Practical experimental validation of fractional-coding molecular/DNA ANNs

Establish experimental validation of the practical feasibility of artificial neural networks implemented via fractional coding with bimolecular chemical reaction networks and DNA strand displacement, moving beyond simulations to demonstrate in-vitro or in-vivo operation and performance for perceptrons and multi-layer ANNs.

Background

The paper presents a theoretical framework for implementing perceptrons and artificial neural networks using molecular computing and DNA via fractional coding. It introduces a molecular divider, exact sigmoid and tanh implementations, scaled inner products that accommodate arbitrary weights, and maps these chemical reaction networks to DNA strand displacement. The authors demonstrate functionality through simulations for perceptrons and an ANN classifier.

Despite these advances, the authors emphasize that their results are based on simulations and that practical use in laboratory or biological settings has not yet been demonstrated. They explicitly state that validation of practical utility remains outstanding and suggest that future work should focus on experimental demonstration of the proposed theoretical framework.