Abstract:A critical feature in signal processing is the ability to interpret correlations in time series signals, such as speech. Machine learning systems process this contextual information by tracking internal states in recurrent neural networks (RNNs), but these can cause memory and processor bottlenecks in applications from edge devices to data centers, motivating research into new analog inference architectures. But whereas photonic accelerators, in particular, have demonstrated big leaps in uni-directional feedforward deep neural network (DNN) inference, the bi-directional architecture of RNNs presents a unique challenge: the need for a short-term memory that (i) programmably transforms optical waveforms with phase coherence , (ii) minimizes added noise, and (iii) enables programmable readily scales to large neuron counts. Here, we address this challenge by introducing an optoacoustic recurrent operator (OREO) that simultaneously meets (i,ii,iii). Specifically, we experimentally demonstrate an OREO that contextualizes and computes the information carried by a sequence of optical pulses via acoustic waves. We show that the acoustic waves act as a link between the different optical pulses, capturing the optical information and using it to manipulate the subsequent operations. Our approach can be controlled completely optically on a pulse-by-pulse basis, offering simple reconfigurability for a use case-specific optimization. We use this feature to demonstrate a recurrent drop-out, which excludes optical input pulses from the recurrent operation. We furthermore apply OREO as an acceptor to recognize up-to $27$ patterns in a sequence of optical pulses. Finally, we introduce a DNN architecture that uses the OREO as bi-directional perceptrons to enable new classes of DNNs in coherent optical signal processing.

Optical Rectifying Linear Units for Back-Propagation Learning in a Deep Holographic Convolutional Neural Network

The Realization Of Parallel Subtraction Using Lclv In Optical Neural Network

Optical Analog Computing Devices Designed by Deep Neural Network

Programmable low-threshold optical nonlinear activation functions for photonic neural networks

Optical Implementation of Neural Learning Algorithms Based on Cross-gain Modulation in a Semiconductor Optical Amplifier

Reconfigurable All-Optical Integrated Nonlinear Activator with Switchable Response Functions for Photonic Neural Networks

All-Chalcogenide Programmable All-Optical Deep Neural Networks

Deep learning with coherent VCSEL neural networks

Training neural networks with end-to-end optical backpropagation

An optoacoustic field-programmable perceptron for recurrent neural networks

Unitary learning for diffractive deep neural network

Unitary Learning for Deep Diffractive Neural Network

<i>In situ</i> optical backpropagation training of diffractive optical neural networks

Optical coherent dot-product chip for sophisticated deep learning regression

Integrated Photonic FFT for Optical Convolutions towards Efficient and High-Speed Neural Networks

In Situ Optical Backpropagation Training of Diffractive Optical Neural Networks

Low-power scalable multilayer optoelectronic neural networks enabled with incoherent light

Backpropagation through nonlinear units for all-optical training of neural networks

Deep Learning with Passive Optical Nonlinear Mapping

Non-volatile Reconfigurable Digital Optical Diffractive Neural Network Based on Phase Change Material

11 TeraFLOPs per second photonic convolutional accelerator for deep learning optical neural networks