Here's how world’s first Photonic Neural Network could lead to superfast computing

Science Image Credit

In the world of technology, neural networks are have been used to create machines that displays a huge range of skills that were earlier showcased only by human beings. For instance: identifying objects, recognising faces, translation from one language to another. Therefore, the focus has shifted in creating circuits that operate more like neurons, so-called neuromorphic chips, as it store a lot of hope for improvement in the arena of artificial intelligence.

And now, Alexander Tait along with his other team mates at Princeton University in New Jersey have built the world’s first integrated silicon photonic neuromorphic chip that is capable of computing at ultrafast speeds.

According to MIT Technology Review, computer science has been deriving its hope from optical computing because photons have more bandwidth than electrons. This can help in processing more data at a much faster pace. But computing areas like analog signal processing derives its juice from ultrafast data processing that only photonic chips can provide.

“Photonic neural networks leveraging silicon photonic platforms could access new regimes of ultrafast information processing for radio, control, and scientific computing,” said Tait along with his team.

The main target is to create an optical device, wherein each node has the same response characteristics as a neuron. Then, these nodes take the form of tiny circular waveguides carved into a silicon substrate in which light can circulate. Once released, this light modulates the output of a laser working at threshold, a regime in which small changes in the incoming light have a dramatic impact on the laser’s output.

It is worth noting that each node in the system works with wave division multiplexing technique. The light from all the nodes can be summed by total power detection before being fed into the laser. And the laser output is fed back into the nodes to develop a feedback circuit with a non-linear character.

Then, the researchers measured the output and revealed that it is mathematically equivalent to a device known as a continuous-time recurrent neural network. “This result suggests that programming tools for CTRNNs could be applied to larger silicon photonic neural networks,” said the scientists.

They further showcased how this can be done using a network consisting of 49 photonic nodes. This photonic neural network can be used to solve the mathematical problem of emulating a certain kind of differential equation and compare it to an ordinary central processing unit. The results show just how fast photonic neural nets can be. “The effective hardware acceleration factor of the photonic neural network is estimated to be 1,960 × in this task,” said Tait and team.

This research paves way for an entirely new industry that could bring optical computing into the mainstream for the first time. “Silicon photonic neural networks could represent first forays into a broader class of silicon photonic systems for scalable information processing,” said the researchers.