Emergent Induction of Deterministic Context-Free L-system Grammar

L-system is a bio-inspired computational model to capture growth process of plants. This paper proposes a new noise-tolerant grammatical induction LGIC2 for deterministic context-free L-systems. LGIC2 induces L-system grammars from a transmuted string mY, employing an emergent approach in order to enforce its noise tolerance. In the method, frequently appearing substrings are extracted from mY to form grammar candidates. A grammar candidate is used to generate a string Z; however, the number of grammar candidates gets huge, meaning enormous computational cost. Thus, how to prune grammar candidates is vital here. We introduce a couple of techniques such as pruning by frequency, pruning by goodness of fit, and pruning by contractive embedding. Finally, several candidates having the strongest similarities between mY and Z are selected as the final solutions. Our experiments using insertion-type transmutation showed that LGIC2 worked very nicely, much better than an enumerative method LGIC1.