Neural networks consisting of DNA

Recent years have seen an increasing amount of work (some of which is also covered in this book) on implementing machine learning methods in physical systems, and the concept of intelligent matter [18] is closely related to this idea. While many approaches of this type employ electronic, magnetic, or photonic systems, it is in principle a relatively natural idea to use soft and biological matter as a basis for physical neural networks. After all, artificial neural networks are inspired by the brain, and the brain is a soft matter system. DNA, the carrier of genetic information, naturally suggests itself for such approaches. It is a soft matter system that has evolved specifically for the purpose1 of storing and processing information.