scispace - formally typeset
Search or ask a question
Journal ArticleDOI

Deep learning with coherent nanophotonic circuits

01 Jul 2017-Vol. 11, Iss: 7, pp 441-446
TL;DR: A new architecture for a fully optical neural network is demonstrated that enables a computational speed enhancement of at least two orders of magnitude and three order of magnitude in power efficiency over state-of-the-art electronics.
Abstract: Artificial Neural Networks have dramatically improved performance for many machine learning tasks. We demonstrate a new architecture for a fully optical neural network that enables a computational speed enhancement of at least two orders of magnitude and three orders of magnitude in power efficiency over state-of-the-art electronics.
Citations
More filters
Journal ArticleDOI
TL;DR: In this paper, anisotropic all-dielectric metamaterials open a new degree of freedom in total internal reflection to shorten the decay length of evanescent waves.
Abstract: Ultra-compact, densely integrated optical components manufactured on a CMOS-foundry platform are highly desirable for optical information processing and electronic-photonic co-integration. However, the large spatial extent of evanescent waves arising from nanoscale confinement, ubiquitous in silicon photonic devices, causes significant cross-talk and scattering loss. Here, we demonstrate that anisotropic all-dielectric metamaterials open a new degree of freedom in total internal reflection to shorten the decay length of evanescent waves. We experimentally show the reduction of cross-talk by greater than 30 times and the bending loss by greater than 3 times in densely integrated, ultra-compact photonic circuit blocks. Our prototype all-dielectric metamaterial-waveguide achieves a low propagation loss of approximately 3.7±1.0 dB/cm, comparable to those of silicon strip waveguides. Our approach marks a departure from interference-based confinement as in photonic crystals or slot waveguides, which utilize nanoscale field enhancement. Its ability to suppress evanescent waves without substantially increasing the propagation loss shall pave the way for all-dielectric metamaterial-based dense integration.

122 citations

Journal ArticleDOI
TL;DR: In this article, the phase change dynamics of Ge2Sb2Te5 (GST) embedded on top of a microring resonator is exploited to alleviate the energy constraints of PCMs in electrical domain.
Abstract: The rapid growth of brain-inspired computing coupled with the inefficiencies in the CMOS implementations of neuromrphic systems has led to intense exploration of efficient hardware implementations of the functional units of the brain, namely, neurons and synapses. However, efforts have largely been invested in implementations in the electrical domain with potential limitations of switching speed, packing density of large integrated systems and interconnect losses. As an alternative, neuromorphic engineering in the photonic domain has recently gained attention. In this work, we propose a purely photonic operation of an Integrate-and-Fire Spiking neuron, based on the phase change dynamics of Ge2Sb2Te5 (GST) embedded on top of a microring resonator, which alleviates the energy constraints of PCMs in electrical domain. We also show that such a neuron can be potentially integrated with on-chip synapses into an all-Photonic Spiking Neural network inferencing framework which promises to be ultrafast and can potentially offer a large operating bandwidth.

119 citations

Journal ArticleDOI
TL;DR: A bimodal artificial sensory neuron is developed to implement the sensory fusion processes and enhanced recognition capability achieved on fused visual/haptic cues is confirmed by simulation of a multi-transparency pattern recognition task.
Abstract: Human behaviors are extremely sophisticated, relying on the adaptive, plastic and event-driven network of sensory neurons. Such neuronal system analyzes multiple sensory cues efficiently to establish accurate depiction of the environment. Here, we develop a bimodal artificial sensory neuron to implement the sensory fusion processes. Such a bimodal artificial sensory neuron collects optic and pressure information from the photodetector and pressure sensors respectively, transmits the bimodal information through an ionic cable, and integrates them into post-synaptic currents by a synaptic transistor. The sensory neuron can be excited in multiple levels by synchronizing the two sensory cues, which enables the manipulating of skeletal myotubes and a robotic hand. Furthermore, enhanced recognition capability achieved on fused visual/haptic cues is confirmed by simulation of a multi-transparency pattern recognition task. Our biomimetic design has the potential to advance technologies in cyborg and neuromorphic systems by endowing them with supramodal perceptual capabilities.

