scholarly journals Optical neuromorphic processing with soliton crystal Kerr microcombs: Scaling the network in size and speed

2022 ◽  
Author(s):  
David Moss
Keyword(s):  
Neural Networks ◽  
Facial Image ◽  
Board Games ◽  
Learning Tasks ◽  
Optical Neural Networks ◽  
Significant Interest ◽  
Spatial Dimensions ◽  
Handwritten Digit ◽  
Output Neurons ◽  
Hierarchical Features

Convolutional neural networks (CNNs), inspired by biological visual cortex systems, are a powerful category of artificial neural networks that can extract the hierarchical features of raw data to greatly reduce the network parametric complexity and enhance the predicting accuracy. They are of significant interest for machine learning tasks such as computer vision, speech recognition, playing board games and medical diagnosis [1-7]. Optical neural networks offer the promise of dramatically accelerating computing speed to overcome the inherent bandwidth bottleneck of electronics. Here, we demonstrate a universal optical vector convolutional accelerator operating beyond 10 Tera-OPS (TOPS - operations per second), generating convolutions of images of 250,000 pixels with 8-bit resolution for 10 kernels simultaneously — enough for facial image recognition. We then use the same hardware to sequentially form a deep optical CNN with ten output neurons, achieving successful recognition of full 10 digits with 900 pixel handwritten digit images with 88% accuracy. Our results are based on simultaneously interleaving temporal, wavelength and spatial dimensions enabled by an integrated microcomb source. We show that this approach is scalable and trainable to much more complex networks for demanding applications such as unmanned vehicle and real-time video recognition.Keywords: Optical neural networks, neuromorphic processor, microcomb, convolutional accelerator

scholarly journals 11.0 Tera-FLOP/second photonic convolutional accelerator for deep learning optical neural networks

2020 ◽  
Author(s):  
David Moss
Keyword(s):  
Neural Networks ◽  
Board Games ◽  
Learning Tasks ◽  
Optical Neural Networks ◽  
Significant Interest ◽  
Spatial Dimensions ◽  
Video Recognition ◽  
Handwritten Digit ◽  
Output Neurons ◽  
Hierarchical Features

Convolutional neural networks (CNNs), inspired by biological visual cortex systems, are a powerful category of artificial neural networks that can extract the hierarchical features of raw data to greatly reduce the network parametric complexity and enhance the predicting accuracy. They are of significant interest for machine learning tasks such as computer vision, speech recognition, playing board games and medical diagnosis [1-7]. Optical neural networks offer the promise of dramatically accelerating computing speed to overcome the inherent bandwidth bottleneck of electronics. Here, we demonstrate a universal optical vector convolutional accelerator operating beyond 10 Tera-FLOPS (floating point operations per second), generating convolutions of images of 250,000 pixels with 8-bit resolution for 10 kernels simultaneously — enough for facial image recognition. We then use the same hardware to sequentially form a deep optical CNN with ten output neurons, achieving successful recognition of full 10 digits with 900 pixel handwritten digit images with 88% accuracy. Our results are based on simultaneously interleaving temporal, wavelength and spatial dimensions enabled by an integrated microcomb source. This approach is scalable and trainable to much more complex networks for demanding applications such as unmanned vehicle and real-time video recognition.

scholarly journals Neuromorphic processing at Tera-OP/s speeds with soliton crystal microcombs

2021 ◽  
Author(s):  
Mengxi Tan ◽  
Xingyuan Xu ◽  
David Moss
Keyword(s):  
Neural Networks ◽  
Board Games ◽  
Learning Tasks ◽  
Optical Neural Networks ◽  
Significant Interest ◽  
Spatial Dimensions ◽  
Video Recognition ◽  
Handwritten Digit ◽  
Output Neurons ◽  
Hierarchical Features

