We developed three coupled van der Pol oscillators with different memristor coupling strengths. We investigate the synchronization state changing the coupling strength of one memristor. Compared with different resistor coupling strengths, amplitude death as the new synchronization state is obtained. In addition, we evaluated the synchronization dependence on the parameter of one memristor to analyze the relative phase difference, amplitude, and power consumption.

Pulse compression radar uses codes for phase modulation. Barker codes and optimal binary codes are examples of binary codes with small sidelobes. Binary codes compressed to several sub-pulses have also been proposed. One problem with searching for binary codes is that as the code length increases, so does the search time. Therefore, we propose a search method for binary codes compressed to several sub-pulses using quantum annealing. In addition, a search experiment using the proposed method was conducted using an Ising machine inspired by quantum annealing technology. As a result, it was found that the proposed method can search for binary codes compressed to several sub-pulses.

In this paper, we examine security for edge AI devices that detect anomaly behavior due to machine failures. Typical edge AI devices perform only inference, but the inference accuracy may be degraded when sensing data changes depending on the environment in which the devices are deployed. One countermeasure against this problem is on-device learning, in which an AI updates its learning model in accordance with its environment. However, on-device learning AIs face a variety of security threats: for example, the correct inference cannot be performed if the deployed AI model is manipulated by the attacker. Although threats including invasive/non-invasive attacks via physical access have been mentioned in the past, relatively few reports have investigated experimental attacks. In this work, we demonstrated data poisoning by exploiting physical attacks on sensors implemented on edge AI. A system for detecting abnormal vibrations was evaluated using an autoencoder that utilizes sensing data from an accelerometer attached to a cooling fan as input. Accelerometer values can be tampered with during an acoustic wave injection attack, so we exploited this kind of attack. The anomaly detector that learned the tampered data could not detect abnormal fan vibrations correctly.

The surge in online shopping has increased the demand for home delivery, leading to a shortage of delivery drivers. This situation compels logistics companies to devise efficient delivery routes, considering delivery times, number of vehicles, and fuel costs. This issue is called the Vehicle Routing Problem with Time Windows (VRPTW). Recent research indicates that the chaotic search (CS) method with adaptive penalty coefficients (APCs) can efficiently solve VRPTW. The CS method with APCs (CS-APC) achieves an efficient solution search by diversifying and centralizing solutions in accordance with the search situation. However, when the solution diversifies towards the end of the search, the variance of solutions obtained by CS-APC increases. In this study, we propose a CS method using APCs based on temperature annealing (APCT). The use of APCT strengthens solution centralization via the search time temperature, thereby improving the quality of solutions. Numerical experiments confirm that our proposed method finds solutions with fewer vehicles and shorter travel distances than when using the conventional method.

An adaptive ant colony optimization with node clustering (AACO-NC) has been proposed as a method for solving the Traveling Salesman Problem (TSP). The AACO-NC first constructs a route and then improves the route by using a modified k-Opt method. However, the modified k-Opt method increases the computational complexity for searching neighborhood solutions. In this work, we investigate the performance of AACO-NC when applying the 2-Opt method instead of the modified k-Opt method to reduce the computational complexity.

We employ the small limit argument approximation (SLAA) to streamline the mathematical representation of a conventional Rician fading channel within the proposed single-input single-output (SISO) system model. The SLAA on the zeroth-ordered modified Bessel function of the first kind facilitates the transition of a newly modified Rician fading channel into the low signal-to-noise ratio (SNR) regime, where either the transmission power is low or the bandwidth is large. We reveal that, for a SISO system with an instantaneous SNR set at 20 dB and a Rician factor set at 0.05, the channel capacity amounts to 7.1 bits/s/Hz.

Digitally controlled color shift keying (DCSK) can accurately represent multilevel optical intensities by turning multiple RGB-LEDs on and off. On the other hand, the channel capacity per LED of DCSK is lower than that of the conventional color shift keying (CSK), although the number of RGB-LEDs is increased. Therefore, we propose a spatially modulated DCSK (SDCSK) scheme to realize a higher channel capacity. This SDCSK combines DCSK and a generalized space shift keying (GSSK) scheme. GSSK is a unique form of spatial modulation (SM). We also consider the new DCSK constellation for increasing the amount of SM data of the SDCSK scheme. As a result of numerical analysis, the channel capacity of SDCSK increased by 3 bit/s/Hz compared with that of DCSK when the received SNR was larger than 57 dB.

IoT-based acoustic data collection systems have been proposed for detecting anomalies in mechanical equipment. However, traditional systems using audible sound do not work well due to interference from various sound sources. To solve this problem, we have previously proposed an ultrasonic data collection system, but the clear and high-quality sound required for anomaly detection results in an increase in data volume. This paper proposes a method for extracting only the necessary frequency band based on the periodicity of the target signal, aiming to reduce the amount of data. Our experiments demonstrate that the proposed method efficiently determines the appropriate frequency band in which the characteristics of the target equipment appear, thereby enabling the amount of data to be reduced by more than 45%.

