Researchers at Vienna University of Technology (TU Wien) have developed an image sensor that can be trained to recognize (in a matter of nanoseconds) certain objects.
The image sensor is made from a 2D material: tungsten diselenide, which is composed of only three atomic layers. The individual photodetectors, the “pixels” of the camera system, are all connected to a small number of output elements that provide the result of object recognition.
The chip represents an artificial neural network capable of learning. The data does not have to be read out and processed by a computer, but rather the chip itself provides information about what it is currently seeing, within nanoseconds. The system is designed to work like neurons within the brain: when one cell is active it can influence the activity of neighboring nerve cells. Artificial learning on the computer works according to the same principle. A network of neurons is simulated digitally, and the strength with which one node of this network influences the other is changed until the network shows the desired behavior.
“Typically the image data is read out pixel by pixel and then processed on the computer,” professor Thomas Mueller said. “We, on the other hand, integrate the neural network with its artificial intelligence directly into the hardware of the image sensor. This makes object recognition many orders of magnitude faster.”
Each individual detector element is able to be adjusted, which allows control of the way in which detected signals affect the output signal, said Lukas Mennel, first author of the study.
“All we have to do is simply adjust a local electric field directly at the photodetector,” he said.
This adaptation is done externally, with the help of a computer program. One can, for example, use the sensor to record different letters and change the sensitivities of the individual pixels step by step until a certain letter always leads exactly to a corresponding output signal. This is how the neural network in the chip is configured — making some connections in the network stronger and others weaker.
Once the learning process is complete, the computer is no longer needed. The neural network is then able to work alone. If a certain letter is presented to the sensor, it generates the trained output signal within 50 nanoseconds — for example, a numerical code representing the letter that the chip has just recognized.
“Our test chip is still small at the moment, but you can easily scale up the technology depending on the task you want to solve,” Mueller said. “In principle, the chip could also be trained to distinguish apples from bananas, but we see its use more in scientific experiments or other specialized applications.”
The technology can be usefully applied wherever extremely high speed is required “From fracture mechanics to particle detection — in many research areas, short events are investigated,” Muelle said. “Often it is not necessary to keep all the data about this event, but rather to answer a very specific question: Does a crack propagate from left to right? Which of several possible particles has just passed by? This is exactly what our technology is good for.”
The research was published in Nature (www.doi.org/10.1038/s41586-020-2038-x).