Evolvability and Redundancy in Shared Grammar Evolution

摘要: Shared grammar evolution (SGE) is a novel scheme for representing and evolving population of variable- length programs as shared set grammatical productions. Productions that fail to contribute selected solutions can be retained several generations beyond their last use. The ensuing redundancy its effects are assessed in this paper on two circuit design tasks associated with random number generation: finding recurrent maximum period, reproducing De Bruijn counter from seed/output pairs. In both instances, increasing leads significantly higher success rates, outperforming comparable increases size. results support previous studies have shown representational beneficial evolutionary search. However, promotes an increase further by encouraging the creation large offspring, evaluation which computationally costly. This observation should generalize any unconstrained representation therefore represents notable drawback evolution.