American researchers have created a living computer. It is the size of a palm and can recognize certain patterns.
Our title image is a symbolic image.
Researchers at the University of Illinois have used tens of thousands of living brain cells from mice to build a computer that can recognize light and electrical patterns. The team presented its findings at the American Institute of Physics (via newscientist.com).
The computer showcased is about the size of a palm. Mice have been playing a significant role in technology for a while now:
A computer made from mouse cells is expected to be more energy-efficient in the future
How does the system work? The scientists grew the computer from 80,000 reprogrammed mouse stem cells and placed these cells between fiber optics on a grid of electrodes.
The cells were kept alive during the experiment in an incubator. The team trained the computer by repeatedly illuminating ten different patterns of electrical impulses for one hour, then recording and analyzing the signals sent by the neurons. In between, the system needed a 30-minute break.
How effective is the mouse computer? In the first tests, the system still struggled. The accuracy of the tests is measured by the F-score, which ranges from 0 to 1, with 1 being the best value. In the first test, the score was 0.6. Only with a mixture of chemicals and electrical impulses could the accuracy be increased to 0.98.
Why do something like this? The idea behind it is that the division of labor between the biological cells and electrical components is meant to reduce the time and energy required to train conventional neural networks.
Therefore, the researchers hope to develop more efficient devices using the cells. However, they are still at the very beginning of the research. The team explains that the living computer made from mouse cells cannot yet keep up with modern neural networks.
Non-researchers might think of dystopian films while watching, such as A.I. – Artificial Intelligence by Steven Spielberg. We are probably still far from such an android.
However, the aim is to get closer to the model of neural networks. Because all artificial neural networks have a biological template.
