Surrogate gradient learning in spiking networks trained on event-based cytometry dataset.

Spiking neural networks (SNNs) are bio-inspired neural networks that - to an extent - mimic the workings of our brains. In a similar fashion, event-based vision sensors try to replicate a biological eye as closely as possible. In this work, we integrate both technologies for the purpose of classifying micro-particles in the context of label-free flow cytometry. We follow up on our previous work in which we used simple logistic regression with binary labels. Although this model was able to achieve an accuracy of over 98%, our goal is to utilize the system for a wider variety of cells, some of which may have less noticeable morphological variations. Therefore, a more advanced machine learning model like the SNNs discussed here would be required. This comes with the challenge of training such networks, since they typically suffer from vanishing gradients. We effectively apply the surrogate gradient method to overcome this issue achieving over 99% classification accuracy on test data for a four-class problem. Finally, rather than treating the neural network as a black box, we explore the dynamics inside the network and make use of that to enhance its accuracy and sparsity.

Enhancing the performance of coherent Ising machines in the large-noise regime with a fifth-order nonlinearity.

Coherent Ising machines (CIMs), leveraging the bistable physical properties of coherent light to emulate Ising spins, exhibit great potential as hardware accelerators for tackling complex combinatorial optimization problems. Recent advances have demonstrated that the performance of CIMs can be enhanced either by incorporating large random noise or higher-order nonlinearities, yet their combined effects on CIM performance remain mainly unexplored. In this work, we develop a numerical CIM model that utilizes a tunable fifth-order polynomial nonlinear dynamic function under large noise levels, which has the potential to be implemented in all-optical platforms. We propose a normal form of a CIM model that allows for both supercritical and subcritical pitchfork bifurcation operational regimes, with fifth-order nonlinearity and tunable hyperparameters to control the Ising spin dynamics. In the benchmark studies, we simulate various sets of MaxCut problems using our fifth-order polynomial CIM model. The results show a significant performance improvement, achieving an average of 59.5% improvement in median time-to-solution (TTS) and an average of 6 times improvement in median success rate (SR) for dense Maxcut problems in the BiqMac library, compared to the commonly used third-order polynomial CIM model with low noise. The fifth-order polynomial CIM model in the large-noise regime also shows better performance trends as the problem size scales up. These findings reveal the enhancements on the computational performance of Ising machines in the large-nose regime from fifth-order nonlinearity, showing important implications for both simulation and hardware perspectives.

A recurrent Gaussian quantum network for online processing of quantum time series.

Over the last decade, researchers have studied the interplay between quantum computing and classical machine learning algorithms. However, measurements often disturb or destroy quantum states, requiring multiple repetitions of data processing to estimate observable values. In particular, this prevents online (real-time, single-shot) processing of temporal data as measurements are commonly performed during intermediate stages. Recently, it was proposed to sidestep this issue by focusing on tasks with quantum output, eliminating the need for detectors. Inspired by reservoir computers, a model was proposed where only a subset of the internal parameters are trained while keeping the others fixed at random values. Here, we also process quantum time series, but we do so using a Recurrent Gaussian Quantum Network (RGQN) of which all internal interactions can be trained. As expected, this increased flexibility yields higher performance in benchmark tasks. Building on this, we show that the RGQN can tackle two quantum communication tasks, while also removing some hardware restrictions of the currently available methods. First, our approach is more resource efficient to enhance the transmission rate of quantum channels that experience certain memory effects. Second, it can counteract similar memory effects if they are unwanted, a task that could previously only be solved when redundantly encoded input signals could be provided. Finally, we run a small-scale version of the last task on Xanadu's photonic processor Borealis.

Emerging opportunities and challenges for the future of reservoir computing.

Reservoir computing originates in the early 2000s, the core idea being to utilize dynamical systems as reservoirs (nonlinear generalizations of standard bases) to adaptively learn spatiotemporal features and hidden patterns in complex time series. Shown to have the potential of achieving higher-precision prediction in chaotic systems, those pioneering works led to a great amount of interest and follow-ups in the community of nonlinear dynamics and complex systems. To unlock the full capabilities of reservoir computing towards a fast, lightweight, and significantly more interpretable learning framework for temporal dynamical systems, substantially more research is needed. This Perspective intends to elucidate the parallel progress of mathematical theory, algorithm design and experimental realizations of reservoir computing, and identify emerging opportunities as well as existing challenges for large-scale industrial adoption of reservoir computing, together with a few ideas and viewpoints on how some of those challenges might be resolved with joint efforts by academic and industrial researchers across multiple disciplines.

