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
TL;DR: In this paper, the authors demonstrate the capability of the previously unexplored material Sb2Se3 for ultralow-loss programmable silicon photonics, which enables an unprecedented optical phase control exceeding 10π radians in a Mach-Zehnder interferometer.
Abstract: The next generation of silicon-based photonic processors and neural and quantum networks need to be adaptable, reconfigurable, and programmable. Phase change technology offers proven nonvolatile electronic programmability; however, the materials used to date have shown prohibitively high optical losses, which are incompatible with integrated photonic platforms. Here, we demonstrate the capability of the previously unexplored material Sb2Se3 for ultralow-loss programmable silicon photonics. The favorable combination of large refractive index contrast and ultralow losses seen in Sb2Se3 facilitates an unprecedented optical phase control exceeding 10π radians in a Mach-Zehnder interferometer. To demonstrate full control over the flow of light, we introduce nanophotonic digital patterning as a previously unexplored conceptual approach with a footprint orders of magnitude smaller than state-of-the-art interferometer meshes. Our approach enables a wealth of possibilities in high-density reconfiguration of optical functionalities on silicon chip.
77 citations
TL;DR: In this paper, a singular-value decomposition approach is proposed to find the best orthogonal channels for communicating between surfaces or volumes, or for optimally describing the inputs and outputs of a complicated optical system or wave scatterer.
Abstract: Modes generally provide an economical description of waves, reducing complicated wave functions to finite numbers of mode amplitudes, as in propagating fiber modes and ideal laser beams. But finding a corresponding mode description for counting the best orthogonal channels for communicating between surfaces or volumes, or for optimally describing the inputs and outputs of a complicated optical system or wave scatterer, requires a different approach. The singular-value decomposition approach we describe here gives the necessary optimal source and receiver "communication modes" pairs and device or scatterer input and output "mode-converter basis function" pairs. These define the best communication or input/output channels, allowing precise counting and straightforward calculations. Here we introduce all the mathematics and physics of this approach, which works for acoustic, radio-frequency and optical waves, including full vector electromagnetic behavior, and is valid from nanophotonic scales to large systems. We show several general behaviors of communications modes, including various heuristic results. We also establish a new "M-gauge" for electromagnetism that clarifies the number of vector wave channels and allows a simple and general quantization. This approach also gives a new modal "M-coefficient" version of Einstein's A&B coefficient argument and revised versions of Kirchhoff's radiation laws. The article is written in a tutorial style to introduce the approach and its consequences.
74 citations
TL;DR: Recently developed molecular biosensing approaches based on the combination of dielectric metasurfaces and imaging detection are highlighted in comparison to traditional plasmonic geometries, and the unique potential of artificial intelligence techniques for nanophotonic sensor design and data analysis is emphasized.
Abstract: Molecular spectroscopy provides unique information on the internal structure of biological materials by detecting the characteristic vibrational signatures of their constituent chemical bonds at infrared frequencies. Nanophotonic antennas and metasurfaces have driven this concept towards few-molecule sensitivity by confining incident light into intense hot spots of the electromagnetic fields, providing strongly enhanced light-matter interaction. In this Minireview, recently developed molecular biosensing approaches based on the combination of dielectric metasurfaces and imaging detection are highlighted in comparison to traditional plasmonic geometries, and the unique potential of artificial intelligence techniques for nanophotonic sensor design and data analysis is emphasized. Because of their spectrometer-less operation principle, such imaging-based approaches hold great promise for miniaturized biosensors in practical point-of-care or field-deployable applications.
74 citations
01 Feb 2020
TL;DR: A new approach for using the intelligence aspects of artificial intelligence for knowledge discovery rather than device optimization in electromagnetic (EM) nanostructures is presented that combines the dimensionality reduction technique with convex‐hull and one‐class support‐vector‐machine (SVM) algorithms to find the range of the feasible responses in the latent response space of the EM nanostructure.
Abstract: We present here a new approach for using the intelligence aspects of artificial intelligence for knowledge discovery rather than device optimization in electromagnetic (EM) nanostructures. This approach uses training data obtained through full-wave EM simulations of a series of nanostructures to train geometric deep learning algorithms to assess the range of feasible responses as well as the feasibility of a desired response from a class of EM nanostructures. To facilitate the knowledge discovery and reduce the computation complexity, our approach combines the dimensionality reduction technique (using an autoencoder) with convex-hull and one-class support-vector-machine (SVM) algorithms to find the range of the feasible responses in the latent (or the reduced) response space of the EM nanostructure. We show that by using a small set of training instances (compared to all possible structures), our approach can provide better than 95% accuracy in assessing the feasibility of a given response. More importantly, the one-class SVM algorithm can be trained to provide the degree of feasibility (or unfeasibility) of a response from a given nanostructure. This important information can be used to modify the initial structure to an alternative one that can enable an initially unfeasible response. To show the applicability of our approach, we apply it to two important classes of binary metasurfaces (MSs), formed by array of plasmonic nanostructures, and periodic MSs formed by an array of dielectric nanopillars. In addition to theoretical results, we show the experimental results obtained by fabricating several MSs of the second class. Our theoretical and experimental results confirm the unique features of this approach for knowledge discovery in EM nanostructures.
74 citations
TL;DR: In this article, the phase modulation and low optical loss of Sb2S3 are experimentally demonstrated for the first time in integrated photonic platforms at both 750nm and 1550nm.
Abstract: Phase change materials (PCMs) have long been used as a storage medium in rewritable compact disk and later in random access memory. In recent years, the integration of PCMs with nanophotonic structures has introduced a new paradigm for non-volatile reconfigurable optics. However, the high loss of the archetypal PCM Ge2Sb2Te5 in both visible and telecommunication wavelengths has fundamentally limited its applications. Sb2S3 has recently emerged as a wide-bandgap PCM with transparency windows ranging from 610nm to near-IR. In this paper, the strong optical phase modulation and low optical loss of Sb2S3 are experimentally demonstrated for the first time in integrated photonic platforms at both 750nm and 1550nm. As opposed to silicon, the thermo-optic coefficient of Sb2S3 is shown to be negative, making the Sb2S3-Si hybrid platform less sensitive to thermal fluctuation. Finally, a Sb2S3 integrated non-volatile microring switch is demonstrated which can be tuned electrically between a high and low transmission state with a contrast over 30dB. Our work experimentally verified the prominent phase modification and low loss of Sb2S3 in wavelength ranges relevant for both solid-state quantum emitter and telecommunication, enabling potential applications such as optical field programmable gate array, post-fabrication trimming, and large-scale integrated quantum photonic network.
73 citations
References
More filters
Proceedings Article•
03 Dec 2012TL;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
[...]
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
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))....
[...]
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
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]....
[...]