118 citations

Journal ArticleDOI
10 Feb 2021-Nature
TL;DR: This result paves the way for the development and proliferation of low-cost, compact and high-performance 3D imaging cameras that could be used in applications from robotics and autonomous navigation to augmented reality and healthcare.
Abstract: Accurate three-dimensional (3D) imaging is essential for machines to map and interact with the physical world1,2 Although numerous 3D imaging technologies exist, each addressing niche applications with varying degrees of success, none has achieved the breadth of applicability and impact that digital image sensors have in the two-dimensional imaging world3–10 A large-scale two-dimensional array of coherent detector pixels operating as a light detection and ranging system could serve as a universal 3D imaging platform Such a system would offer high depth accuracy and immunity to interference from sunlight, as well as the ability to measure the velocity of moving objects directly11 Owing to difficulties in providing electrical and photonic connections to every pixel, previous systems have been restricted to fewer than 20 pixels12–15 Here we demonstrate the operation of a large-scale coherent detector array, consisting of 512 pixels, in a 3D imaging system Leveraging recent advances in the monolithic integration of photonic and electronic circuits, a dense array of optical heterodyne detectors is combined with an integrated electronic readout architecture, enabling straightforward scaling to arbitrarily large arrays Two-axis solid-state beam steering eliminates any trade-off between field of view and range Operating at the quantum noise limit16,17, our system achieves an accuracy of 31 millimetres at a distance of 75 metres when using only 4 milliwatts of light, an order of magnitude more accurate than existing solid-state systems at such ranges Future reductions of pixel size using state-of-the-art components could yield resolutions in excess of 20 megapixels for arrays the size of a consumer camera sensor This result paves the way for the development and proliferation of low-cost, compact and high-performance 3D imaging cameras that could be used in applications from robotics and autonomous navigation to augmented reality and healthcare A compact, high-performance silicon photonics-based light detection and ranging system for three-dimensional imaging is developed that should be amenable to low-cost mass manufacturing

118 citations

Journal ArticleDOI
TL;DR: In this article, an integrated phase-change memory cell that can be electrically or optically switched between binary or multilevel states is presented. But it cannot be read out both optically and electrically, offering a new strategy for merging computing and communications technologies.
Abstract: Modern-day computers rely on electrical signaling for the processing and storage of data, which is bandwidth-limited and power hungry. This fact has long been realized in the communications field, where optical signaling is the norm. However, exploiting optical signaling in computing will require new on-chip devices that work seamlessly in both electrical and optical domains, without the need for repeated electrical-to-optical conversion. Phase-change devices can, in principle, provide such dual electrical-optical operation, but assimilating both functionalities into a single device has so far proved elusive owing to conflicting requirements of size-limited electrical switching and diffraction-limited optical response. Here, we combine plasmonics, photonics, and electronics to deliver an integrated phase-change memory cell that can be electrically or optically switched between binary or multilevel states. Crucially, this device can also be simultaneously read out both optically and electrically, offering a new strategy for merging computing and communications technologies.

116 citations

References
More filters
Proceedings Article
03 Dec 2012
TL;DR: The state-of-the-art performance of CNNs was achieved by Deep Convolutional Neural Networks (DCNNs) as discussed by the authors, which consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax.
Abstract: We trained a large, deep convolutional neural network to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 different classes. On the test data, we achieved top-1 and top-5 error rates of 37.5% and 17.0% which is considerably better than the previous state-of-the-art. The neural network, which has 60 million parameters and 650,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax. To make training faster, we used non-saturating neurons and a very efficient GPU implementation of the convolution operation. To reduce overriding in the fully-connected layers we employed a recently-developed regularization method called "dropout" that proved to be very effective. We also entered a variant of this model in the ILSVRC-2012 competition and achieved a winning top-5 test error rate of 15.3%, compared to 26.2% achieved by the second-best entry.

73,978 citations

Journal ArticleDOI
28 May 2015-Nature
TL;DR: Deep learning is making major advances in solving problems that have resisted the best attempts of the artificial intelligence community for many years, and will have many more successes in the near future because it requires very little engineering by hand and can easily take advantage of increases in the amount of available computation and data.
Abstract: Deep learning allows computational models that are composed of multiple processing layers to learn representations of data with multiple levels of abstraction. These methods have dramatically improved the state-of-the-art in speech recognition, visual object recognition, object detection and many other domains such as drug discovery and genomics. Deep learning discovers intricate structure in large data sets by using the backpropagation algorithm to indicate how a machine should change its internal parameters that are used to compute the representation in each layer from the representation in the previous layer. Deep convolutional nets have brought about breakthroughs in processing images, video, speech and audio, whereas recurrent nets have shone light on sequential data such as text and speech.

