While machine learning has been around a long time, deep learning has taken on a life of its own lately. The reason for that has mostly to do with the increasing amounts of computing power that have become widely availablealong with the burgeoning quantities of data that can be easily harvested and used to train neural networks.
The amount of computing power at people's fingertips started growing in leaps and bounds at the turn of the millennium, when graphical processing units (GPUs) began to be harnessed for nongraphical calculations, a trend that has become increasingly pervasive over the past decade. But the computing demands of deep learning have been rising even faster. This dynamic has spurred engineers to develop electronic hardware accelerators specifically targeted to deep learning, Google's Tensor Processing Unit (TPU) being a prime example.
Here, I will describe a very different approach to this problemusing optical processors to carry out neural-network calculations with photons instead of electrons. To understand how optics can serve here, you need to know a little bit about how computers currently carry out neural-network calculations. So bear with me as I outline what goes on under the hood.
Almost invariably, artificial neurons are constructed using special software running on digital electronic computers of some sort. That software provides a given neuron with multiple inputs and one output. The state of each neuron depends on the weighted sum of its inputs, to which a nonlinear function, called an activation function, is applied. The result, the output of this neuron, then becomes an input for various other neurons.
Reducing the energy needs of neural networks might require computing with light
For computational efficiency, these neurons are grouped into layers, with neurons connected only to neurons in adjacent layers. The benefit of arranging things that way, as opposed to allowing connections between any two neurons, is that it allows certain mathematical tricks of linear algebra to be used to speed the calculations.
While they are not the whole story, these linear-algebra calculations are the most computationally demanding part of deep learning, particularly as the size of the network grows. This is true for both training (the process of determining what weights to apply to the inputs for each neuron) and for inference (when the neural network is providing the desired results).
What are these mysterious linear-algebra calculations? They aren't so complicated really. They involve operations on matrices, which are just rectangular arrays of numbersspreadsheets if you will, minus the descriptive column headers you might find in a typical Excel file.
This is great news because modern computer hardware has been very well optimized for matrix operations, which were the bread and butter of high-performance computing long before deep learning became popular. The relevant matrix calculations for deep learning boil down to a large number of multiply-and-accumulate operations, whereby pairs of numbers are multiplied together and their products are added up.
Over the years, deep learning has required an ever-growing number of these multiply-and-accumulate operations. Consider LeNet, a pioneering deep neural network, designed to do image classification. In 1998 it was shown to outperform other machine techniques for recognizing handwritten letters and numerals. But by 2012 AlexNet, a neural network that crunched through about 1,600 times as many multiply-and-accumulate operations as LeNet, was able to recognize thousands of different types of objects in images.
Advancing from LeNet's initial success to AlexNet required almost 11 doublings of computing performance. During the 14 years that took, Moore's law provided much of that increase. The challenge has been to keep this trend going now that Moore's law is running out of steam. The usual solution is simply to throw more computing resourcesalong with time, money, and energyat the problem.
As a result, training today's large neural networks often has a significant environmental footprint. One 2019 study found, for example, that training a certain deep neural network for natural-language processing produced five times the CO2 emissions typically associated with driving an automobile over its lifetime.
Improvements in digital electronic computers allowed deep learning to blossom, to be sure. But that doesn't mean that the only way to carry out neural-network calculations is with such machines. Decades ago, when digital computers were still relatively primitive, some engineers tackled difficult calculations using analog computers instead. As digital electronics improved, those analog computers fell by the wayside. But it may be time to pursue that strategy once again, in particular when the analog computations can be done optically.
It has long been known that optical fibers can support much higher data rates than electrical wires. That's why all long-haul communication lines went optical, starting in the late 1970s. Since then, optical data links have replaced copper wires for shorter and shorter spans, all the way down to rack-to-rack communication in data centers. Optical data communication is faster and uses less power. Optical computing promises the same advantages.
