The team of East China Normal University proposed the systematic molecular implementation of convolutional neural network with DNA regulation circuit

Complex biomolecular circuits make cells with intelligent behavior , Can survive before the evolution of neural brain . since 1990 It was first shown in the mid-s DNA Since the calculation , Liquid phase synthesis DNA Circuits have been developed as computing hardware , To perform calculations similar to neural networks utilizing the collective characteristics of complex biochemical systems . However , Expand this based on DNA Neural networks to support more powerful computing are still challenging .

ad locum , The research team of East China Normal University proposed a systematic molecular implementation of convolutional neural network algorithm , The algorithm has a synthesis based on a simple door opening and closing architecture DNA Regulating circuit . The researchers based on DNA The weight sharing convolutional neural network can be realized at the same time 144 Bit input parallel product accumulation operation （Multiply Accumulate,MAC）, And identify as many as 8 Patterns of categories .

Besides , This system can be combined with other DNA Circuit connection , To build a layered network , Identify as many as... Through a two-step method 32 Patterns of categories ： Rough classification of languages , Then it is classified into specific handwritten symbols . The researchers also reduced the calculation time from a few hours to a few minutes by using a simple cyclic freeze-thaw method . be based on DNA Regulating circuit of , It is a step towards the realization of molecular computers with high computing power and the ability to classify complex and noisy information .

The research 「 Molecular convolutional neural networks with DNA regulatory circuits」 entitled , On 2022 year 7 month 4 Published on 《 Nature Machine Intelligence》.

Thesis link ：https://www.nature.com/articles/s42256-022-00502-7