Demonstration of a silicon photonic neural network

Tait, Alexander N.; Zhou, Ellen; Wu, Allie X.; Nahmias, Mitchell A.; Lima, Thomas Ferreira de; Shastri, Bhavin J.; Prucnal, Paul R.

doi:10.1109/phosst.2016.7548726

Cited by 4 publications

(4 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[41]. Figure 4E shows a modulator with continuous nonlinearity, the first demonstration of which in a PNN and recurrent network is presented in [42]. The pros and cons between the schemes in Figure 4D and E are the same ones brought up by the on-chip vs. off-chip light source debate, currently under way in the silicon photonics community.…”

Section: O/e/o Pnnsmentioning

confidence: 81%

Progress in neuromorphic photonics

et al. 2017

Self Cite

View full text Add to dashboard Cite

As society's appetite for information continues to grow, so does our need to process this information with increasing speed and versatility. Many believe that the one-size-fits-all solution of digital electronics is becoming a limiting factor in certain areas such as data links, cognitive radio, and ultrafast control. Analog photonic devices have found relatively simple signal processing niches where electronics can no longer provide sufficient speed and reconfigurability. Recently, the landscape for commercially manufacturable photonic chips has been changing rapidly and now promises to achieve economies of scale previously enjoyed solely by microelectronics. By bridging the mathematical prowess of artificial neural networks to the underlying physics of optoelectronic devices, neuromorphic photonics could breach new domains of information processing demanding significant complexity, low cost, and unmatched speed. In this article, we review the progress in neuromorphic photonics, focusing on photonic integrated devices. The challenges and design rules for optoelectronic instantiation of artificial neurons are presented. The proposed photonic architecture revolves around the processing network node composed of two parts: a nonlinear element and a network interface. We then survey excitable lasers in the recent literature as candidates for the nonlinear node and microring-resonator weight banks as the network interface. Finally, we compare metrics between neuromorphic electronics and neuromorphic photonics and discuss potential applications.

show abstract

Section: O/e/o Pnnsmentioning

confidence: 81%

Progress in neuromorphic photonics

et al. 2017

Self Cite

View full text Add to dashboard Cite

show abstract

“…Such networks would be even more useful in applications requiring high-speed nonlinear computation. The first proof of concept for an OEO neural network came from Princeton [6] and demonstrated neuromorphic dynamics. An outline of the network can be seen in Figure 3.…”

Section: Motivationmentioning

confidence: 99%

“…In [6]: # Initialization calCurrentChan = np.array( [3,4,5,6]) Create an empty calibration model to be filled, using the ascribed channels for validation purposes.…”

Section: B Code Appendicesmentioning

confidence: 99%

Design and Control of a Photonic Neural Network Applied to High-Bandwidth Classification

Gordon

2018

Preprint

View full text Add to dashboard Cite

Neural networks can very effectively perform multidimensional nonlinear classification. However, electronic networks suffer from significant bandwidth limitations due to carrier lifetimes and capacitive coupling. This project investigates photonic neural networks that can get around these limitations by performing both the activation function and weighted addition in the optical domain using microring resonators. These optical microring resonators provide both nonlinearity and superior fan-in without compromising bandwidth. The ability to thermally calibrate networks of cascaded axons and dendrites and train such a network to solve nonlinear classification problems are demonstrated using theory and simulations. The former is also demonstrated experimentally on a two-channel axon cascaded into a two-channel dendrite, showing good agreement between simulation and experiment. In addition, the use of transverse modes to increase the size of each photonic layer is examined. Simulations that determined the optimal waveguide geometry for using these modes were experimentally validated.First and foremost, I would like to express my deepest gratitude to Alex Tait 'GS, my primary day-to-day mentor and advisor. Without him, absolutely none of this work would have been possible. I would also like to extend my thanks to the entire Lightwave Communications Laboratory for their advice, support, and infrastructure. This project was also made possible by generous funding from the School of Enginering and Applied Science and the Department of Electrical Engineering. Finally, I would like to acknowledge all of my friends and peers: the fighting ELE Class of 2017, off which I could always bounce ideas, and my friends and roommates, who were willing to listen to my ramblings and help edit this document. Thank you to everybody for helping make this a success.This work is dedicated to the loving memory of my parents, Lauren and David Gordon.

show abstract

“…In addition, significant research activities on the implementation of optical communications between different types of optical spiking neurons were reported in [24,25]. Following the first demonstration of the integration of a small all-optical NN in silicon [26], more studies of implementations of all-optical synapses [27][28][29] have been published. Because adaptability is an important feature of NNs, the main goal of the design of optical synapses is to model the biological learning rules.…”

Section: Introductionmentioning

confidence: 99%

Optical Axons for Electro-Optical Neural Networks

Hulea

Ghassemlooy

Rajbhandari

et al. 2020

Sensors

View full text Add to dashboard Cite

Recently, neuromorphic sensors, which convert analogue signals to spiking frequencies, have been reported for neurorobotics. In bio-inspired systems these sensors are connected to the main neural unit to perform post-processing of the sensor data. The performance of spiking neural networks has been improved using optical synapses, which offer parallel communications between the distanced neural areas but are sensitive to the intensity variations of the optical signal. For systems with several neuromorphic sensors, which are connected optically to the main unit, the use of optical synapses is not an advantage. To address this, in this paper we propose and experimentally verify optical axons with synapses activated optically using digital signals. The synaptic weights are encoded by the energy of the stimuli, which are then optically transmitted independently. We show that the optical intensity fluctuations and link’s misalignment result in delay in activation of the synapses. For the proposed optical axon, we have demonstrated line of sight transmission over a maximum link length of 190 cm with a delay of 8 μs. Furthermore, we show the axon delay as a function of the illuminance using a fitted model for which the root mean square error (RMS) similarity is 0.95.

show abstract

Demonstration of a silicon photonic neural network

Cited by 4 publications

References 5 publications

Progress in neuromorphic photonics

Progress in neuromorphic photonics

Design and Control of a Photonic Neural Network Applied to High-Bandwidth Classification

Optical Axons for Electro-Optical Neural Networks

Contact Info

Product

Resources

About