But there is a big difference between communicating data and computing with it. And this is where analog optical approaches hit a roadblock. Conventional computers are based on transistors, which are highly nonlinear circuit elementsmeaning that their outputs aren't just proportional to their inputs, at least when used for computing. Nonlinearity is what lets transistors switch on and off, allowing them to be fashioned into logic gates. This switching is easy to accomplish with electronics, for which nonlinearities are a dime a dozen. But photons follow Maxwell's equations, which are annoyingly linear, meaning that the output of an optical device is typically proportional to its inputs.
The trick is to use the linearity of optical devices to do the one thing that deep learning relies on most: linear algebra.
To illustrate how that can be done, I'll describe here a photonic device that, when coupled to some simple analog electronics, can multiply two matrices together. Such multiplication combines the rows of one matrix with the columns of the other. More precisely, it multiplies pairs of numbers from these rows and columns and adds their products togetherthe multiply-and-accumulate operations I described earlier. My MIT colleagues and I published a paper about how this could be done in 2019. We're working now to build such an optical matrix multiplier.
Optical data communication is faster and uses less power. Optical computing promises the same advantages.
The basic computing unit in this device is an optical element called a beam splitter. Although its makeup is in fact more complicated, you can think of it as a half-silvered mirror set at a 45-degree angle. If you send a beam of light into it from the side, the beam splitter will allow half that light to pass straight through it, while the other half is reflected from the angled mirror, causing it to bounce off at 90 degrees from the incoming beam.
Now shine a second beam of light, perpendicular to the first, into this beam splitter so that it impinges on the other side of the angled mirror. Half of this second beam will similarly be transmitted and half reflected at 90 degrees. The two output beams will combine with the two outputs from the first beam. So this beam splitter has two inputs and two outputs.
To use this device for matrix multiplication, you generate two light beams with electric-field intensities that are proportional to the two numbers you want to multiply. Let's call these field intensities x and y. Shine those two beams into the beam splitter, which will combine these two beams. This particular beam splitter does that in a way that will produce two outputs whose electric fields have values of (x + y)/2 and (x y)/2.
In addition to the beam splitter, this analog multiplier requires two simple electronic componentsphotodetectorsto measure the two output beams. They don't measure the electric field intensity of those beams, though. They measure the power of a beam, which is proportional to the square of its electric-field intensity.
Why is that relation important? To understand that requires some algebrabut nothing beyond what you learned in high school. Recall that when you square (x + y)/2 you get (x2 + 2xy + y2)/2. And when you square (x y)/2, you get (x2 2xy + y2)/2. Subtracting the latter from the former gives 2xy.
Pause now to contemplate the significance of this simple bit of math. It means that if you encode a number as a beam of light of a certain intensity and another number as a beam of another intensity, send them through such a beam splitter, measure the two outputs with photodetectors, and negate one of the resulting electrical signals before summing them together, you will have a signal proportional to the product of your two numbers.
Simulations of the integrated Mach-Zehnder interferometer found in Lightmatter's neural-network accelerator show three different conditions whereby light traveling in the two branches of the interferometer undergoes different relative phase shifts (0 degrees in a, 45 degrees in b, and 90 degrees in c).Lightmatter
My description has made it sound as though each of these light beams must be held steady. In fact, you can briefly pulse the light in the two input beams and measure the output pulse. Better yet, you can feed the output signal into a capacitor, which will then accumulate charge for as long as the pulse lasts. Then you can pulse the inputs again for the same duration, this time encoding two new numbers to be multiplied together. Their product adds some more charge to the capacitor. You can repeat this process as many times as you like, each time carrying out another multiply-and-accumulate operation.
Using pulsed light in this way allows you to perform many such operations in rapid-fire sequence. The most energy-intensive part of all this is reading the voltage on that capacitor, which requires an analog-to-digital converter. But you don't have to do that after each pulseyou can wait until the end of a sequence of, say, N pulses. That means that the device can perform N multiply-and-accumulate operations using the same amount of energy to read the answer whether N is small or large. Here, N corresponds to the number of neurons per layer in your neural network, which can easily number in the thousands. So this strategy uses very little energy.
