Towards Optical Neural Networks with Fabrication Noise Immunity

Fang, Michael Y.-S.; Manipatruni, Sasikanth; Wierzynski, Casimir; Khosrowshahi, Amir

doi:10.1364/cleo_si.2019.sf1n.6

Cited by 1 publication

(7 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition, several researchers have developed different formats of Optical Neural Networks (ONN), including the diffraction based and interferometer based networks [16] [17]. In particular, the paper by Fang has reported successful implementation of ONNs utilizing Mach-Zehnder Interferometers (MZI).…”

Section: B Prior Researchmentioning

confidence: 99%

“…These networks encoded the linear transformations typical of perceptrons through differently structured arrays of MZIs. We have used the models and codes used in this paper for benchmarking purposes, and built further on the structure reported [17] [18].…”

Section: B Prior Researchmentioning

confidence: 99%

“…We pre-processed the MNIST image data for use with the quantum models by superimposing the top half of the image over the bottom half through a translation step in order to generate 392 complex numbers, which are also single photon Fock states, as will be explained later [17]. Specifically, if a is the matrix of pixel values, then a x,y is the real part and a x+14,y is the complex part of each complex number for x ≤ 14 and y ≤ 28.…”

Section: A Datasetmentioning

confidence: 99%

“…As another standard for benchmarking comparisons, we used the optical neural networks presented by Fang et al [17]. We used the libraries published in this work as the basis for our model and implemented them in the PyTorch framework [18].…”

Section: Optical Neural Network (Onn)mentioning

confidence: 99%

“…In the figure, the first unitary, diagonal, second unitary, and amplification in each layer are denoted by V , Σ, U, and β , respectively. Finally, the layers before the end are capped of with a ModSoftplus layer, which is a discrete sigmoidal nonlinearity, and the last layer is ended with a ModSquared discrete nonlinearity and a LogSoftmax, which is a smooth sigmoid that produces the final predictions [17].…”

Section: Optical Neural Network (Onn)mentioning

confidence: 99%

See 4 more Smart Citations

Quantum Optical Convolutional Neural Network: A Novel Image Recognition Framework for Quantum Computing

Parthasarathy,

Bhowmik

2020

Preprint

View full text Add to dashboard Cite

Large machine learning models based on Convolutional Neural Networks (CNNs) with rapidly increasing number of parameters, trained with massive amounts of data, are being deployed in a wide array of computer vision tasks from self-driving cars to medical imaging. The insatiable demand for computing resources required to train these models is fast outpacing the advancement of classical computing hardware, and new frameworks including Optical Neural Networks (ONNs) and quantum computing are being explored as future alternatives.In this work, we report a novel quantum computing based deep learning model, the Quantum Optical Convolutional Neural Network (QOCNN), to alleviate the computational bottleneck in future computer vision applications. Using the popular MNIST dataset, we have benchmarked this new architecture against a traditional CNN based on the seminal LeNet model. We have also compared the performance with previously reported ONNs, namely the GridNet and ComplexNet, as well as a Quantum Optical Neural Network (QONN) that we built by combining the ComplexNet with quantum based sinusoidal nonlinearities. In essence, our work extends the prior research on QONN by adding quantum convolution and pooling layers preceding it.We have evaluated all the models by determining their accuracies, confusion matrices, Receiver Operating Characteristic (ROC) curves, and Matthews Correlation Coefficients. The performance of the models were similar overall, and the ROC curves indicated that the new QOCNN model is robust. Finally, we estimated the gains in computational efficiencies from executing this novel framework on a quantum computer. We conclude that switching to a quantum computing based approach to deep learning may result in comparable accuracies to classical models, while achieving unprecedented boosts in computational performances and drastic reduction in power consumption.

show abstract

Section: B Prior Researchmentioning

confidence: 99%

Section: B Prior Researchmentioning

confidence: 99%

Section: A Datasetmentioning

confidence: 99%

Section: Optical Neural Network (Onn)mentioning

confidence: 99%

Section: Optical Neural Network (Onn)mentioning

confidence: 99%

See 3 more Smart Citations

Quantum Optical Convolutional Neural Network: A Novel Image Recognition Framework for Quantum Computing

Parthasarathy,

Bhowmik

2020

Preprint

View full text Add to dashboard Cite

show abstract

Towards Optical Neural Networks with Fabrication Noise Immunity

Cited by 1 publication

References 0 publications

Quantum Optical Convolutional Neural Network: A Novel Image Recognition Framework for Quantum Computing

Quantum Optical Convolutional Neural Network: A Novel Image Recognition Framework for Quantum Computing

Contact Info

Product

Resources

About