Experimental results on nonlinear distortion compensation using photonic reservoir computing with a single set of weights for different wavelengths.

Photonics-based computing approaches in combination with wavelength division multiplexing offer a potential solution to modern data and bandwidth needs. This paper experimentally takes an important step towards wavelength division multiplexing in an integrated waveguide-based photonic reservoir computing platform by using a single set of readout weights for up to at least 3 ITU-T channels to efficiently scale the data bandwidth when processing a nonlinear signal equalization task on a 28 Gbps modulated on-off keying signal. Using multiple-wavelength training, we obtain bit error rates well below that of the [Formula: see text] forward error correction limit at high fiber input powers of 18 dBm, which result in high nonlinear distortion. The results of the reservoir chip are compared to a tapped delay line filter and clearly show that the system performs nonlinear equalization. This was achieved using only limited post processing which in future work can be implemented in optical hardware as well.

Threshold plasticity of SOI-GST microring resonators.

Spiking Neural Networks, also known as third generation Artificial Neural Networks, have widely attracted more attention because of their advantages of behaving more biologically interpretable and being more suitable for hardware implementation. Apart from using traditional synaptic plasticity, neural networks can also be based on threshold plasticity, achieving similar functionality. This can be implemented using e.g. the Bienenstock, Cooper and Munro rule. This is a classical unsupervised learning mechanism in which the threshold is closely related to the output of the post-synaptic neuron. We show in simulations that the threshold characteristics of the nonlinear effects of a microring resonator integrated with Ge2Sb2Te5 demonstrate some complex dependencies on the intracavity refractive index, attenuation, and wavelength detuning of the incident optical pulse, and exhibit class II excitability. We also show that we are able to modify the threshold power of the microring resonator by the changes of the refractive index and loss of Ge2Sb2Te5, due to transitions between the crystalline and amorphous states. Simulations show that the presented device exhibits both excitatory and inhibitory learning behavior, either lowering or raising the threshold.

Integrated photonic reservoir computing with an all-optical readout.

Integrated photonic reservoir computing has been demonstrated to be able to tackle different problems because of its neural network nature. A key advantage of photonic reservoir computing over other neuromorphic paradigms is its straightforward readout system, which facilitates both rapid training and robust, fabrication variation-insensitive photonic integrated hardware implementation for real-time processing. We present our recent development of a fully-optical, coherent photonic reservoir chip integrated with an optical readout system, capitalizing on these benefits. Alongside the integrated system, we also demonstrate a weight update strategy that is suitable for the integrated optical readout hardware. Using this online training scheme, we successfully solved 3-bit header recognition and delayed XOR tasks at 20 Gbps in real-time, all within the optical domain without excess delays.

Programmable low-power consumption all-optical nonlinear activation functions using a micro-ring resonator with phase-change materials.

A programmable hardware implementation of all-optical nonlinear activation functions for different scenarios and applications in all-optical neural networks is essential. We demonstrate a programmable, low-loss all-optical activation function device based on a silicon micro-ring resonator loaded with phase change materials. Four different nonlinear activation functions of Relu, ELU, Softplus and radial basis functions are implemented for incident signal light of the same wavelength. The maximum power consumption required to switch between the four different nonlinear activation functions in calculation is only 1.748 nJ. The simulation of classification of hand-written digit images also shows that they can perform well as alternative nonlinear activation functions. The device we design can serve as nonlinear units in photonic neural networks, while its nonlinear transfer function can be flexibly programmed to optimize the performance of different neuromorphic tasks.

Rigorous dynamic model of a silicon ring resonator with phase change material for a neuromorphic node.

The photonics platform has been considered increasingly promising for neuromorphic computing, due to its potential in providing low latency and energy efficient large-scale parallel connectivity. Phase change materials (PCMs) have been recently employed to introduce all-optical non-volatile memory in integrated photonic circuits, especially finding application as non-volatile weighting element in photonic artificial neural networks. Interestingly, these weighting elements can potentially be used as building blocks for large-scale networks that can autonomously adapt to their input, i.e. presenting the property of plasticity, similarly to the biological brain. In this work, we develop a computationally efficient dynamical model of a silicon ring resonator (RR) enhanced by a phase change material, namely Ge2Sb2Te5 (GST). We do so starting from two existing dynamical models (of a silicon RR and of a GST thin film on a straight silicon waveguide), but extending the optical equations to properly account for the high absorption and asymmetry in the ring due to the phase change material. Our model accounts for silicon nonlinear effects due to free carriers and temperature, as well as for the phase change of GST, whose energy efficiency and optical contrast can be enhanced by the RR resonant behaviour. We also restructure the optical equations so that the model can be efficiently employed in a modular way within a commercial software for system-level photonics simulations. Moreover, exploiting the developed model, we explore several design parameters and show that both speed and energy efficiency of memory operations can be enhanced by factors from six to ten. Also, we show that the achievable optical contrast due to GST phase change can be increased by more than a factor ten by leveraging the resonant properties of the RR, at the expense of higher optical loss. Finally, by exploiting the nonlinear dynamics arising in silicon RR networks, we show that a strong contrast is achievable while preserving energy efficiency.