Sometimes you can save energy on the input side of things, too. That's because the same value is often used as an input to multiple neurons. Rather than that number being converted into light multiple timesconsuming energy each timeit can be transformed just once, and the light beam that is created can be split into many channels. In this way, the energy cost of input conversion is amortized over many operations.
Splitting one beam into many channels requires nothing more complicated than a lens, but lenses can be tricky to put onto a chip. So the device we are developing to perform neural-network calculations optically may well end up being a hybrid that combines highly integrated photonic chips with separate optical elements.
I've outlined here the strategy my colleagues and I have been pursuing, but there are other ways to skin an optical cat. Another promising scheme is based on something called a Mach-Zehnder interferometer, which combines two beam splitters and two fully reflecting mirrors. It, too, can be used to carry out matrix multiplication optically. Two MIT-based startups, Lightmatter and Lightelligence, are developing optical neural-network accelerators based on this approach. Lightmatter has already built a prototype that uses an optical chip it has fabricated. And the company expects to begin selling an optical accelerator board that uses that chip later this year.
Another startup using optics for computing is Optalysis, which hopes to revive a rather old concept. One of the first uses of optical computing back in the 1960s was for the processing of synthetic-aperture radar data. A key part of the challenge was to apply to the measured data a mathematical operation called the Fourier transform. Digital computers of the time struggled with such things. Even now, applying the Fourier transform to large amounts of data can be computationally intensive. But a Fourier transform can be carried out optically with nothing more complicated than a lens, which for some years was how engineers processed synthetic-aperture data. Optalysis hopes to bring this approach up to date and apply it more widely.
Theoretically, photonics has the potential to accelerate deep learning by several orders of magnitude.
There is also a company called Luminous, spun out of Princeton University, which is working to create spiking neural networks based on something it calls a laser neuron. Spiking neural networks more closely mimic how biological neural networks work and, like our own brains, are able to compute using very little energy. Luminous's hardware is still in the early phase of development, but the promise of combining two energy-saving approachesspiking and opticsis quite exciting.
There are, of course, still many technical challenges to be overcome. One is to improve the accuracy and dynamic range of the analog optical calculations, which are nowhere near as good as what can be achieved with digital electronics. That's because these optical processors suffer from various sources of noise and because the digital-to-analog and analog-to-digital converters used to get the data in and out are of limited accuracy. Indeed, it's difficult to imagine an optical neural network operating with more than 8 to 10 bits of precision. While 8-bit electronic deep-learning hardware exists (the Google TPU is a good example), this industry demands higher precision, especially for neural-network training.
There is also the difficulty integrating optical components onto a chip. Because those components are tens of micrometers in size, they can't be packed nearly as tightly as transistors, so the required chip area adds up quickly. A 2017 demonstration of this approach by MIT researchers involved a chip that was 1.5 millimeters on a side. Even the biggest chips are no larger than several square centimeters, which places limits on the sizes of matrices that can be processed in parallel this way.
There are many additional questions on the computer-architecture side that photonics researchers tend to sweep under the rug. What's clear though is that, at least theoretically, photonics has the potential to accelerate deep learning by several orders of magnitude.
Based on the technology that's currently available for the various components (optical modulators, detectors, amplifiers, analog-to-digital converters), it's reasonable to think that the energy efficiency of neural-network calculations could be made 1,000 times better than today's electronic processors. Making more aggressive assumptions about emerging optical technology, that factor might be as large as a million. And because electronic processors are power-limited, these improvements in energy efficiency will likely translate into corresponding improvements in speed.
