Optical learning neurochip with internal analog memory

Graphene/MoS2-xOx/graphene photomemristor with tunable non-volatile responsivities for neuromorphic vision processing

Conventional artificial intelligence (AI) machine vision technology, based on the von Neumann architecture, uses separate sensing, computing, and storage units to process huge amounts of vision data generated in sensory terminals. The frequent movement of redundant data between sensors, processors and memory, however, results in high-power consumption and latency. A more efficient approach is to offload some of the memory and computational tasks to sensor elements that can perceive and process the optical signal simultaneously. Here, we proposed a non-volatile photomemristor, in which the reconfigurable responsivity can be modulated by the charge and/or photon flux through it and further stored in the device. The non-volatile photomemristor has a simple two-terminal architecture, in which photoexcited carriers and oxygen-related ions are coupled, leading to a displaced and pinched hysteresis in the current-voltage characteristics. For the first time, non-volatile photomemristors implement computationally complete logic with photoresponse-stateful operations, for which the same photomemristor serves as both a logic gate and memory, using photoresponse as a physical state variable instead of light, voltage and memresistance. The polarity reversal of photomemristors shows great potential for in-memory sensing and computing with feature extraction and image recognition for neuromorphic vision.

Practical optoelectronic neural network realizations based on the fault tolerance of the backpropagation algorithm

This paper describes how the fault tolerance of the backpropagation algorithm can be used to accommodate the realistic (nonideal) transfer characteristics of the optical communication links used, between neural layers, in optoelectronic neural networks. In particular the authors demonstrate that networks, utilizing MSM (metal-semiconductor-metal) photodiodes (PDs) and either LED (light emitting diode) or MQW (multiple quantum well) laser transmitters within these intraneural links, are capable of performing satisfactorily even in the presence of such nonideal device phenomena as: 60% optical crosstalk, 50% optoelectronic device variation, or a thresholded (I(th)=0.5*I(max)) laser output characteristic. Subsequent to this, the authors then show how it is possible to use this fault tolerance to simplify the neuron architecture, to the extent that it consists only of MSM PDs a current amplifier, and an MQW laser. The overall neuron transfer function is then a first-order approximation to the original sigmoidal function.

Neural network using longitudinal modes of an injection laser with external feedback

A new optical neural-network concept using the control of the modes of an injection laser by external feedback is described by a simple laser model. This approach uses the wavelength dispersed longitudinal modes of the laser as neurons and the amount of external feedback as connection weights. The predictions of the simple model are confirmed both with extensive numerical examples using the laser rate equations and also by experiments with GaAlAs injection lasers. The inputs and connection weights to this laser neural network are provided by external masks which control the amount of feedback reaching the laser. Stochastic learning is used to obtain weight masks for a small three-input and four-output neural net for the numerical and experimental examples. Winner-take-all and exclusive-or operations are obtained on the input set with different weight masks. Both of these operations are also obtained in experiments with a three-input/four-output laser neural network operating at an estimated speed greater than 10 GCPS. The eventual speed of this type of neural network hardware is expected to reach well within TCPS range if it is built in an optoelectronic integrated circuit with dimensions in the order of a mm. Different neural-network architectures possible with this approach are discussed.

Introduction to the Bioelectronic Devices Project in Japan

The Bioelectronic Devices Project was organized in Japan as a 10-year national project and it has been working toward developing fundamental key technologies for designing and assembling innovative information-processing devices by realizing the excellent functions specifically found in molecular assemblies and information processing of living organisms. The project is now in the third year of the second phase (the 8th year of its 10-year duration) and researchers have been trying to elucidate the specific characters of the prototype devices. The outline of this challenging project is given with the latest experimental results.

Optical implementation of a translation-invariant second-order neural network for multiple-pattern classification

A novel approach to the optical implementation of second-order neural networks that can recognize multiple patterns is reported. The systems issues, especially the accuracy required for the weighted interconnections, are discussed for numeric character (0-9) recognition. It is shown that the accuracy of the weighted interconnections has a far greater influence on the network performance during training than on classification. To lessen the problem, we introduce an adaptive learning rule, whereby the optical power is adjusted during training. Finally, numeric character recognition using an experimental system with a liquid-crystal display is demonstrated.