Influence of the input signal's phase modulation on the performance of optical delay-based reservoir computing using semiconductor lasers.

In photonic reservoir computing, semiconductor lasers with delayed feedback have shown to be suited to efficiently solve difficult and time-consuming problems. The input data in this system is often optically injected into the reservoir. Based on numerical simulations, we show that the performance depends heavily on the way that information is encoded in this optical injection signal. In our simulations we compare different input configurations consisting of Mach-Zehnder modulators and phase modulators for injecting the signal. We observe far better performance on a one-step ahead time-series prediction task when modulating the phase of the injected signal rather than only modulating its amplitude.

Wavelength dimension in waveguide-based photonic reservoir computing.

Existing work on coherent photonic reservoir computing (PRC) mostly concentrates on single-wavelength solutions. In this paper, we discuss the opportunities and challenges related to exploiting the wavelength dimension in integrated photonic reservoir computing systems. Different strategies are presented to be able to process several wavelengths in parallel using the same readout. Additionally, we present multiwavelength training techniques that allow to increase the stable operating wavelength range by at least a factor of two. It is shown that a single-readout photonic reservoir system can perform with ≈0% BER on several WDM channels in parallel for bit-level tasks and nonlinear signal equalization. This even when taking manufacturing deviations and laser wavelength drift into account.

Comparing different nonlinearities in readout systems for optical neuromorphic computing networks.

Nonlinear activation is a crucial building block of most machine-learning systems. However, unlike in the digital electrical domain, applying a saturating nonlinear function in a neural network in the analog optical domain is not as easy, especially in integrated systems. In this paper, we first investigate in detail the photodetector nonlinearity in two main readout schemes: electrical readout and optical readout. On a 3-bit-delayed XOR task, we show that optical readout trained with backpropagation gives the best performance. Furthermore, we propose an additional saturating nonlinearity coming from a deliberately non-ideal voltage amplifier after the detector. Compared to an all-optical nonlinearity, these two kinds of nonlinearities are extremely easy to obtain at no additional cost, since photodiodes and voltage amplifiers are present in any system. Moreover, not having to design ideal linear amplifiers could relax their design requirements. We show through simulation that for long-distance nonlinear fiber distortion compensation, using only the photodiode nonlinearity in an optical readout delivers BER improvements over three orders of magnitude. Combined with the amplifier saturation nonlinearity, we obtain another three orders of magnitude improvement of the BER.

Experimental realization of integrated photonic reservoir computing for nonlinear fiber distortion compensation.

Nonlinearity mitigation in optical fiber networks is typically handled by electronic Digital Signal Processing (DSP) chips. Such DSP chips are costly, power-hungry and can introduce high latencies. Therefore, optical techniques are investigated which are more efficient in both power consumption and processing cost. One such a machine learning technique is optical reservoir computing, in which a photonic chip can be trained on certain tasks, with the potential advantages of higher speed, reduced power consumption and lower latency compared to its electronic counterparts. In this paper, experimental results are presented where nonlinear distortions in a 32 GBPS OOK signal are mitigated to below the 0.2 × 10-3 FEC limit using a photonic reservoir. Furthermore, the results of the reservoir chip are compared to a tapped delay line filter to clearly show that the system performs nonlinear equalisation.

Addressing limited weight resolution in a fully optical neuromorphic reservoir computing readout.

Using optical hardware for neuromorphic computing has become more and more popular recently, due to its efficient high-speed data processing capabilities and low power consumption. However, there are still some remaining obstacles to realizing the vision of a completely optical neuromorphic computer. One of them is that, depending on the technology used, optical weighting elements may not share the same resolution as in the electrical domain. Moreover, noise of the weighting elements are important considerations as well. In this article, we investigate a new method for improving the performance of optical weighting components, even in the presence of noise and in the case of very low resolution. Our method utilizes an iterative training procedure and is able to select weight connections that are more robust to quantization and noise. As a result, even with only 8 to 32 levels of resolution, in noisy weighting environments, the method can outperform both nearest rounding low-resolution weighting and random rounding weighting by up to several orders of magnitude in terms of bit error rate and can deliver performance very close to full-resolution weighting elements.