Many of the concepts in analog optical computing are decades old. Some even predate silicon computers. Schemes for optical matrix multiplication, and even for optical neural networks, were first demonstrated in the 1970s. But this approach didn't catch on. Will this time be different? Possibly, for three reasons.
First, deep learning is genuinely useful now, not just an academic curiosity. Second, we can't rely on Moore's Law alone to continue improving electronics. And finally, we have a new technology that was not available to earlier generations: integrated photonics. These factors suggest that optical neural networks will arrive for real this timeand the future of such computations may indeed be photonic.
Originally posted here:
D-Waves 500-Qubit Machine Hits the Cloud - IEEE Spectrum
- Netflixs 3 Body Problem: The science explained by an astrophysicist - Vox.com - March 24th, 2024 [March 24th, 2024]
- Entanglement entropies of nuclear systems gro - EurekAlert - March 24th, 2024 [March 24th, 2024]
- The Quest for a Theory of Everything Scientists Put Einstein to the Test - SciTechDaily - March 24th, 2024 [March 24th, 2024]
- Vibrating atoms are seen 'tuning' the energy of a single electron - Earth.com - March 24th, 2024 [March 24th, 2024]
- Innovator Spotlight: Joseph Maciejko | The Quad - University of Alberta - March 24th, 2024 [March 24th, 2024]
- A Breakthrough in the Control of Quantum Phenomena at Room Temperature Has Been Achieved, Researchers Say - The Debrief - February 16th, 2024 [February 16th, 2024]
- The End of the Quantum Ice Age: Room Temperature Breakthrough - SciTechDaily - February 16th, 2024 [February 16th, 2024]
- Quantum computer outperformed by new traditional computing - Earth.com - February 16th, 2024 [February 16th, 2024]
- URI program to help STEM professionals pivot into quantum information science careers - The University of Rhode Island - February 16th, 2024 [February 16th, 2024]
- Quantum realm controlled at room temperature for the first time - Earth.com - February 16th, 2024 [February 16th, 2024]
- Quantum Breakthrough: New Method Preserves Information Against All Odds - SciTechDaily - February 16th, 2024 [February 16th, 2024]
- Quantum computers get new design that makes them more "useful" - Earth.com - February 16th, 2024 [February 16th, 2024]
- Beyond Classical Physics: Scientists Discover New State of Matter With Chiral Properties - SciTechDaily - February 16th, 2024 [February 16th, 2024]
- Quantum research sheds light on the mystery of high-temperature superconductivity - Tech Explorist - February 16th, 2024 [February 16th, 2024]
- Unlocking the Mysteries of Quantum Many-Body Systems: A Look at Quantum Simulators and Universal Scaling ... - Medriva - February 16th, 2024 [February 16th, 2024]
- Functioning quantum internet makes giant stride closer to reality - Earth.com - February 13th, 2024 [February 13th, 2024]
- Exploring New Futures in Space: A Revolutionary Integration of Neuroscience, Quantum Physics, and Space Exploration - SETI Institute - February 13th, 2024 [February 13th, 2024]
- Uncovering the Quantum Plateau: Significance and Implications | Nature Physics - Medriva - February 13th, 2024 [February 13th, 2024]
- The State of the Art in Quantum Computing - Medium - February 13th, 2024 [February 13th, 2024]
- Beyond the Visible Universe: New Research Reveals How Gravity Influences the Quantum Realm - SciTechDaily - February 13th, 2024 [February 13th, 2024]
- Leader of IBM's Quantum Safe Team to speak at URI - University of Rhode Island - September 23rd, 2023 [September 23rd, 2023]
