Reservoir computing (RC) refers to an artificial neural network framework that exhibits temporal dynamic behavior, and its computational structure can be implemented in physical systems. However, the network performance of conventional RC is limited in terms of power scale, readability, spatial and temporal scalabilities, mass producibility, and precision. Therefore, we propose an RC with atomic switches. An atomic switch is a new type of nanodevice that exhibits superior power scale, readability, spatial and temporal scalabilities, and mass producibility. In the proposed RC architecture, we arranged atomic switches sequentially in a ring formation and used time-division multiplexing. The resistance of the atomic switches in the proposed architecture changes nonlinearly with a change in input and memorizes the input; therefore, this RC architecture is expected to have high precision and a large memory capacity (MC). In this study, we simulated this architecture and compared it with the conventional architecture in terms of precision and MC. The results showed that the proposed RC architecture had higher precision and greater MC.

We can observe marked point processes in a wide variety of natural phenomena. Thus, it is important to establish theories and methods for analyzing such observed marked point processes. In a nonlinear time series analysis, the theories are the embedding theorems and the methods are the state space reconstruction from the observed time series. Although it has been revealed that we can reconstruct a state space from a point process without marked values, it has not been clarified yet whether we can reconstruct the state space from a marked point process. Therefore, in this study, we numerically investigated whether a state space could be reconstructed from a marked point process using a delayed coordinate system. To assess the issue, we evaluated the similarity between the interpoint distance distributions on an attractor of the original dynamical system that generated the marked point process and an attractor reconstructed from the marked point process. Results of numerical experiments show that the interpoint distance distributions in the original and reconstructed state spaces are similar, implying that it is possible to reconstruct nonlinear dynamical systems from the observed marked point processes.

In ARIB STD T-109, the inter-vehicle communication (IVC) systems are based on the carrier sense multiple access/orthogonal frequency division multiplexing (CSMA/OFDM) scheme. However, there is concern that communication performance is degraded owing to hidden terminals. On the other hand, to reduce this effect, research has been conducted on simultaneous communication using direct spread-code division multiple access (DS-CDMA). However, when this method is applied in urban environment where there are many reflections and diffractions, there is concern regarding the degradation of communication performance by interference waves. In this paper, the performances of DS-CDMA IVC with various forms of primary modulation are compared and the performance of DS-CDMA IVC with successive interference canceller (SIC) is evaluated to reduce the effects of interfering waves.

In the IoT, a small amount of payment (i.e., micropayment) enables the trading of sensor data collected by IoT devices. The IOTA cryptocurrency, which achieves high-speed transactions without transaction fees, shows the potential to realize such micropayment. In this work, we introduce and implement an IOTA-based micropayment system for IoT devices (i.e., Raspberry Pi), assuming an air quality monitoring application. We then evaluate the latency performance and power consumption.

Sound quality metrics (SQMs), such as sharpness and fluctuation strength, are used for objectively evaluating sound quality related to human perception. Recent research has revealed the auditory characteristics, such as the level-dependency and asymmetry, of auditory filter shapes. However, the existing models of the SQMs do not take into account such auditory characteristics. It is unclear whether these models can accurately account for sound quality. This paper proposes computational models of sharpness and fluctuation strength using proposed loudness models composed of the gammatone auditory filterbank and gammachirp auditory filterbank. These two metrics (sharpness and fluctuation strength) for several test signals were calculated using the proposed models then compared with those from perceptual data. The results indicate that the sharpness calculated with the proposed computational models of sharpness was similar to that from the perceptual data on sharpness and was more accurate than the baseline models. Also, the fluctuation strength calculated with the proposed computational models of fluctuation strength was almost similar to that from the perceptual data on fluctuation strength.

This paper deals with single-channel speaker-dependent speech separation. While discriminative approaches using deep neural networks (DNNs) have recently proved powerful, generative approaches, including methods based on non-negative matrix factorization (NMF), are still attractive because of their flexibility in handling the mismatch between training and test conditions. Although NMF-based methods work reasonably well for particular sound sources, one limitation is that they can fail to work for sources with spectrograms that do not comply with the NMF model. To address this problem, attempts have recently been made to replace the NMF model with DNNs. With a similar motivation to these attempts, we propose in this paper a variational autoencoder (VAE)-based monaural source separation (VASS) method using a conditional VAE (CVAE) for source spectrogram modeling. We further propose an extension of the VASS method, called the discriminative VASS (DVASS) method, which uses a discriminative criterion for model training so that the separated signals directly become optimal. Experimental results revealed that the VASS method performed better than an NMF-based method, and the DVASS method performed better than the VASS method.

In this paper, we describe a tomato recognition algorithm and a fruit grasping mechanism for a tomato-harvesting robot that is designed to operate in a horticultural facility. In the tomato recognition algorithm, image processing is performed on the color images captured by a depth camera to detect red ripe large tomato and oval mini-tomato fruits. Then, the spatial coordinates of the fruit center and the diameter of the fruit are calculated using the depth information acquired by the depth camera. For the grasping mechanism, we designed and fabricated a mechanism for grasping an oval mini-tomato, which is necessary in the harvesting process. Experiments using these systems were conducted in a horticultural facility where tomatoes are grown, and the performances of both the tomato recognition and grasping motions were verified.