Linear programmable nanophotonic processors

Harris, Nicholas C.; Carolan, Jacques; Bunandar, Darius; Prabhu, Mihika; Hochberg, Michael; Baehr‐Jones, Tom; Fanto, Michael L.; Smith, A. Matthew; Tison, Christopher C.; Alsing, Paul M.; Englund, Dirk

doi:10.1364/optica.5.001623

Cited by 311 publications

(228 citation statements)

References 78 publications

Supporting

Mentioning

212

Contrasting

Order By: Relevance

“…Also, its simple and scalable design warrants a near-term implementation in optical experiments and is apt for embedding in miniaturized devices compatible with quantum technologies. Indeed, all required optical components have already been experimentally demonstrated on integrated circuits [27][28][29][30][31][32].…”

Section: Discussionmentioning

confidence: 99%

“…For instance, when one clip-to-clip connection 2 To fulfill all the above desiderata, a top-down approach could be to adopt well-established meshes of optical elements for programmable multifunctional nanophotonics hardware [32]. However, their versatility comes at the cost of higher computational resources (should one iteratively adjust their settings according to externally-processed unitary decompositions [31]) or of a less intuitive dependency of the output on the internal components [48,49]. 3 Non-ideal detection efficiency can be counteracted with a control feedback, by sending again a photon to the same decision tree if, in the previous time bin, no photon was collected.…”

Section: Decision Trees As Linear Optical Circuitsmentioning

confidence: 99%

“…The proposed architecture, based on single-photon evolution on a mesh of tunable beamsplitters, is simple, scalable, and a first integration in quantum optical experiments appears to be within the reach of near-term technology.OPEN ACCESS RECEIVED promising features as compared to electronic processors [27][28][29][30]. For instance, nanosecond-scale routing and reconfigurability have already been demonstrated [31][32][33], while encoding information in photons enables decision-making at the speed of light, only limited by the generation and detection rates. Moreover, the use of phase-change materials for in-memory information processing [34] promises to enhance the energy efficiency, since their properties can be modified without continuous external intervention [35,36].…”

mentioning

confidence: 99%

See 2 more Smart Citations

Photonic architecture for reinforcement learning

et al. 2020

View full text Add to dashboard Cite

The last decade has seen an unprecedented growth in artificial intelligence and photonic technologies, both of which drive the limits of modern-day computing devices. In line with these recent developments, this work brings together the state of the art of both fields within the framework of reinforcement learning. We present the blueprint for a photonic implementation of an active learning machine incorporating contemporary algorithms such as SARSA, Q-learning, and projective simulation. We numerically investigate its performance within typical reinforcement learning environments, showing that realistic levels of experimental noise can be tolerated or even be beneficial for the learning process. Remarkably, the architecture itself enables mechanisms of abstraction and generalization, two features which are often considered key ingredients for artificial intelligence. The proposed architecture, based on single-photon evolution on a mesh of tunable beamsplitters, is simple, scalable, and a first integration in quantum optical experiments appears to be within the reach of near-term technology.OPEN ACCESS RECEIVED promising features as compared to electronic processors [27][28][29][30]. For instance, nanosecond-scale routing and reconfigurability have already been demonstrated [31][32][33], while encoding information in photons enables decision-making at the speed of light, only limited by the generation and detection rates. Moreover, the use of phase-change materials for in-memory information processing [34] promises to enhance the energy efficiency, since their properties can be modified without continuous external intervention [35,36]. Importantly, since the architecture uses single photons, decision-making is fueled by genuine quantum randomness. This feature marks a fundamental departure from pseudorandom number generation in conventional devices. (ii) The second contribution is the development of a specific variant of PS based on binary decision trees (tree-PS, or t-PS for short), which is closely connected to the standard PS and suitable for the implementation on a photonic circuit. Furthermore, we discuss how this variant enables key features of artificial intelligence, namely abstraction and generalization [37,38].The Article is structured as follows. In section 2 we summarize the theoretical framework of RL, exemplified by three common approaches: SARSA, Q-learning, and PS. In section 3 we describe the blueprint for a fully integrated, photonic RL agent. We then numerically investigate its performance within two standard RL tasks and under realistic experimental imperfections in section 4. Finally, in section 5 we discuss promising features of this architecture within the context of t-PS.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Decision Trees As Linear Optical Circuitsmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Photonic architecture for reinforcement learning