- University Assistant Predoctoral, Physics job with UNIVERSITY OF ... - Times Higher Education - September 23rd, 2023 [September 23rd, 2023]
- Zentropy A New Theory That Could Transform Material Science - SciTechDaily - September 23rd, 2023 [September 23rd, 2023]
- Researchers Studying the Quantum Realm Observe Alice in ... - The Debrief - September 23rd, 2023 [September 23rd, 2023]
- Augusta University graduate starts business in the artificial ... - Jagwire Augusta - September 23rd, 2023 [September 23rd, 2023]
- Quantum Echoes: A Revolutionary Method to Store Information as Sound Waves - SciTechDaily - August 14th, 2023 [August 14th, 2023]
- 'Quantum superchemistry' observed for the 1st time ever - Space.com - August 14th, 2023 [August 14th, 2023]
- Quantum Avalanche A Phenomenon That May Revolutionize Microelectronics and Supercomputing - SciTechDaily - August 14th, 2023 [August 14th, 2023]
- Applications of quantum mechanics at the beach - Symmetry magazine - August 14th, 2023 [August 14th, 2023]
- Book Review: On the Origin of Time Stephen Hawking's Final Theory - Moose Jaw Today - August 14th, 2023 [August 14th, 2023]
- Harnessing Quantum Technologies: The Next Big Leap in Global ... - Fagen wasanni - August 14th, 2023 [August 14th, 2023]
- The quantum avalanche - At the Vienna University of Technology, it ... - Chemie.de - August 14th, 2023 [August 14th, 2023]
- Semiconductors: The Linchpin of AI in Quantum Computing - Fagen wasanni - August 14th, 2023 [August 14th, 2023]
- The Promising Collaboration Between AI and Quantum Computing - Fagen wasanni - August 14th, 2023 [August 14th, 2023]
- String theory physicist changed quantum field theory - USC Dornsife College of Letters, Arts and Sciences - August 14th, 2023 [August 14th, 2023]
- QUANTUM SUPERCOMPUTERS. The words Quantum and ... - Medium - August 14th, 2023 [August 14th, 2023]
- Fourteen MIT School of Science professors receive tenure for 2022 ... - MIT News - August 14th, 2023 [August 14th, 2023]
- The Fascinating World of Quantum Integrated Circuits: The Next Big ... - Fagen wasanni - August 14th, 2023 [August 14th, 2023]
- Conclusive Evidence for Modified Gravity: Collapse of Newton's and ... - SciTechDaily - August 14th, 2023 [August 14th, 2023]
- Physicists Open New Path to an Exotic Form of Superconductivity - SciTechDaily - August 14th, 2023 [August 14th, 2023]
- The Principle of Least Action Now Exists in the Quantum Realm - Popular Mechanics - June 10th, 2023 [June 10th, 2023]
- Quantum materials: Electron spin measured for the first time - EurekAlert - June 10th, 2023 [June 10th, 2023]
- Life in a hologram | MIT News | Massachusetts Institute of Technology - MIT News - June 10th, 2023 [June 10th, 2023]
- If Black Holes Evaporate, Everything Evaporates - Universe Today - June 10th, 2023 [June 10th, 2023]
- Clever Ant-Man Easter Egg Links The Movie to the Real World's ... - Startefacts - June 10th, 2023 [June 10th, 2023]
- Quantum Cryptography: The Cutting Edge of Secure Communication - CityLife - June 10th, 2023 [June 10th, 2023]
- This 17-year-old works to make quantum mainstream - Indiatimes.com - June 10th, 2023 [June 10th, 2023]
- The multiverse is doomed and even Spider-Man and The Flash can't save it - Yahoo Entertainment - June 10th, 2023 [June 10th, 2023]
- Physics of Time Travel: A Scientific Perspective - Mirage News - June 10th, 2023 [June 10th, 2023]
- Quantum Spin Liquids: The Future of Superconductors - EnergyPortal.eu - June 10th, 2023 [June 10th, 2023]
