AI Neural Network made of DNA can identify ‘molecular handwriting’


Artificial Intelligence is advancing day by day and here is another proof of it. Recently, a group of Scientists from the California Institute of Technology has made a new neural network out of DNA strands! Yes, you heard it right! Using the DNA Strands, Scientists have made up a network which will help them to perform pattern recognition.

The new AI will perfectly identify numbers which are encoded into the patterns using Machine Learning and Artifical Intelligence. It will make the biological machines work with Artificial Intelligence making it more powerful than ever before.

What Are Artificial Neural Networks?

Artificial Neural Networks are basically mathematical models. They are inspired by the human brain and can perform complex functions. Just like their own counterpart, Artificial Neural Networks are simpler but can do complex calculations. They are built around a network of neurons and hence are capable to process complex information easily.

The New AI Network Created

Artificial Intelligence is generally known as pattern recognition tools. They are used to study and identify the pattern present within the data provided. The scientists of California Institute have found a similar way to use the AI Network. It will use the biologically engineered strands of DNA to process complex data. They made the case ease by making the AI network recognize different distinct patterns. With the new AI neural network, the main goal is to design them with intelligent behaviors and AI made out of DNA.

In the whitepaper of the project, the team has mentioned that they have used the bio-engineered strands of DNA to act as bits. It will help them to represent the human handwriting in digital format. For such AI tools, the basic test is to use them on human handwriting. Since it is not structured and vary a lot, it will help in determining the actual capacity of the network. The AI Network must be taught on how to recognize these numbers and understand the variations. Then the network can use its fed in data or memory to output the result of the comparison.


Lulu Qian, Assistant Professor at California Insititute of Technology said that “We have created and designed biochemical circuits. It will function like a network of neurons which will classify molecular information. Substantially, our product is better at recognizing more complex data than any other present in the world”.

Molecular numbers are more complex than human handwriting. They are made up of nearly 20 different and unique DNA strands. It is made using 100 different molecules, each of which is used to represent an individual pixel of 10 * 10 pattern. They are mixed together and kept in a test tube. When the AI Neural Network was tested, it was able to recognize nearly nine categories each representing a possible handwriting of digits 0 to 9. But, there is more to it. Researchers are trying to develop the neural network to learn from its memory and add to it. Thus, it will become smarter on its own and identify more different patterns.


Please enter your comment!
Please enter your name here