in procedural way. The proposed incorporates only general vision-related background knowledge does not require any task-specific information. describes algorithm, provides firm rationale for its design, proves competitiveness with human-designed recognition systems an extensive experimental evaluation, on demanding real-world task object synthetic aperture radar (SAR) imagery.