Abstract Convolutional neural networks (CNNs), inspired by biological visual cortex systems, are a powerful category of artificial neural networks that can extract the hierarchical features of raw data to greatly reduce the network parametric complexity and enhance the predicting accuracy. They are of significant interest for machine learning tasks such as computer vision, speech recognition, playing board games and medical diagnosis [1-7]. Optical neural networks offer the promise of dramatically accelerating computing speed to overcome the inherent bandwidth bottleneck of electronics. Here, we demonstrate a universal optical vector convolutional accelerator operating beyond 10 Tera-OPS (TOPS - operations per second), generating convolutions of images of 250,000 pixels with 8-bit resolution for 10 kernels simultaneously — enough for facial image recognition. We then use the same hardware to sequentially form a deep optical CNN with ten output neurons, achieving successful recognition of full 10 digits with 900 pixel handwritten digit images with 88% accuracy. Our results are based on simultaneously interleaving temporal, wavelength and spatial dimensions enabled by an integrated microcomb source. This approach is scalable and trainable to much more complex networks for demanding applications such as unmanned vehicle and real-time video recognition.

scholarly journals Neuromorphic processing at Tera-OP/s with soliton crystal microcombs

2021 ◽  
Author(s):  
David Moss
Keyword(s):  
Neural Networks ◽  
Board Games ◽  
Learning Tasks ◽  
Optical Neural Networks ◽  
Significant Interest ◽  
Spatial Dimensions ◽  
Video Recognition ◽  
Handwritten Digit ◽  
Output Neurons ◽  
Hierarchical Features

Convolutional neural networks (CNNs), inspired by biological visual cortex systems, are a powerful category of artificial neural networks that can extract the hierarchical features of raw data to greatly reduce the network parametric complexity and enhance the predicting accuracy. They are of significant interest for machine learning tasks such as computer vision, speech recognition, playing board games and medical diagnosis [1-7]. Optical neural networks offer the promise of dramatically accelerating computing speed to overcome the inherent bandwidth bottleneck of electronics. Here, we demonstrate a universal optical vector convolutional accelerator operating beyond 10 Tera-OPS (TOPS - operations per second), generating convolutions of images of 250,000 pixels with 8-bit resolution for 10 kernels simultaneously — enough for facial image recognition. We then use the same hardware to sequentially form a deep optical CNN with ten output neurons, achieving successful recognition of full 10 digits with 900 pixel handwritten digit images with 88% accuracy. Our results are based on simultaneously interleaving temporal, wavelength and spatial dimensions enabled by an integrated microcomb source. This approach is scalable and trainable to much more complex networks for demanding applications such as unmanned vehicle and real-time video recognition.

scholarly journals 11 Tera-FLOP/s photonic convolutional accelerator for optical neural networks

2021 ◽  
Author(s):  
David Moss
Keyword(s):  
Neural Networks ◽  
Board Games ◽  
Learning Tasks ◽  
Optical Neural Networks ◽  
Significant Interest ◽  
Spatial Dimensions ◽  
Video Recognition ◽  
Handwritten Digit ◽  
Output Neurons ◽  
Hierarchical Features

Convolutional neural networks (CNNs), inspired by biological visual cortex systems, are a powerful category of artificial neural networks that can extract the hierarchical features of raw data to greatly reduce the network parametric complexity and enhance the predicting accuracy. They are of significant interest for machine learning tasks such as computer vision, speech recognition, playing board games and medical diagnosis [1-7]. Optical neural networks offer the promise of dramatically accelerating computing speed to overcome the inherent bandwidth bottleneck of electronics. Here, we demonstrate a universal optical vector convolutional accelerator operating beyond 10 Tera-FLOPS (floating point operations per second), generating convolutions of images of 250,000 pixels with 8-bit resolution for 10 kernels simultaneously — enough for facial image recognition. We then use the same hardware to sequentially form a deep optical CNN with ten output neurons, achieving successful recognition of full 10 digits with 900 pixel handwritten digit images with 88% accuracy. Our results are based on simultaneously interleaving temporal, wavelength and spatial dimensions enabled by an integrated microcomb source. This approach is scalable and trainable to much more complex networks for demanding applications such as unmanned vehicle and real-time video recognition.

scholarly journals 11.0 Tera-FLOP/second photonic convolutional accelerator for deep learning optical neural networks

