Liane Bernstein
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Liane Bernstein has a total of 39 co-authors. Here are the top co-authors by number of shared publications:
- Alexander Sludds
Massachusetts Institute of Technology
8 shared publications - Dirk Englund
Massachusetts Institute of Technology, Cambridge, USA
8 shared publications - Ryan Hamerly
Research Laboratory of Electronics, MIT, Cambridge, USA
7 shared publications - Zaijun Chen
Research Laboratory of Electronics, MIT, Cambridge, MA 02139, USA
4 shared publications - Saumil Bandyopadhyay
Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge, MA 02139, USA
3 shared publications - Zhizhen Zhong
Massachusetts Institute of Technology
3 shared publications
Latest Publications
- 27 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (22) , 7795-7806
Netcast: Low-Power Edge Computing With WDM-Defined Optical Neural Networks
This paper analyzes the performance and energy efficiency of Netcast, a recently proposed optical neural-network architecture designed for edge computing. Netcast performs deep neural network inference by dividing the computational task into two steps, which are split between the (cloud) server and (edge) client: (1) the server employs a wavelength-multiplexed modulator array to encode the network's weights onto an optical signal in an analog time-wavelength basis, and (2) the client obtains the desired matrix-vector product through modulation and time-integrated detection. The simultaneous use of wavelength multiplexing, broadband modulation, and integration detection allows large neural networks to be run at the client by effectively pushing the energy and memory requirements back to the server. The performance and energy efficiency are fundamentally limited by crosstalk and detector noise, respectively. We derive analytic expressions for these limits and perform numerical simulations to verify these bounds.
- 27 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (22) , 7795-7806
Netcast: Low-Power Edge Computing With WDM-Defined Optical Neural Networks
This paper analyzes the performance and energy efficiency of Netcast, a recently proposed optical neural-network architecture designed for edge computing. Netcast performs deep neural network inference by dividing the computational task into two steps, which are split between the (cloud) server and (edge) client: (1) the server employs a wavelength-multiplexed modulator array to encode the network's weights onto an optical signal in an analog time-wavelength basis, and (2) the client obtains the desired matrix-vector product through modulation and time-integrated detection. The simultaneous use of wavelength multiplexing, broadband modulation, and integration detection allows large neural networks to be run at the client by effectively pushing the energy and memory requirements back to the server. The performance and energy efficiency are fundamentally limited by crosstalk and detector noise, respectively. We derive analytic expressions for these limits and perform numerical simulations to verify these bounds.
- Top Cited Papers
- 17 July 2023
- Published by Springer Nature in Nature Photonics
- Vol. 17 (8) , 723-730
Deep learning with coherent VCSEL neural networks
- 17 July 2023
- Published by Springer Nature in Nature Photonics
- Vol. 17 (8) , 723-730
Top Cited PapersDeep learning with coherent VCSEL neural networks
- Open Access
- 23 June 2023
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 9 (25) , eadg7904
Single-shot optical neural network
Analog optical and electronic hardware has emerged as a promising alternative to digital electronics to improve the efficiency of deep neural networks (DNNs). However, previous work has been limited in scalability (input vector lengthK≈ 100 elements) or has required nonstandard DNN models and retraining, hindering widespread adoption. Here, we present an analog, CMOS–compatible DNN processor that uses free-space optics to reconfigurably distribute an input vector and optoelectronics for static, updatable weighting and the nonlinearity—withK≈ 1000 and beyond. We demonstrate single-shot-per-layer classification of the MNIST, Fashion-MNIST, and QuickDraw datasets with standard fully connected DNNs, achieving respective accuracies of 95.6, 83.3, and 79.0% without preprocessing or retraining. We also experimentally determine the fundamental upper bound on throughput (∼0.9 exaMAC/s), set by the maximum optical bandwidth before substantial increase in error. Our combination of wide spectral and spatial bandwidths enables highly efficient computing for next-generation DNNs.