- Interview: Three Books That Make Tess Gunty Angry - The New York Times - June 10th, 2023 [June 10th, 2023]
- Events Calendar School of Mathematics and Statistics Colloquium ... - Carleton University - June 10th, 2023 [June 10th, 2023]
- Graphene and Quantum Computing: A Match Made in Heaven - CityLife - June 10th, 2023 [June 10th, 2023]
- A Quantum Computer Simulation Has Reversed Time And Physics May Never Be The Same - Twisted Sifter - June 2nd, 2023 [June 2nd, 2023]
- Realizing the Einstein-Podolsky-Rosen Paradox for Atomic Clouds - Physics - June 2nd, 2023 [June 2nd, 2023]
- The US and UK team up to advance quantum information science - Fermi National Accelerator Laboratory - June 2nd, 2023 [June 2nd, 2023]
- How plants can perform feats of quantum mechanics - Big Think - June 2nd, 2023 [June 2nd, 2023]
- Physicists Make Matter out of Light to Find Quantum Singularities - Scientific American - June 2nd, 2023 [June 2nd, 2023]
- Eventually everything will evaporate, not only black holes - Science Daily - June 2nd, 2023 [June 2nd, 2023]
- Julius-Maximillians-Universitt Wrzburg Researchers Use ... - HPCwire - June 2nd, 2023 [June 2nd, 2023]
- TNTs The Lazarus Project Uses Suspense Trapping to Ask Smart ... - Roger Ebert - June 2nd, 2023 [June 2nd, 2023]
- Quantum Exponential: building a cutting edge quantum technology ... - The Armchair Trader - June 2nd, 2023 [June 2nd, 2023]
- IMDEA Software and IMDEA Networks work to deploy in the ... - EurekAlert - June 2nd, 2023 [June 2nd, 2023]
- Ian Hacking, Eminent Philosopher of Science and Much Else, Dies ... - The New York Times - June 2nd, 2023 [June 2nd, 2023]
- Does mass increase when nearing the speed of light? - Big Think - June 2nd, 2023 [June 2nd, 2023]
- Answering Questions about Boring Numbers, Disasters, Fusion, and ... - Scientific American - June 2nd, 2023 [June 2nd, 2023]
- Spiderman: Across the Spider-verse | Reel World | timesnewspapers ... - Webster-Kirkwood Times, Inc. - June 2nd, 2023 [June 2nd, 2023]
- There's a Secret Way to Get to Absolute Zero. Scientists Just Found It. - Popular Mechanics - May 6th, 2023 [May 6th, 2023]
- Photon Precision: How Quantum Physicists Shattered the Bounds of Sensitivity - SciTechDaily - May 6th, 2023 [May 6th, 2023]
- Do we live in a hologram? Why physics is still mesmerised by this idea - New Scientist - May 6th, 2023 [May 6th, 2023]
- Is Ultimate Truth an Equation? Nah. The Stute - The Stute - May 6th, 2023 [May 6th, 2023]
- UChicago Lab Creates 'Quantum Casino,' a Win-Win to Educate and ... - Polsky Center for Entrepreneurship and Innovation - May 6th, 2023 [May 6th, 2023]
- Physics - Tweezers in Three Dimensions - Physics - May 6th, 2023 [May 6th, 2023]
- Brave new world: On the edge of a second quantum revolution - University of Cape Town News - May 6th, 2023 [May 6th, 2023]
- Researchers pull back the quantum curtain on 'Weyl fermions' - Phys.org - May 6th, 2023 [May 6th, 2023]
- Scale separation: Breaking down unsolvable problems into solvable ones - Phys.org - May 6th, 2023 [May 6th, 2023]
- Postdoctoral Research Associate in Quantum Optics job with ... - Times Higher Education - May 6th, 2023 [May 6th, 2023]
- Australia's first quantum strategy predicts $6 billion in revenue and ... - SmartCompany - May 6th, 2023 [May 6th, 2023]
- Nature's Quantum Secret: Link Discovered Between Photosynthesis ... - SciTechDaily - May 6th, 2023 [May 6th, 2023]
- Two ERC proof of concept grants for the University of Bonn - EurekAlert - May 6th, 2023 [May 6th, 2023]