Search

Collecting time series data spatially distributed in many locations is often important for analyzing climate change and its impacts on ecosystems. However, comprehensive spatial data collection is not always feasible, requiring us to predict climate variables at some locations. This study focuses on a prediction of climatic elements, specifically near-surface temperature and pressure, at a target location apart from a data observation point. Our approach uses two prediction methods: reservoir computing (RC), known as a machine learning framework with low computational requirements, and vector autoregression models (VAR), recognized as a statistical method for analyzing time series data. Our results show that the accuracy of the predictions degrades with the distance between the observation and target locations. We quantitatively estimate the distance in which effective predictions are possible. We also find that in the context of climate data, a geographical distance is associated with data correlation, and a strong data correlation significantly improves the prediction accuracy with RC. In particular, RC outperforms VAR in predicting highly correlated data within the predictive range. These findings suggest that machine learning-based methods can be used more effectively to predict climatic elements in remote locations by assessing the distance to them from the data observation point in advance. Our study on low-cost and accurate prediction of climate variables has significant value for climate change strategies.

Dynamical behaviors of complex interacting systems, including brain activities, financial price movements, and physical collective phenomena, are associated with underlying interactions between the system's components. The issue of uncovering interaction relations in such systems using observable dynamics is called relational inference. In this study, we propose a Diffusion model for Relational Inference (DiffRI), inspired by a self-supervised method for probabilistic time series imputation. DiffRI learns to infer the probability of the presence of connections between components through conditional diffusion modeling.

Accurate epidemic forecasting is a critical task in controlling disease transmission. Many deep learning-based models focus only on static or dynamic graphs when constructing spatial information, ignoring their relationship. Additionally, these models often rely on recurrent structures, which can lead to error accumulation and computational time consumption. To address the aforementioned problems, we propose a novel model called Backbone-based Dynamic Graph Spatio-Temporal Network (BDGSTN). Intuitively, the continuous and smooth changes in graph structure, make adjacent graph structures share a basic pattern. To capture this property, we use adaptive methods to generate static backbone graphs containing the primary information and temporal models to generate dynamic temporal graphs of epidemic data, fusing them to generate a backbone-based dynamic graph. To overcome potential limitations associated with recurrent structures, we introduce a linear model DLinear to handle temporal dependencies and combine it with dynamic graph convolution for epidemic forecasting. Extensive experiments on two datasets demonstrate that BDGSTN outperforms baseline models and ablation comparison further verifies the effectiveness of model components. Furthermore, we analyze and measure the significance of backbone and temporal graphs by using information metrics from different aspects. Finally, we compare model parameter volume and training time to confirm the superior complexity and efficiency of BDGSTN.

The Dissemination Process Classification (DPC) is a popular application of temporal graph classification. The aim of DPC is to classify different spreading patterns of information or pestilence within a community represented by discrete-time temporal graphs. Recently, a reservoir computing-based model named Dynamical Graph Echo State Network (DynGESN) has been proposed for processing temporal graphs with relatively high effectiveness and low computational costs. In this study, we propose a novel model which combines a novel data augmentation strategy called snapshot merging with the DynGESN for dealing with DPC tasks. In our model, the snapshot merging strategy is designed for forming new snapshots by merging neighboring snapshots over time, and then multiple reservoir encoders are set for capturing spatiotemporal features from merged snapshots. After those, the logistic regression is adopted for decoding the sum-pooled embeddings into the classification results. Experimental results on six benchmark DPC datasets show that our proposed model has better classification performances than the DynGESN and several kernel-based models.

The authors have numerically studied how to enhance reservoir computing performance by thoroughly extracting their spin-wave device potential for higher-dimensional information generation. The reservoir device has a 1-input exciter and 120-output detectors on the top of a continuous magnetic garnet film for spin-wave transmission. For various nonlinear and fading-memory dynamic phenomena distributing in the film space, small in-plane magnetic fields were used to prepare stripe domain structures and various damping constants at the film sides and bottom were explored. The ferromagnetic resonant frequency and relaxation time of spin precession clearly characterized the change in spin dynamics with the magnetic field and damping constant. The common input signal for reservoir computing was a 1 GHz cosine wave with random 6-valued amplitude modulation. A basic 120-dimensional reservoir output vector was obtained from time-series signals at the 120 output detectors under each of the three magnetic field conditions. Then, 240- and 360-dimensional reservoir output vectors were also constructed by concatenating two and three basic ones, respectively. In nonlinear autoregressive moving average (NARMA) prediction tasks, the computational performance was enhanced as the dimension of the reservoir output vector becomes higher and a significantly low prediction error was achieved for the 10th-order NARMA using the 360-dimensional vector and optimum damping constant. The results are clear evidence that the collection of diverse output signals efficiently increases the dimensionality effective for reservoir computing, i.e., reservoir-state richness. This paper demonstrates that performance enhancement through various configuration settings is a practical approach for on-chip reservoir computing devices with small numbers of real output nodes.

