UPenn engineers create chip that can train AI with light waves

UPenn engineers create chip that can train AI with light waves

(NewsNation) — Engineers at the University of Pennsylvania developed a new chip that uses light waves to perform the math needed to train artificial intelligence models.

This chip could potentially “radically accelerate the processing speed of computers while also reducing their energy consumption,” the university said in a news release.

Not only that, but it could also improve people’s privacy online. With the chip able to make many computations simultaneously, sensitive information wouldn’t need to be stored in a computer’s “working memory,” making them “virtually unhackable,” the university said.

Chips used today are limited, according to the university, as they are essentially based on the same principles as chips used in the earliest days of the 1960s’ “computing revolution.”

But this new silicon-photonic chip is the first to bring together Professor Nader Engheta’s research in manipulating materials at the nanoscale to perform mathematical computations using light.

Light is the fastest possible means of communication, according to the university.

Engheta’s group worked with UPenn professor in Electrical and Systems Engineering Firooz Aflatouni, whose research has pioneered nanoscale silicon devices.

In a paper in Nature Photonics, the researchers say their goal was to develop a platform for performing “vector-matrix multiplication.”

Vector-matrix multiplication is a mathematical operation that is crucial to the development of the neural networks that power AI tools.

To do this, the engineers made the silicon in the chip thinner by about 150 nanometers, but only in specific regions.

This height variation caused the light to scatter in specific patterns, giving engineers the ability to control it.

Source link

Denial of responsibility! NewsConcerns is an automatic aggregator of the all world’s media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, all materials to their authors. If you are the owner of the content and do not want us to publish your materials, please contact us by email – [email protected]. The content will be deleted within 24 hours.

Leave a Comment