Simulating self-learning in photorefractive optical reservoir computers.

Photorefractive materials exhibit an interesting plasticity under the influence of an optical field. By extending the finite-difference time-domain method to include the photorefractive effect, we explore how this property can be exploited in the context of neuromorphic computing for telecom applications. By first priming the photorefractive material with a random bit stream, the material reorganizes itself to better recognize simple patterns in the stream. We demonstrate this by simulating a typical reservoir computing setup, which gets a significant performance boost on performing the XOR on two consecutive bits in the stream after this initial priming step.

Machine learning issues and opportunities in ultrafast particle classification for label-free microflow cytometry.

Machine learning offers promising solutions for high-throughput single-particle analysis in label-free imaging microflow cytomtery. However, the throughput of online operations such as cell sorting is often limited by the large computational cost of the image analysis while offline operations may require the storage of an exceedingly large amount of data. Moreover, the training of machine learning systems can be easily biased by slight drifts of the measurement conditions, giving rise to a significant but difficult to detect degradation of the learned operations. We propose a simple and versatile machine learning approach to perform microparticle classification at an extremely low computational cost, showing good generalization over large variations in particle position. We present proof-of-principle classification of interference patterns projected by flowing transparent PMMA microbeads with diameters of [Formula: see text] and [Formula: see text]. To this end, a simple, cheap and compact label-free microflow cytometer is employed. We also discuss in detail the detection and prevention of machine learning bias in training and testing due to slight drifts of the measurement conditions. Moreover, we investigate the implications of modifying the projected particle pattern by means of a diffraction grating, in the context of optical extreme learning machine implementations.

A training algorithm for networks of high-variability reservoirs.

Physical reservoir computing approaches have gained increased attention in recent years due to their potential for low-energy high-performance computing. Despite recent successes, there are bounds to what one can achieve simply by making physical reservoirs larger. Therefore, we argue that a switch from single-reservoir computing to multi-reservoir and even deep physical reservoir computing is desirable. Given that error backpropagation cannot be used directly to train a large class of multi-reservoir systems, we propose an alternative framework that combines the power of backpropagation with the speed and simplicity of classic training algorithms. In this work we report our findings on a conducted experiment to evaluate the general feasibility of our approach. We train a network of 3 Echo State Networks to perform the well-known NARMA-10 task, where we use intermediate targets derived through backpropagation. Our results indicate that our proposed method is well-suited to train multi-reservoir systems in an efficient way.

Demonstrating delay-based reservoir computing using a compact photonic integrated chip.

Photonic delay-based reservoir computing (RC) has gained considerable attention lately, as it allows for simple technological implementations of the RC concept that can operate at high speed. In this paper, we discuss a practical, compact and robust implementation of photonic delay-based RC, by integrating a laser and a 5.4 cm delay line on an InP photonic integrated circuit. We demonstrate the operation of this chip with 23 nodes at a speed of 0.87 GSa/s, showing performances that is similar to previous non-integrated delay-based setups. We also investigate two other post-processing methods to obtain more nodes in the output layer. We show that these methods improve the performance drastically, without compromising the computation speed.

Highly parallel simulation and optimization of photonic circuits in time and frequency domain based on the deep-learning framework PyTorch.

We propose a new method for performing photonic circuit simulations based on the scatter matrix formalism. We leverage the popular deep-learning framework PyTorch to reimagine photonic circuits as sparsely connected complex-valued neural networks. This allows for highly parallel simulation of large photonic circuits on graphical processing units in time and frequency domain while all parameters of each individual component can easily be optimized with well-established machine learning algorithms such as backpropagation.

Study on the limit of detection in MZI-based biosensor systems.

Mach-Zehnder interferometers are integrated photonic sensors that have yielded excellent detection limits down to 10-7 RIU. They are of particular interest due to their large design freedom, allowing for example application in promising point-of-care compatible read-out schemes. The attainable detection limit of such sensors can interact with the sensor design in different ways, depending on the dominant origin of noise which can either be influencing a single sensor arm, both sensor arms or can be unrelated to the sensor itself. In this work, the interaction of these three noise regimes with the sensor design is examined. The regimes are combined into a framework that predicts the limit of detection as a function of sensor design. A set of experimental results confirms the validity of this obtained theoretical framework. This analysis provides a blueprint for optimization of MZI photonic sensors under any combination of read-out method and measurement circumstances.