2020 ◽  
Author(s):  
David Moss
Keyword(s):  
Neural Networks ◽  
Board Games ◽  
Learning Tasks ◽  
Optical Neural Networks ◽  
Significant Interest ◽  
Spatial Dimensions ◽  
Video Recognition ◽  
Handwritten Digit ◽  
Output Neurons ◽  
Hierarchical Features

Convolutional neural networks (CNNs), inspired by biological visual cortex systems, are a powerful category of artificial neural networks that can extract the hierarchical features of raw data to greatly reduce the network parametric complexity and enhance the predicting accuracy. They are of significant interest for machine learning tasks such as computer vision, speech recognition, playing board games and medical diagnosis [1-7]. Optical neural networks offer the promise of dramatically accelerating computing speed to overcome the inherent bandwidth bottleneck of electronics. Here, we demonstrate a universal optical vector convolutional accelerator operating beyond 10 Tera-FLOPS (floating point operations per second), generating convolutions of images of 250,000 pixels with 8-bit resolution for 10 kernels simultaneously — enough for facial image recognition. We then use the same hardware to sequentially form a deep optical CNN with ten output neurons, achieving successful recognition of full 10 digits with 900 pixel handwritten digit images with 88% accuracy. Our results are based on simultaneously interleaving temporal, wavelength and spatial dimensions enabled by an integrated microcomb source. This approach is scalable and trainable to much more complex networks for demanding applications such as unmanned vehicle and real-time video recognition.

scholarly journals Photonic convolutional accelerator operating at Tera-OPs speeds for neural networks with Kerr microcombs

2021 ◽  
Author(s):  
David Moss
Keyword(s):  
Neural Networks ◽  
Board Games ◽  
Learning Tasks ◽  
Optical Neural Networks ◽  
Significant Interest ◽  
Spatial Dimensions ◽  
Video Recognition ◽  
Handwritten Digit ◽  
Output Neurons ◽  
Hierarchical Features

<p>Convolutional neural networks (CNNs), inspired by biological visual cortex systems, are a powerful category of artificial neural networks that can extract the hierarchical features of raw data to greatly reduce the network parametric complexity and enhance the predicting accuracy. They are of significant interest for machine learning tasks such as computer vision, speech recognition, playing board games and medical diagnosis [1-7]. Optical neural networks offer the promise of dramatically accelerating computing speed to overcome the inherent bandwidth bottleneck of electronics. Here, we demonstrate a universal optical vector convolutional accelerator operating beyond 10 Tera-OPS (TOPS - operations per second), generating convolutions of images of 250,000 pixels with 8-bit resolution for 10 kernels simultaneously — enough for facial image recognition. We then use the same hardware to sequentially form a deep optical CNN with ten output neurons, achieving successful recognition of full 10 digits with 900 pixel handwritten digit images with 88% accuracy. Our results are based on simultaneously interleaving temporal, wavelength and spatial dimensions enabled by an integrated microcomb source. This approach is scalable and trainable to much more complex networks for demanding applications such as unmanned vehicle and real-time video recognition. <i> </i></p>

scholarly journals Photonic convolutional accelerator operating at Tera-OPs speeds for neural networks with Kerr microcombs

2021 ◽  
Author(s):  
David Moss
Keyword(s):  
Neural Networks ◽  
Board Games ◽  
Learning Tasks ◽  
Optical Neural Networks ◽  
Significant Interest ◽  
Spatial Dimensions ◽  
Video Recognition ◽  
Handwritten Digit ◽  
Output Neurons ◽  
Hierarchical Features

