Tsinghua University scientists have developed efficient all-analog electronic photo chips

Scientists at China’s Tsinghua University have recently developed an all-analog electronic photo chip. According to a report published by the Xinhua News Agency, the chip will provide superior performance in computer vision tasks, faster processing and greater energy efficiency compared to current chips.

The research findings, which offer an alternative to conventional technologies relying on analog-to-digital conversion, are published in the journal Nature.

Analog and digital signals are two types of signals that carry information. Analog signals change continuously, like light rays forming an image, while digital signals are discontinuous, like binary numbers, Xinhua reported.

In vision-based computing tasks such as image recognition and object recognition, environmental signals are analog and must be converted to digital signals for processing by artificial intelligence neural networks, systems trained to recognize patterns and relationships in a dataset. However, analog-to-digital conversion is time-consuming and energy-consuming, limiting the speed and efficiency of neural network performance. Photonic computing, which uses analog light signals, is one of the most promising approaches to address this problem, the report said.

In the new study, the researchers designed an integrated photoelectronic processor to take advantage of light, in the form of photons, and electrons, as found in electrical currents, in an all-analog manner. The result is called the “all-analog chip combining electronics and light” or ACCEL.

“We maximized the advantages of light and electricity under all-analog signals, avoiding the disadvantages of analog-to-digital conversion and breaking the bottleneck of power consumption and speed,” said Fang Lu, a researcher from the Tsinghua team.

Experiments have shown that ACCEL is able to identify and classify objects with a degree of accuracy comparable to digital neural networks. In addition, it classifies high-resolution images from various scenes of everyday life more than 3,000 times faster and with 4,000,000 times less energy consumption than a first-class GPU (graphics processing unit).

ACCEL has potential for future applications in areas such as unmanned systems, industrial inspections and large-scale artificial intelligence models, Dai Qinghai, director of the School of Information Science and Technology at Tsinghua University, said in the report.

However, for now, the team has only developed prototypes based on the principles of opto-electronic integration for specific computational functions. This indicates that the chip is still in the prototype stage and is not yet ready for practical use. More research and development is needed to create intelligent visual computing chips with general-purpose capabilities for wide-ranging real-world applications, Dai said.

An industry analyst, speaking on condition of anonymity, told the Global Times on Sunday: “Research progress may indicate a possible reduction in reliance on photolithography machines.”