Neural networks could be implemented more quickly using new photonic technology
“Deep learning” computer systems, based on artificial neural networks that mimic the way the brain learns from an accumulation of examples, have become a hot topic in computer science. In addition to enabling technologies such as face- and voice-recognition software, these systems could scour vast amounts of medical data to find patterns that could be useful diagnostically, or scan chemical formulas for possible new pharmaceuticals.
But the computations these systems must carry out are highly complex and demanding, even for the most powerful computers.
Now, a team of researchers at MIT and elsewhere has developed a new approach to such computations, using light instead of electricity, which they say could vastly improve the speed and efficiency of certain deep learning computations. Their results appear today in the journal Nature Photonics in a paper by MIT postdoc Yichen Shen, graduate student Nicholas Harris, professors Marin Solja?i? and Dirk Englund, and eight others.
Solja?i? says that many researchers over the years have made claims about optics-based computers, but that “people dramatically over-promised, and it backfired.” While many proposed uses of such photonic computers turned out not to be practical, a light-based neural-network system developed by this team “may be applicable for deep-learning for some applications,” he says.
Traditional computer architectures are not very efficient when it comes to the kinds of calculations needed for certain important neural-network tasks. Such tasks typically involve repeated multiplications of matrices, which can be very computationally intensive in conventional CPU or GPU chips.
After years of research, the MIT team has come up with a way of performing these operations optically instead. “This chip, once you tune it, can carry out matrix multiplication with, in principle, zero energy, almost instantly,” Solja?i? says. “We’ve demonstrated the crucial building blocks but not yet the full system.”
By way of analogy, Solja?i? points out that even an ordinary eyeglass lens carries out a complex calculation (the so-called Fourier transform) on the light waves that pass through it. The way light beams carry out computations in the new photonic chips is far more general but has a similar underlying principle. The new approach uses multiple light beams directed in such a way that their waves interact with each other, producing interference patterns that convey the result of the intended operation. The resulting device is something the researchers call a programmable nanophotonic processor.
The result, Shen says, is that the optical chips using this architecture could, in principle, carry out calculations performed in typical artificial intelligence algorithms much faster and using less than one-thousandth as much energy per operation as conventional electronic chips. “The natural advantage of using light to do matrix multiplication plays a big part in the speed up and power savings, because dense matrix multiplications are the most power hungry and time consuming part in AI algorithms” he says.
The new programmable nanophotonic processor, which was developed in the Englund lab by Harris and collaborators, uses an array of waveguides that are interconnected in a way that can be modified as needed, programming that set of beams for a specific computation. “You can program in any matrix operation,” Harris says. The processor guides light through a series of coupled photonic waveguides. The team’s full proposal calls for interleaved layers of devices that apply an operation called a nonlinear activation function, in analogy with the operation of neurons in the brain.
To demonstrate the concept, the team set the programmable nanophotonic processor to implement a neural network that recognizes four basic vowel sounds. Even with this rudimentary system, they were able to achieve a 77 percent accuracy level, compared to about 90 percent for conventional systems. There are “no substantial obstacles” to scaling up the system for greater accuracy, Solja?i? says.
Englund adds that the programmable nanophotonic processor could have other applications as well, including signal processing for data transmission. “High-speed analog signal processing is something this could manage” faster than other approaches that first convert the signal to digital form, since light is an inherently analog medium. “This approach could do processing directly in the analog domain,” he says.
The team says it will still take a lot more effort and time to make this system useful; however, once the system is scaled up and fully functioning, it can find many user cases, such as data centers or security systems. The system could also be a boon for self-driving cars or drones, says Harris, or “whenever you need to do a lot of computation but you don’t have a lot of power or time.”
Learn more: New system allows optical “deep learning”
The Latest on: Neural networks
- FPGA-based neural network accelerator outperforms GPUs on October 5, 2018 at 9:52 am
Xilinx Developer Forum: Claimed to be the highest performance convolutional neural network (CNN) on an fpga, Omnitek’s CNN is available now. The deep learning processing unit (DPU) is future-proofed, ... […]
- Neuton: A new, disruptive neural network framework for AI applications on October 4, 2018 at 7:01 am
This does not happen often. Deep learning is the hottest technology today, with countless applications and deep investment from the usual suspects. To have something new released from someone who is n... […]
- Bell Integrator's Neuton Is A New Neural Network Framework For AI And Machine Learning on October 4, 2018 at 6:09 am
The biggest tech subject of the day is one that fascinates, bemuses, befuddles and confuses us. What is artificial intelligence or AI as it is better known? What is machine-to-machine learning? What d... […]
- Baidu is testing neural networks that can match job seekers to jobs on October 2, 2018 at 11:15 am
The Chinese company’s technology identifies candidates whose skills meet employers’ needs. The news: In a recent paper, Baidu researchers showed how their neural net can work out from their résumés wh... […]
- Neural Network Optimization with Sparse Computing and Facebook Glow on October 1, 2018 at 12:50 pm
“The Mostly Complete Chart of Neural Networks, Explained” was published in August 2017, and identified 27 different types of neural networks. The number of neural network types has increased since the... […]
- New Paper: Survey of FPGAs for Convolutional Neural Networks on September 30, 2018 at 8:06 am
Sparsh Mittal has just published a new paper on the use of FPGAs for Convolutional Neural Networks. Deep convolutional neural networks (CNNs) have recently shown very high accuracy in a wide range of ... […]
- Neural networks detect defects in metal 3D printing on September 27, 2018 at 5:23 pm
For years, Lawrence Livermore National Laboratory engineers and scientists have used an array of sensors and imaging techniques to analyze the physics and processes behind metal 3D printing in an ongo... […]
- A new way to explain neural networks on September 26, 2018 at 9:33 am
Explaining exactly how artificial neural networks (ANN) work in a mathless way can sometimes feel like a lost cause, though. They’re often likened to neural pathways in the human brain, but that ... […]
- Cadence’s Deep-Neural-Network Processor Pushes to 3.4 TMACs/W on September 26, 2018 at 9:26 am
Cadence extended its machine-learning (ML) offering with the Tensilica DNA 100 deep-neural-network processor (see figure), which incorporates Tensilica DSP support to handle new network layers. It tar... […]
- What the Heck is an Artificial Neural Network? on September 24, 2018 at 11:33 am
What is an artificial neural network? They power many technologies such as Apple’s A12 Bionic chip, but what are they? Forbes breaks it down. Artificial neural networks use different layers of mathema... […]
via Google News and Bing News