Sou Nobukawa

Yusuke Sakemi, Sou Nobukawa, Toshitaka Matsuki et al.

Reservoir computing (RC) can efficiently process time-series data by transferring the input signal to randomly connected recurrent neural networks (RNNs), which are referred to as a reservoir. The high-dimensional representation of time-series data in the reservoir significantly simplifies subsequent learning tasks. Although this simple architecture allows fast learning and facile physical implementation, the learning performance is inferior to that of other state-of-the-art RNN models. In this paper, to improve the learning ability of RC, we propose self-modulated RC (SM-RC), which extends RC by adding a self-modulation mechanism. The self-modulation mechanism is realized with two gating variables: an input gate and a reservoir gate. The input gate modulates the input signal, and the reservoir gate modulates the dynamical properties of the reservoir. We demonstrated that SM-RC can perform attention tasks where input information is retained or discarded depending on the input signal. We also found that a chaotic state emerged as a result of learning in SM-RC. This indicates that self-modulation mechanisms provide RC with qualitatively different information-processing capabilities. Furthermore, SM-RC outperformed RC in NARMA and Lorentz model tasks. In particular, SM-RC achieved a higher prediction accuracy than RC with a reservoir 10 times larger in the Lorentz model tasks. Because the SM-RC architecture only requires two additional gates, it is physically implementable as RC, providing a new direction for realizing edge AI.

Hayato Nihei, Sou Nobukawa, Yusuke Sakemi et al.

Reservoir computing (RC) establishes the basis for the processing of time-series data by exploiting the high-dimensional spatiotemporal response of a recurrent neural network to an input signal. In particular, RC trains only the output layer weights. This simplicity has drawn attention especially in Edge Artificial Intelligence (AI) applications. Edge AI enables time-series anomaly detection in real time, which is important because detection delays can lead to serious incidents. However, achieving adequate anomaly-detection performance with RC alone may require an unacceptably large reservoir on resource-constrained edge devices. Without enlarging the reservoir, attention mechanisms can improve accuracy, although they may require substantial computation and undermine the learning efficiency of RC. In this study, to improve the anomaly detection performance of RC without sacrificing learning efficiency, we propose a spectral residual RC (SR-RC) that integrates the spectral residual (SR) method - a learning-free, bottom-up attention mechanism - with RC. We demonstrated that SR-RC outperformed conventional RC and logistic-regression models based on values extracted by the SR method across benchmark tasks and real-world time-series datasets. Moreover, because the SR method, similarly to RC, is well suited for hardware implementation, SR-RC suggests a practical direction for deploying RC as Edge AI for time-series anomaly detection.

Yusuke Sakemi, Yuji Okamoto, Takashi Morie et al.

Large-scale deep learning models are increasingly constrained by their immense energy consumption, limiting their scalability and applicability for edge intelligence. In-memory computing (IMC) offers a promising solution by addressing the von Neumann bottleneck inherent in traditional deep learning accelerators, significantly reducing energy consumption. However, the analog nature of IMC introduces hardware nonidealities that degrade model performance and reliability. This paper presents a novel approach to directly train physical models of IMC, formulated as ordinary-differential-equation (ODE)-based physical neural networks (PNNs). To enable the training of large-scale networks, we propose a technique called differentiable spike-time discretization (DSTD), which reduces the computational cost of ODE-based PNNs by up to 20 times in speed and 100 times in memory. We demonstrate that such large-scale networks enhance the learning performance by exploiting hardware nonidealities on the CIFAR-10 dataset. The proposed bottom-up methodology is validated through the post-layout SPICE simulations on the IMC circuit with nonideal characteristics using the sky130 process. The proposed PNN approach reduces the discrepancy between the model behavior and circuit dynamics by at least an order of magnitude. This work paves the way for leveraging nonideal physical devices, such as non-volatile resistive memories, for energy-efficient deep learning applications.