46,982 citations

Journal ArticleDOI
26 Feb 2015-Nature
TL;DR: This work bridges the divide between high-dimensional sensory inputs and actions, resulting in the first artificial agent that is capable of learning to excel at a diverse array of challenging tasks.
Abstract: The theory of reinforcement learning provides a normative account, deeply rooted in psychological and neuroscientific perspectives on animal behaviour, of how agents may optimize their control of an environment. To use reinforcement learning successfully in situations approaching real-world complexity, however, agents are confronted with a difficult task: they must derive efficient representations of the environment from high-dimensional sensory inputs, and use these to generalize past experience to new situations. Remarkably, humans and other animals seem to solve this problem through a harmonious combination of reinforcement learning and hierarchical sensory processing systems, the former evidenced by a wealth of neural data revealing notable parallels between the phasic signals emitted by dopaminergic neurons and temporal difference reinforcement learning algorithms. While reinforcement learning agents have achieved some successes in a variety of domains, their applicability has previously been limited to domains in which useful features can be handcrafted, or to domains with fully observed, low-dimensional state spaces. Here we use recent advances in training deep neural networks to develop a novel artificial agent, termed a deep Q-network, that can learn successful policies directly from high-dimensional sensory inputs using end-to-end reinforcement learning. We tested this agent on the challenging domain of classic Atari 2600 games. We demonstrate that the deep Q-network agent, receiving only the pixels and the game score as inputs, was able to surpass the performance of all previous algorithms and achieve a level comparable to that of a professional human games tester across a set of 49 games, using the same algorithm, network architecture and hyperparameters. This work bridges the divide between high-dimensional sensory inputs and actions, resulting in the first artificial agent that is capable of learning to excel at a diverse array of challenging tasks.

23,074 citations


"Deep learning with coherent nanopho..." refers background or methods in this paper

  • ...The computational resolution of ONNs is limited by practical non-idealities, including (1) thermal crosstalk between phase shifters in interferometers, (2) optical coupling drift, (3) the finite precision with which an optical phase can be set (16 bits in our case), (4) photodetection noise and (5) finite photodetection dynamic range (30 dB in our case)....

    [...]

  • ...(3) Once a neural network is trained, the architecture can be passive, and computation on the optical signals will be performed without additional energy input....

    [...]

  • ...We used four instances of the OIU to realize the following matrix transformations in the spatial-mode basis: (1) U((1))Σ((1)), (2) V((1)), (3) U((2))Σ((2)) and (4) V((2))....

    [...]

  • ...Transformations (1) and (2) realize the first matrix M((1)), and (3) and (4) implement M((2))....

    [...]

Journal ArticleDOI
28 Jul 2006-Science
TL;DR: In this article, an effective way of initializing the weights that allows deep autoencoder networks to learn low-dimensional codes that work much better than principal components analysis as a tool to reduce the dimensionality of data is described.
Abstract: High-dimensional data can be converted to low-dimensional codes by training a multilayer neural network with a small central layer to reconstruct high-dimensional input vectors. Gradient descent can be used for fine-tuning the weights in such "autoencoder" networks, but this works well only if the initial weights are close to a good solution. We describe an effective way of initializing the weights that allows deep autoencoder networks to learn low-dimensional codes that work much better than principal components analysis as a tool to reduce the dimensionality of data.

16,717 citations

Journal ArticleDOI
TL;DR: This historical survey compactly summarizes relevant work, much of it from the previous millennium, review deep supervised learning, unsupervised learning, reinforcement learning & evolutionary computation, and indirect search for short programs encoding deep and large networks.

14,635 citations


"Deep learning with coherent nanopho..." refers methods in this paper

  • ...ANNs can be trained by feeding training data into the input layer and then computing the output by forward propagation; weighting parameters in each matrix are subsequently optimized using back propagation [16]....

    [...]