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Nonlinear activation functions are regarded as one of the key reasons for the power of deep learning compared to classical machine learning methods. Adding nonlinearity into a photonic network substantially changes the functionality compared with previously demonstrated linear photonic circuits [21], [43], [44]. Several different kinds of optical nonlinearities have been proposed for implementation in optical neural networks, such as saturable absorption, optical bistability and two-photon absorption, to name a few [45]- [47].…”

Section: B Convolution Pooling and Activation Layersmentioning

confidence: 99%

“…Programmable integrated photonics has made significant advances, potentially eliminating the need for such a hybrid system [21]. As compared to bulk optics, integrated photonics is a scalable solution in terms of alignment stability and total network size.…”

Section: Introductionmentioning

confidence: 99%

Photonic convolutional neural networks using integrated diffractive optics

Ong¹,

Ang²,

Ooi³

et al. 2020

Preprint

View full text Add to dashboard Cite

With recent rapid advances in photonic integrated circuits, it has been demonstrated that programmable photonic chips can be used to implement artificial neural networks. Convolutional neural networks (CNN) are a class of deep learning methods that have been highly successful in applications such as image classification and speech processing. We present an architecture to implement a photonic CNN using the Fourier transform property of integrated star couplers. We show, in computer simulation, high accuracy image classification using the MNIST dataset. We also model component imperfections in photonic CNN and show that the performance degradation can be recovered in a programmable chip. Our proposed architecture provides a large reduction in physical footprint compared to current implementations as it utilizes the natural advantages of optics and hence offers a scalable pathway towards integrated photonic deep learning processors.

show abstract

A New Family of Ultralow Loss Reversible Phase‐Change Materials for Photonic Integrated Circuits: Sb₂S₃ and Sb₂Se₃

Delaney

Zeimpekis

Lawson

et al. 2020

Adv Funct Materials

362

246

View full text Add to dashboard Cite

Phase‐change materials (PCMs) are seeing tremendous interest for their use in reconfigurable photonic devices; however, the most common PCMs exhibit a large absorption loss in one or both states. Here, Sb2S3 and Sb2Se3 are demonstrated as a class of low loss, reversible alternatives to the standard commercially available chalcogenide PCMs. A contrast of refractive index of Δn = 0.60 for Sb2S3 and Δn = 0.77 for Sb2Se3 is reported, while maintaining very low losses (k < 10−5) in the telecommunications C‐band at 1550 nm. With a stronger absorption in the visible spectrum, Sb2Se3 allows for reversible optical switching using conventional visible wavelength lasers. Here, a stable switching endurance of better than 4000 cycles is demonstrated. To deal with the essentially zero intrinsic absorption losses, a new figure of merit (FOM) is introduced taking into account the measured waveguide losses when integrating these materials onto a standard silicon photonics platform. The FOM of 29 rad phase shift per dB of loss for Sb2Se3 outperforms Ge2Sb2Te5 by two orders of magnitude and paves the way for on‐chip programmable phase control. These truly low‐loss switchable materials open up new directions in programmable integrated photonic circuits, switchable metasurfaces, and nanophotonic devices.

show abstract

Linear programmable nanophotonic processors

Cited by 311 publications

References 78 publications

Photonic architecture for reinforcement learning

Photonic architecture for reinforcement learning

Photonic convolutional neural networks using integrated diffractive optics

A New Family of Ultralow Loss Reversible Phase‐Change Materials for Photonic Integrated Circuits: Sb₂S₃ and Sb₂Se₃

Contact Info

Product

Resources

About

Linear programmable nanophotonic processors

Cited by 311 publications

References 78 publications

Photonic architecture for reinforcement learning

Photonic architecture for reinforcement learning

Photonic convolutional neural networks using integrated diffractive optics

A New Family of Ultralow Loss Reversible Phase‐Change Materials for Photonic Integrated Circuits: Sb2S3 and Sb2Se3

Contact Info

Product

Resources

About

A New Family of Ultralow Loss Reversible Phase‐Change Materials for Photonic Integrated Circuits: Sb₂S₃ and Sb₂Se₃