- 23 June 2023
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 9 (25) , eadg7904
Open AccessSingle-shot optical neural network
Analog optical and electronic hardware has emerged as a promising alternative to digital electronics to improve the efficiency of deep neural networks (DNNs). However, previous work has been limited in scalability (input vector lengthK≈ 100 elements) or has required nonstandard DNN models and retraining, hindering widespread adoption. Here, we present an analog, CMOS–compatible DNN processor that uses free-space optics to reconfigurably distribute an input vector and optoelectronics for static, updatable weighting and the nonlinearity—withK≈ 1000 and beyond. We demonstrate single-shot-per-layer classification of the MNIST, Fashion-MNIST, and QuickDraw datasets with standard fully connected DNNs, achieving respective accuracies of 95.6, 83.3, and 79.0% without preprocessing or retraining. We also experimentally determine the fundamental upper bound on throughput (∼0.9 exaMAC/s), set by the maximum optical bandwidth before substantial increase in error. Our combination of wide spectral and spatial bandwidths enables highly efficient computing for next-generation DNNs.
- Top Cited Papers
- 21 October 2022
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 378 (6617) , 270-276
Delocalized photonic deep learning on the internet’s edge
Advanced machine learning models are currently impossible to run on edge devices such as smart sensors and unmanned aerial vehicles owing to constraints on power, processing, and memory. We introduce an approach to machine learning inference based on delocalized analog processing across networks. In this approach, named Netcast, cloud-based “smart transceivers” stream weight data to edge devices, enabling ultraefficient photonic inference. We demonstrate image recognition at ultralow optical energy of 40 attojoules per multiply (100 watts) cloud computers.
- 21 October 2022
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 378 (6617) , 270-276
Top Cited PapersDelocalized photonic deep learning on the internet’s edge
Advanced machine learning models are currently impossible to run on edge devices such as smart sensors and unmanned aerial vehicles owing to constraints on power, processing, and memory. We introduce an approach to machine learning inference based on delocalized analog processing across networks. In this approach, named Netcast, cloud-based “smart transceivers” stream weight data to edge devices, enabling ultraefficient photonic inference. We demonstrate image recognition at ultralow optical energy of 40 attojoules per multiply (100 watts) cloud computers.
- Open Access
Single-Shot Optical Neural Network
As deep neural networks (DNNs) grow to solve increasingly complex problems, they are becoming limited by the latency and power consumption of existing digital processors. For improved speed and energy efficiency, specialized analog optical and electronic hardware has been proposed, however, with limited scalability (input vector length $K$ of hundreds of elements). Here, we present a scalable, single-shot-per-layer analog optical processor that uses free-space optics to reconfigurably distribute an input vector and integrated optoelectronics for static, updatable weighting and the nonlinearity -- with $K \approx 1,000$ and beyond. We experimentally test classification accuracy of the MNIST handwritten digit dataset, achieving 94.7% (ground truth 96.3%) without data preprocessing or retraining on the hardware. We also determine the fundamental upper bound on throughput ($\sim$0.9 exaMAC/s), set by the maximum optical bandwidth before significant increase in error. Our combination of wide spectral and spatial bandwidths in a CMOS-compatible system enables highly efficient computing for next-generation DNNs.Open AccessSingle-Shot Optical Neural Network
As deep neural networks (DNNs) grow to solve increasingly complex problems, they are becoming limited by the latency and power consumption of existing digital processors. For improved speed and energy efficiency, specialized analog optical and electronic hardware has been proposed, however, with limited scalability (input vector length $K$ of hundreds of elements). Here, we present a scalable, single-shot-per-layer analog optical processor that uses free-space optics to reconfigurably distribute an input vector and integrated optoelectronics for static, updatable weighting and the nonlinearity -- with $K \approx 1,000$ and beyond. We experimentally test classification accuracy of the MNIST handwritten digit dataset, achieving 94.7% (ground truth 96.3%) without data preprocessing or retraining on the hardware. We also determine the fundamental upper bound on throughput ($\sim$0.9 exaMAC/s), set by the maximum optical bandwidth before significant increase in error. Our combination of wide spectral and spatial bandwidths in a CMOS-compatible system enables highly efficient computing for next-generation DNNs. - Open Access
