Deep learning has profoundly reshaped the technology landscape in numerous scientific areas and industrial sectors. This technology advancement is, nevertheless, confronted with severe bottlenecks in digital computing. Optical neural network presents a promising solution due to the ultra-high computing speed and energy efficiency. In this work, we present systematic study of unitary optical neural network (UONN) as an approach towards optical deep learning. Our results show that the UONN can be trained to high accuracy through special unitary gradient descent optimization, and the UONN is robust against physical imperfections and noises, hence it is more suitable for physical implementation than existing ONNs.
Deep learning has profoundly reshaped the technology landscape in numerous scientific areas and industrial sectors. This technology advancement is, nevertheless, confronted with severe bottlenecks in digital computing. Optical neural network presents a promising solution due to the ultra-high computing speed and energy efficiency. In this work, we present systematic study of unitary optical neural network (UONN) as an approach towards optical deep learning. Our results show that the UONN can be trained to high accuracy through special unitary gradient descent optimization, and the UONN is robust against physical imperfections and noises, hence it is more suitable for physical implementation than existing ONNs.
This paper explores the application of the parameter-shift rule (PSR) for computing gradients in unitary optical neural networks (UONNs). While backpropagation has been fundamental to training conventional neural networks, its implementation in optical neural networks faces significant challenges due to the physical constraints of optical systems. We demonstrate how PSR, which calculates gradients by evaluating functions at shifted parameter values, can be effectively adapted for training UONNs constructed from Mach-Zehnder interferometer meshes. We develop a theoretical framework for two distinct scenarios, real-valued outputs and complex-valued outputs of optical networks, which derives exact analytical gradients directly from hardware measurements for phase parameters in optical neural networks. For the first real-valued scenario, the gradient calculation directly conforms to PSR. For the second complex-valued scenario, we improved PSR to obtain the gradient. Then we simulate the training of optical neural networks based on optical PSR and both scenarios complete the classification tasks well. The optical PSR offers a powerful and analytically precise approach for calculating gradients in unitary optical neural networks. This approach offers a promising alternative to traditional in silico training methods and circumvents the limitations of both finite difference approximations and all-optical backpropagation implementations. We present the theoretical framework and practical methodology for applying PSR to optimize phase parameters in optical neural networks, potentially advancing the development of efficient hardware-based training strategies for optical computing systems.
Large machine learning models based on Convolutional Neural Networks (CNNs) with rapidly increasing numbers of parameters are being deployed in a wide array of computer vision tasks. However, 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 MNIST dataset, we have benchmarked this new architecture against a traditional CNN model based on the seminal LeNet architecture. We have also compared the performance against previously reported ONNs, the GridNet and ComplexNet, and a Quantum Optical Neural Network (QONN) that we built by combining the ComplexNet with quantum-inspired sinusoidal nonlinearities. Our work extends the prior research on QONN by adding quantum convolution and pooling layers. We have evaluated the models using the metrics of accuracies, confusion matrices, Receiver Operating Characteristic curves, and Matthews Correlation Coefficients. The performance of the models were similar, and the metrics 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, concluding that switching to a quantum computing based approach to deep learning may result in comparable accuracies to classical models while simultaneously achieving unprecedented boosts in computational performances and power allocation.
Owing to the high integration, reconfiguration and strong robustness, Mach-Zehnder interferometers (MZIs) based optical neural networks (ONNs) have been widely considered. However, there are few works adding bias, which is important for neural networks, into the ONNs and systematically studying its effect. In this article, we propose a tunable-bias based optical neural network (TBONN) with one unitary matrix layer, which can improve the utilization rate of the MZIs, increase the trainable weights of the network and has more powerful representational capacity than traditional ONNs. By systematically studying its underlying mechanism and characteristics, we demonstrate that TBONN can achieve higher performance by adding more optical biases to the same side beside the inputted signals. For the two-dimensional dataset, the average prediction accuracy of TBONN with 2 biases (97.1%) is 5% higher than that of TBONN with 0 biases (92.1%). Additionally, utilizing TBONN, we propose a novel optical deep Q network (ODQN) algorithm to complete path planning tasks. By implementing simulated experiments, our ODQN shows competitive performance compared with the conventional deep Q network, but accelerates the computation speed by 2.5 times and 4.5 times for 2D and 3D grid worlds, respectively. Further, a more noticeable acceleration will be obtained when applying TBONN to more complex tasks. Also, we demonstrate the strong robustness of TBONN and the imprecision elimination method by using on-chip training.
Neural networks enjoy widespread success in both research and industry and, with the advent of quantum technology, it is a crucial challenge to design quantum neural networks for fully quantum learning tasks. Here we propose a truly quantum analogue of classical neurons, which form quantum feedforward neural networks capable of universal quantum computation. We describe the efficient training of these networks using the fidelity as a cost function, providing both classical and efficient quantum implementations. Our method allows for fast optimisation with reduced memory requirements: the number of qudits required scales with only the width, allowing deep-network optimisation. We benchmark our proposal for the quantum task of learning an unknown unitary and find remarkable generalisation behaviour and a striking robustness to noisy training data.
Optical neural networks (ONNs) have been developed to enhance processing speed and energy efficiency in machine learning by leveraging optical devices for nonlinear activation and establishing connections among neurons. In this work, we propose a quantum optical neural network (QONN) that utilizes atom-cavity neurons with controllable photon absorption and emission. These quantum neurons are designed to replace the electronic components in ONNs, which typically introduce delays and substantial energy consumption during nonlinear activation. To evaluate the performance of the QONN, we apply it to the MNIST digit classification task, considering the effects of photon absorption duration, random atom-cavity detuning, and stochastic photon loss. Additionally, we introduce a convolutional QONN to facilitate a real-world satellite image classification (SAT-6) task. Due to its compact hardware and low power consumption, the QONN offers a promising solution for real-time satellite sensing, reducing communication bandwidth with ground stations and thereby enhancing data security.
Hybrid quantum neural networks represent a novel approach to enhancing computational performance in deep learning tasks. By integrating quantum-inspired techniques with classical neural network architectures, these models aim to leverage the computational advantages of quantum computing while retaining the scalability and interpretability of classical methods. This paper explores the design and implementation of hybrid quantum neural networks, focusing on their ability to optimize computational efficiency in deep learning tasks. We present a comprehensive review of quantum-inspired optimization techniques and their application in neural network training. Through empirical evaluations and comparative studies, we demonstrate the effectiveness of hybrid quantum neural networks in improving training speed, convergence, and generalization performance. This work contributes to the advancement of efficient deep learning methodologies and lays the groundwork for further exploration of quantum-inspired approaches in machine learning. DOI - https://doi.org/10.65525/SVUP.9788199778009.2026.231-234