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 article, the metal microheater is integrated on the side of the waveguide rather than on the top, and the tuning efficiency is significantly enhanced, and response time as fast as 2
Abstract: Thermal tuning acts as one of the most fundamental roles in integrated silicon photonics since it can provide flexibility and reconfigurability. Low tuning power and fast tuning speed are long-term pursuing goals in terms of the performance of the thermal tuning. Here, we propose and experimentally demonstrate an efficient thermal tuning scheme employing the metallic metal heater. The slow-light effect in the photonic crystal waveguide is employed to enhance the performance of the metal microheater. Meanwhile, the metal microheater is integrated on the side of the waveguide rather than on the top. Thanks to both the slow-light effect and the side-integrated microheater, the tuning efficiency is significantly enhanced, and the response time as fast as 2 $\mu \text{s}$ is obtained. Since the thermal tuning with metal heater has been widely applied in silicon photonics, the proposed scheme may provide a valuable solution towards the performance enhancement of the thermal tuning in silicon photonics.
5 citations
Cites methods from "Deep learning with coherent nanopho..."
...As one of the most feasible methods to manipulate the phase and intensity of the light on chip, thermal tuning has been extensively applied in integrated photonics including the largescale photonic phased array antenna [3], integrated microwave photonics [4]–[6], optical artificial neural networks chip [7]...
[...]
TL;DR: In this article , a photoelectrical synaptic system based on high-quality epitaxial Ba0.6Sr0.4TiO3 (BST) films was presented.
Abstract: Electronic synaptic devices with photoelectric sensing function are becoming increasingly important in the development of neuromorphic computing system. Here, we present a photoelectrical synaptic system based on high-quality epitaxial Ba0.6Sr0.4TiO3 (BST) films in which the resistance ramp characteristic of the device provides the possibility to simulate synaptic behavior. The memristor with the Pt/BST/Nb:SrTiO3 structure exhibits reliable I–V characteristics and adjustable resistance modulation characteristics. The device can faithfully demonstrate synaptic functions, such as potentiation and depression, spike time-dependent plasticity, and paired pulse facilitation, and the recognition accuracy of handwritten digits was as high as 92.2%. Interestingly, the functions of visual perception, visual memory, and color recognition of the human eyes have also been realized based on the device. This work will provide a strong candidate for the neuromorphic computing hardware system of photoelectric synaptic devices.
5 citations
TL;DR: In this article , an all-optical implementation of a nonlinear activation function based on germanium silicon hybrid integration was proposed and demonstrated for optical neural networks (ONNs) to achieve more various functions.
Abstract: Nonlinear activation functions are crucial for optical neural networks (ONNs) to achieve more various functions. However, the current nonlinear functions suffer from some dilemma, including high power consumption, high loss, and limited bandwidth. Here, we propose and demonstrate an all-optical implementation of a nonlinear activation function based on germanium silicon hybrid integration. The principle lies in the intrinsic absorption and the carrier-induced refractive index change of germanium in C -band. It has a large operating bandwidth and a response frequency of 70 MHz, with a loss of 4.28 dB and a threshold power of 5.1 mW. Adopting it to the MNIST handwriting data set classification, it shows an improvement in accuracy from 91.6% to 96.8%. This proves that our scheme has great potential for advanced ONN applications.
5 citations
TL;DR: In this paper, a review is given for the recent progresses of Si/2D photodetectors working in the wavelength band from near-infrared to midinfrared, which are attractive for many applications.
Abstract: Two-dimensional materials (2DMs) have been used widely in constructing photodetectors (PDs) because of their advantages in flexible integration and ultrabroad operation wavelength range. Specifically, 2DM PDs on silicon have attracted much attention because silicon microelectronics and silicon photonics have been developed successfully for many applications. 2DM PDs meet the imperious demand of silicon photonics on low-cost, high-performance, and broadband photodetection. In this work, a review is given for the recent progresses of Si/2DM PDs working in the wavelength band from near-infrared to mid-infrared, which are attractive for many applications. The operation mechanisms and the device configurations are summarized in the first part. The waveguide-integrated PDs and the surface-illuminated PDs are then reviewed in details, respectively. The discussion and outlook for 2DM PDs on silicon are finally given.
5 citations
DOI•
22 Nov 2021TL;DR: In this article, the authors investigated the nonlinear dynamics of two coupled fiber ring resonators, coherently driven by a single laser beam, and comprehensively explored the optical switching arising when scanning the detuning of the undriven cavity.
Abstract: We experimentally investigate the nonlinear dynamics of two coupled fiber ring resonators, coherently driven by a single laser beam. We comprehensively explore the optical switching arising when scanning the detuning of the undriven cavity, and show how the driven cavity detuning dramatically changes the resulting hysteresis cycle. By driving the photonic dimer out-of-equilibrium, we observe the occurrence of stable self-switching oscillations near avoided resonance crossings. All results agree well with the driven-dissipative Bose-Hubbard dimer model in the weakly coupled regime.
5 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]....
[...]