Delocalized Photonic Deep Learning on the Internet's Edge
Advances in deep neural networks (DNNs) are transforming science and technology. However, the increasing computational demands of the most powerful DNNs limit deployment on low-power devices, such as smartphones and sensors -- and this trend is accelerated by the simultaneous move towards Internet-of-Things (IoT) devices. Numerous efforts are underway to lower power consumption, but a fundamental bottleneck remains due to energy consumption in matrix algebra, even for analog approaches including neuromorphic, analog memory and photonic meshes. Here we introduce and demonstrate a new approach that sharply reduces energy required for matrix algebra by doing away with weight memory access on edge devices, enabling orders of magnitude energy and latency reduction. At the core of our approach is a new concept that decentralizes the DNN for delocalized, optically accelerated matrix algebra on edge devices. Using a silicon photonic smart transceiver, we demonstrate experimentally that this scheme, termed Netcast, dramatically reduces energy consumption. We demonstrate operation in a photon-starved environment with 40 aJ/multiply of optical energy for 98.8% accurate image recognition and <1 photon/multiply using single photon detectors. Furthermore, we show realistic deployment of our system, classifying images with 3 THz of bandwidth over 86 km of deployed optical fiber in a Boston-area fiber network. Our approach enables computing on a new generation of edge devices with speeds comparable to modern digital electronics and power consumption that is orders of magnitude lower.
Open AccessDelocalized Photonic Deep Learning on the Internet's Edge
Advances in deep neural networks (DNNs) are transforming science and technology. However, the increasing computational demands of the most powerful DNNs limit deployment on low-power devices, such as smartphones and sensors -- and this trend is accelerated by the simultaneous move towards Internet-of-Things (IoT) devices. Numerous efforts are underway to lower power consumption, but a fundamental bottleneck remains due to energy consumption in matrix algebra, even for analog approaches including neuromorphic, analog memory and photonic meshes. Here we introduce and demonstrate a new approach that sharply reduces energy required for matrix algebra by doing away with weight memory access on edge devices, enabling orders of magnitude energy and latency reduction. At the core of our approach is a new concept that decentralizes the DNN for delocalized, optically accelerated matrix algebra on edge devices. Using a silicon photonic smart transceiver, we demonstrate experimentally that this scheme, termed Netcast, dramatically reduces energy consumption. We demonstrate operation in a photon-starved environment with 40 aJ/multiply of optical energy for 98.8% accurate image recognition and <1 photon/multiply using single photon detectors. Furthermore, we show realistic deployment of our system, classifying images with 3 THz of bandwidth over 86 km of deployed optical fiber in a Boston-area fiber network. Our approach enables computing on a new generation of edge devices with speeds comparable to modern digital electronics and power consumption that is orders of magnitude lower.
- 8 March 2022
- Published by Optica Publishing Group in Biomedical Optics Express
- Vol. 13 (4) , 1939-1947
Ultrahigh resolution spectral-domain optical coherence tomography using the 1000–1600 nm spectral band
Ultrahigh resolution optical coherence tomography (UHR-OCT) can image microscopic features that are not visible with the standard OCT resolution of 5-15 µm. In previous studies, high-speed UHR-OCT has been accomplished within the visible (VIS) and near-infrared (NIR-I) spectral ranges, specifically within 550-950 nm. Here, we present a spectral domain UHR-OCT system operating in a short-wavelength infrared (SWIR) range from 1000 to 1600 nm using a supercontinuum light source and an InGaAs-based spectrometer. We obtained an axial resolution of 2.6 µm in air, the highest ever recorded in the SWIR window to our knowledge, with deeper penetration into tissues than VIS or NIR-I light. We demonstrate imaging of conduction fibers of the left bundle branch in freshly excised porcine hearts. These results suggest a potential for deep-penetration, ultrahigh resolution OCT in intraoperative applications.