Memristive systems and devices are potentially available for implementing reservoir computing (RC) systems applied to pattern recognition. However, the computational ability of memristive RC systems depends on intertwined factors such as system architectures and physical properties of memristive elements, which complicates identifying the key factor for system performance. Here we develop a simulation platform for RC with memristor device networks, which enables testing different system designs for performance improvement. Numerical simulations show that the memristor-network-based RC systems can yield high computational performance comparable to that of state-of-the-art methods in three time series classification tasks. We demonstrate that the excellent and robust computation under device-to-device variability can be achieved by appropriately setting network structures, nonlinearity of memristors, and pre/post-processing, which increases the potential for reliable computation with unreliable component devices. Our results contribute to an establishment of a design guide for memristive reservoirs toward a realization of energy-efficient machine learning hardware., Comment: 14 pages, 7 figures, 5 supplementary figures

Machine learning approaches have recently been leveraged as a substitute or an aid for physical/mathematical modeling approaches to dynamical systems. To develop an efficient machine learning method dedicated to modeling and prediction of multiscale dynamics, we propose a reservoir computing (RC) model with diverse timescales by using a recurrent network of heterogeneous leaky integrator (LI) neurons. We evaluate computational performance of the proposed model in two time series prediction tasks related to four chaotic fast-slow dynamical systems. In a one-step-ahead prediction task where input data are provided only from the fast subsystem, we show that the proposed model yields better performance than the standard RC model with identical LI neurons. Our analysis reveals that the timescale required for producing each component of target multiscale dynamics is appropriately and flexibly selected from the reservoir dynamics by model training. In a long-term prediction task, we demonstrate that a closed-loop version of the proposed model can achieve longer-term predictions compared to the counterpart with identical LI neurons depending on the hyperparameter setting., Comment: 7 pages, 6 figures, supplemental material (10 pages)

Modern computation based on the von Neumann architecture is today a mature cutting-edge science. In the Von Neumann architecture, processing and memory units are implemented as separate blocks interchanging data intensively and continuously. This data transfer is responsible for a large part of the power consumption. The next generation computer technology is expected to solve problems at the exascale with 1018 calculations each second. Even though these future computers will be incredibly powerful, if they are based on von Neumann type architectures, they will consume between 20 and 30 megawatts of power and will not have intrinsic physically built-in capabilities to learn or deal with complex data as our brain does. These needs can be addressed by neuromorphic computing systems which are inspired by the biological concepts of the human brain. This new generation of computers has the potential to be used for the storage and processing of large amounts of digital information with much lower power consumption than conventional processors. Among their potential future applications, an important niche is moving the control from data centers to edge devices. The aim of this Roadmap is to present a snapshot of the present state of neuromorphic technology and provide an opinion on the challenges and opportunities that the future holds in the major areas of neuromorphic technology, namely materials, devices, neuromorphic circuits, neuromorphic algorithms, applications, and ethics. The Roadmap is a collection of perspectives where leading researchers in the neuromorphic community provide their own view about the current state and the future challenges. We hope that this Roadmap will be a useful resource to readers outside this field, for those who are just entering the field, and for those who are well established in the neuromorphic community. https://doi.org/10.1088/2634-4386/ac4a83

This paper studies numerically how the signal detector arrangement influences the performance of reservoir computing using spin waves excited in a ferrimagnetic garnet film. This investigation is essentially important since the input information is not only conveyed but also transformed by the spin waves into high-dimensional information space when the waves propagate in the film in a spatially distributed manner. This spatiotemporal dynamics realizes a rich reservoir-computational functionality. First, we simulate spin waves in a rectangular garnet film with two input electrodes to obtain spatial distributions of the reservoir states in response to input signals, which are represented as spin vectors and used for a machine-learning waveform classification task. The detected reservoir states are combined through readout connection weights to generate a final output. We visualize the spatial distribution of the weights after training to discuss the number and positions of the output electrodes by arranging them at grid points, equiangularly circular points or at random. We evaluate the classification accuracy by changing the number of the output electrodes, and find that a high accuracy ($>$ 90\%) is achieved with only several tens of output electrodes regardless of grid, circular or random arrangement. These results suggest that the spin waves possess sufficiently complex and rich dynamics for this type of tasks. Then we investigate in which area useful information is distributed more by arranging the electrodes locally on the chip. Finally, we show that this device has generalization ability for input wave-signal frequency in a certain frequency range. These results will lead to practical design of spin-wave reservoir devices for low-power intelligent computing in the near future.