<p>Convolutional neural networks (CNNs), inspired by biological visual cortex systems, are a powerful category of artificial neural networks that can extract the hierarchical features of raw data to greatly reduce the network parametric complexity and enhance the predicting accuracy. They are of significant interest for machine learning tasks such as computer vision, speech recognition, playing board games and medical diagnosis [1-7]. Optical neural networks offer the promise of dramatically accelerating computing speed to overcome the inherent bandwidth bottleneck of electronics. Here, we demonstrate a universal optical vector convolutional accelerator operating beyond 10 Tera-OPS (TOPS - operations per second), generating convolutions of images of 250,000 pixels with 8-bit resolution for 10 kernels simultaneously — enough for facial image recognition. We then use the same hardware to sequentially form a deep optical CNN with ten output neurons, achieving successful recognition of full 10 digits with 900 pixel handwritten digit images with 88% accuracy. Our results are based on simultaneously interleaving temporal, wavelength and spatial dimensions enabled by an integrated microcomb source. This approach is scalable and trainable to much more complex networks for demanding applications such as unmanned vehicle and real-time video recognition. <i> </i></p>

scholarly journals Tera-OPs Photonic Convolutional Neural NetworksBased on Kerr Microcombs

2021 ◽  
Author(s):  
Mengxi Tan ◽  
Mike Xu ◽  
David Moss
Keyword(s):  
Neural Networks ◽  
Board Games ◽  
Learning Tasks ◽  
Optical Neural Networks ◽  
Significant Interest ◽  
Spatial Dimensions ◽  
Video Recognition ◽  
Handwritten Digit ◽  
Output Neurons ◽  
Hierarchical Features

Convolutional neural networks (CNNs), inspired by biological visual cortex systems, are a powerful category of artificial neural networks that can extract the hierarchical features of raw data to greatly reduce the network parametric complexity and enhance the predicting accuracy. They are of significant interest for machine learning tasks such as computer vision, speech recognition, playing board games and medical diagnosis [1-7]. Optical neural networks offer the promise of dramatically accelerating computing speed to overcome the inherent bandwidth bottleneck of electronics. Here, we demonstrate a universal optical vector convolutional accelerator operating beyond 10 Tera-OPS (TOPS - operations per second), generating convolutions of images of 250,000 pixels with 8-bit resolution for 10 kernels simultaneously &mdash; enough for facial image recognition. We then use the same hardware to sequentially form a deep optical CNN with ten output neurons, achieving successful recognition of full 10 digits with 900 pixel handwritten digit images with 88% accuracy. Our results are based on simultaneously interleaving temporal, wavelength and spatial dimensions enabled by an integrated microcomb source. This approach is scalable and trainable to much more complex networks for demanding applications such as unmanned vehicle and real-time video recognition.

scholarly journals Ultrahigh bandwidth applications of microcombs: Neural networks and optical data transmission

2021 ◽  
Author(s):  
David Moss
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Data Transmission ◽  
Medical Diagnosis ◽  
Optical Data ◽  
Board Games ◽  
Learning Tasks ◽  
Optical Neural Networks ◽  
Data Rates ◽  
Hierarchical Features

<div>We report ultrahigh bandwidth applications of Kerr microcombs to optical neural networks and to optical data transmission, at data rates from 44 Terabits/s (Tb/s) to approaching 100 Tb/s. Convolutional neural networks (CNNs) are a powerful category of artificial neural networks that can extract the hierarchical features of raw data to greatly reduce the network complexity and enhance the accuracy for machine learning tasks such as computer vision, speech recognition, playing board games and medical diagnosis [1-7].</div>

scholarly journals Microcombs for ultrahigh bandwidth optical data transmission and neural networks

2021 ◽  
Author(s):  
David Moss
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Data Transmission ◽  
Medical Diagnosis ◽  
Optical Data ◽  
Board Games ◽  
Learning Tasks ◽  
Optical Neural Networks ◽  
Data Rates ◽  
Hierarchical Features

We report ultrahigh bandwidth applications of Kerr microcombs to optical neural networks and to optical datatransmission, at data rates from 44 Terabits/s (Tb/s) to approaching 100 Tb/s. Convolutional neural networks (CNNs) area powerful category of artificial neural networks that can extract the hierarchical features of raw data to greatly reducethe network complexity and enhance the accuracy for machine learning tasks such as computer vision, speechrecognition, playing board games and medical diagnosis [1-7].

Sign in / Sign up

Export Citation Format

Share Document