- 8 March 2022
- Published by Optica Publishing Group in Biomedical Optics Express
- Vol. 13 (4) , 1939-1947
Ultrahigh resolution spectral-domain optical coherence tomography using the 1000–1600 nm spectral band
Ultrahigh resolution optical coherence tomography (UHR-OCT) can image microscopic features that are not visible with the standard OCT resolution of 5-15 µm. In previous studies, high-speed UHR-OCT has been accomplished within the visible (VIS) and near-infrared (NIR-I) spectral ranges, specifically within 550-950 nm. Here, we present a spectral domain UHR-OCT system operating in a short-wavelength infrared (SWIR) range from 1000 to 1600 nm using a supercontinuum light source and an InGaAs-based spectrometer. We obtained an axial resolution of 2.6 µm in air, the highest ever recorded in the SWIR window to our knowledge, with deeper penetration into tissues than VIS or NIR-I light. We demonstrate imaging of conduction fibers of the left bundle branch in freshly excised porcine hearts. These results suggest a potential for deep-penetration, ultrahigh resolution OCT in intraoperative applications.
- 1 January 2022
- Published by Optica Publishing Group
- Vol. 1, STh5G.7
Scalable Ultralow Latency Photonic Tensor Processor
We propose a photonic processor with the potential to compute deep neural network layer outputs in ∼10 ns. Our proof-of-concept experiments show low loss of classification accuracy on the MNIST dataset; we investigate fundamental throughput limits.
- 1 January 2022
- Published by Optica Publishing Group
- Vol. 1, STh5G.7
Scalable Ultralow Latency Photonic Tensor Processor
We propose a photonic processor with the potential to compute deep neural network layer outputs in ∼10 ns. Our proof-of-concept experiments show low loss of classification accuracy on the MNIST dataset; we investigate fundamental throughput limits.
- Open Access
- 4 February 2021
- Published by Springer Nature in Scientific Reports
- Vol. 11 (1) , 1-12
Freely scalable and reconfigurable optical hardware for deep learning
As deep neural network (DNN) models grow ever-larger, they can achieve higher accuracy and solve more complex problems. This trend has been enabled by an increase in available compute power; however, efforts to continue to scale electronic processors are impeded by the costs of communication, thermal management, power delivery and clocking. To improve scalability, we propose a digital optical neural network (DONN) with intralayer optical interconnects and reconfigurable input values. The path-length-independence of optical energy consumption enables information locality between a transmitter and a large number of arbitrarily arranged receivers, which allows greater flexibility in architecture design to circumvent scaling limitations. In a proof-of-concept experiment, we demonstrate optical multicast in the classification of 500 MNIST images with a 3-layer, fully-connected network. We also analyze the energy consumption of the DONN and find that digital optical data transfer is beneficial over electronics when the spacing of computational units is on the order of $$>10\,\upmu $$ m.- 4 February 2021
- Published by Springer Nature in Scientific Reports
- Vol. 11 (1) , 1-12
Open AccessFreely scalable and reconfigurable optical hardware for deep learning
As deep neural network (DNN) models grow ever-larger, they can achieve higher accuracy and solve more complex problems. This trend has been enabled by an increase in available compute power; however, efforts to continue to scale electronic processors are impeded by the costs of communication, thermal management, power delivery and clocking. To improve scalability, we propose a digital optical neural network (DONN) with intralayer optical interconnects and reconfigurable input values. The path-length-independence of optical energy consumption enables information locality between a transmitter and a large number of arbitrarily arranged receivers, which allows greater flexibility in architecture design to circumvent scaling limitations. In a proof-of-concept experiment, we demonstrate optical multicast in the classification of 500 MNIST images with a 3-layer, fully-connected network. We also analyze the energy consumption of the DONN and find that digital optical data transfer is beneficial over electronics when the spacing of computational units is on the order of $$>10\,\upmu $$ m.