Spin waves propagating through a stripe domain structure and reservoir computing with their spin dynamics have been numerically studied with focusing on the relation between physical phenomena and computing capabilities. Our system utilizes a spin-wave-based device that has a continuous magnetic garnet film and 1-input/72-output electrodes on the top. To control spatially-distributed spin dynamics, a stripe magnetic domain structure and amplitude-modulated triangular input waves were used. The spatially-arranged electrodes detected spin vector outputs with various nonlinear characteristics that were leveraged for reservoir computing. By moderately suppressing nonlinear phenomena, our system achieves 100$\%$ prediction accuracy in temporal exclusive-OR (XOR) problems with a delay step up to 5. At the same time, it shows perfect inference in delay tasks with a delay step more than 7 and its memory capacity has a maximum value of 21. This study demonstrated that our spin-wave-based reservoir computing has a high potential for edge-computing applications and also can offer a rich opportunity for further understanding of the underlying nonlinear physics., Comment: 14 pages, 13 figures

Metapopulation epidemic models describe epidemic dynamics in networks of spatially distant patches connected with pathways for migration of individuals. In the present study, we deal with a susceptible-infected-recovered (SIR) metapopulation model where the epidemic process in each patch is represented by an SIR model and the mobility of individuals is assumed to be a homogeneous diffusion. Our study focuses on two types of patches including high-risk and low-risk ones, in order to evaluate intervention strategies for epidemic control. We theoretically analyze the intervention threshold, indicating the critical fraction of low-risk patches for preventing a global epidemic outbreak. We show that targeted intervention to high-degree patches is more effective for epidemic control than random intervention. The theoretical results are validated by Monte Carlo simulation for synthetic and realistic scale-free patch networks. Our approach is useful for exploring better local interventions aimed at containment of epidemics., Comment: 11pages, 8 figures

Reservoir computing is a computational framework suited for temporal/sequential data processing. It is derived from several recurrent neural network models, including echo state networks and liquid state machines. A reservoir computing system consists of a reservoir for mapping inputs into a high-dimensional space and a readout for pattern analysis from the high-dimensional states in the reservoir. The reservoir is fixed and only the readout is trained with a simple method such as linear regression and classification. Thus, the major advantage of reservoir computing compared to other recurrent neural networks is fast learning, resulting in low training cost. Another advantage is that the reservoir without adaptive updating is amenable to hardware implementation using a variety of physical systems, substrates, and devices. In fact, such physical reservoir computing has attracted increasing attention in diverse fields of research. The purpose of this review is to provide an overview of recent advances in physical reservoir computing by classifying them according to the type of the reservoir. We discuss the current issues and perspectives related to physical reservoir computing, in order to further expand its practical applications and develop next-generation machine learning systems., Comment: 62 pages, 13 figures

In this paper, we show that spatial correlation of renewable energy outputs greatly influences the robustness of power grids. First, we propose a new index for the spatial correlation among renewable energy outputs. We find that the spatial correlation of renewable energy outputs in a short time-scale is as weak as that caused by independent random variables and that in a long time-scale is as strong as that under perfect synchronization. Then, by employing the topology of the power grid in eastern Japan, we analyze the robustness of the power grid with spatial correlation of renewable energy outputs. The analysis is performed by using a realistic differential-algebraic equations model and the result shows that the spatial correlation of the energy resources strongly degrades the robustness of the power grid. Our result suggests that the spatial correlation of the renewable energy outputs should be taken into account when estimating the stability of power grids.

Universal scaling laws form one of the central issues in physics. A non-standard scaling law or a breakdown of a standard scaling law, on the other hand, can often lead to the finding of a new universality class in physical systems. Recently, we found that a statistical quantity related to fluctuations follows a non-standard scaling law with respect to system size in a synchronized state of globally coupled non-identical phase oscillators [Nishikawa et al., Chaos $\boldsymbol{22}$, 013133 (2012)]. However, it is still unclear how widely this non-standard scaling law is observed. In the present paper, we discuss the conditions required for the unusual scaling law in globally coupled oscillator systems, and we validate the conditions by numerical simulations of several different models.

We investigate a critical exponent related to synchronization transition in globally coupled nonidentical phase oscillators. The critical exponents of susceptibility, correlation time, and correlation size are significant quantities to characterize fluctuations in coupled oscillator systems of large but finite size and understand a universal property of synchronization. These exponents have been identified for the sinusoidal coupling but not fully studied for other coupling schemes. Herein, for a general coupling function including a negative second harmonic term in addition to the sinusoidal term, we numerically estimate the critical exponent of the correlation size, denoted by $\nu_+$, in a synchronized regime of the system by employing a non-conventional statistical quantity. First, we confirm that the estimated value of $\nu_+$ is approximately 5/2 for the sinusoidal coupling case, which is consistent with the well-known theoretical result. Second, we show that the value of $\nu_+$ increases with an increase in the strength of the second harmonic term. Our result implies that the critical exponent characterizing synchronization transition largely depends on the coupling function.

We investigate the diffusion coefficient of the time integral of the Kuramoto order parameter in globally coupled nonidentical phase oscillators. This coefficient represents the deviation of the time integral of the order parameter from its mean value on the sample average. In other words, this coefficient characterizes long-term fluctuations of the order parameter. For a system of N coupled oscillators, we introduce a statistical quantity D, which denotes the product of N and the diffusion coefficient. We study the scaling law of D with respect to the system size N. In other well-known models such as the Ising model, the scaling property of D is D \sim O(1) for both coherent and incoherent regimes except for the transition point. In contrast, in the globally coupled phase oscillators, the scaling law of D is different for the coherent and incoherent regimes: D \sim O(1/N^a) with a certain constant a>0 in the coherent regime and D \sim O(1) in the incoherent regime. We demonstrate that these scaling laws hold for several representative coupling schemes.