This study investigates the entrainment phenomenon, which is related to how speakers adapt to their partners in daily conversation. In contrast to previous fragmented studies, we propose a comprehensive approach that integrates various acoustic features and explores entrainment across multiple parameters, the comprehensived considered entrainment can be extracted using a linear model. We hypothesize that interlocutors utilize diverse strategies for modulating their speech features to reflect different dialogue scenarios. We calculate proximity, convergence, and synchrony metrics to differentiate conversation scenarios and propose a linear model to clarify how dialogue strategies modulate and integrate speech features. The results of evaluation with four types of conversations from the Golden Marriage corpus show that the proposed method outperforms principal component analysis in extracting integrated acoustic features and exploring entrainment across multiple parameters by the classification results. This research contributes to understanding how speakers adapt in conversations, offering insights into the intricate dynamics of entrainment and enhancing the accuracy of conversation scenario classification.

While the automation and intellectualization of factories are advancing due to the promotion of Digital Transformation (DX), there are still many processes and tasks that require craftsmen. The efficiency improvement of skill transfer is very important, and a system that quickly develops beginners into skilled workers is necessary. Therefore, it is necessary to have a technique for numerical measurement and visualization of proficiency. Although images are generally used for the recognition of work action, we have shown that action recognition and task segmentation can be achieved to a high degree of accuracy using only information from microphones and inertial measurement devices that are worn by workers. On this basis, a method for automatically visualizing the proficiency of workers in assembly work using inertia and sound sensors was examined.

This paper introduces a method for automatically detecting bolus in ultrasound videos using YOLOv5 to evaluate swallowing function. Neck ultrasound videos show promise for noninvasive swallowing function evaluation to prevent aspiration pneumonia. To effectively evaluate swallowing, it is essential to determine the position of the bolus, appearance frame, and size within the video. Currently, this evaluation relies on visual inspection by an examiner, requiring considerable effort. Therefore, this study aims to automate bolus detection in ultrasound videos using YOLOv5, widely used in object detection. The effectiveness of our method is assessed using ultrasound videos obtained from swallowing experiments involving jelly beverages and water.

Noncontact vital signs sensing is the technology for measuring vitals without attaching measuring instruments to the human body. Compared with the contact type, the noncontact type does not require the stress of mounting a measuring instrument. In addition, it is an effective means for patients who have difficulty attaching measurement instruments. This paper focuses on a millimeter-wave (MMW) radar as a device for noncontact vital signs sensing, and we develop a noncontact respiration measurement system using the MMW radar. We have experimentally measured the respiration rate of a subject using the developed system inside and outside the electromagnetic anechoic chamber. As a result, we have confirmed that the measured respiration rate corresponds to the subject-counted respiration rate. Moreover, there was almost no effect of environmental noise on the respiration rate measurement with this study's experimental environment and parameters.

We proposed a method to reduce the data-mining bias[1] caused by the complexity of machine learning models and to increase their readability and transparency of the models, considering an investment situation in financial business as an example. By investment simulations with real financial data, we could confirm that our proposed method improved investment performance compared with the original random forest[2] even though decision rules were simplified and their number of them was reduced by our proposed method.

In this study, we investigated the impact of relevance boundaries and common words on a fund search system, and constructed a fund search visualization system using density-based clustering, validating it across multiple topics. The results indicate that setting relevance boundaries has a significant impact on search results, and using a visualization system can effectively enhance the search efficiency.

One of the requirements in times of disaster is to obtain information to confirm the safety of victims and to grasp the damage situation in the disaster area. To satisfy this, a system for building a temporary network using drones has been proposed. However, when a user shares his safety information with other users through the drone network, there is a risk of information leakage if the other user is malicious. Furthermore, it should be considered that disaster victims require communication terminals to be as small as possible and to have a long battery life as a requirement. Therefore, a cost-effective and secure information-sharing system that allows access control is required. We present a system that employs (2, 3)-threshold secret sharing and realizes a lightweight access control process specifically designed for drone networks in the event of communication failures during disasters. Furthermore, it incorporates security measures to protect against dishonest users.

In order to achieve high accuracy image navigation in space, it is necessary to perform highly accurate self-position estimation by matching images taken by the probe's camera with map images obtained through preliminary exploration. However, images taken in space are affected by disturbances due to the position of the sun and camera anomalies, and information loss due to compression for data transmission. As a result, the matching of corresponding points between images yields many outliers and the accuracy of self-position estimation is reduced. To address this problem, Random Sample Consensus (RANSAC) [1] is conventionally used to remove outliers. However, RANSAC significantly increases processing time for corresponding point groups with a large proportion of outliers. In this study, we adopt Neural Guided RANSAC (NG-RANSAC) [2] for images with space-specific disturbances and achieve higher accuracy.

Bit-depth expansion is a technique for reconstructing a high-bit image by predicting the missing bits in a low-bit image. With the development of high-bit monitors, corresponding high-bit images are needed to maximize their performance. However, many image data are still in 8-bit format. It is a complicated task to accurately recover lost information by expanding the bit depth while distinguishing between false contours and edges in real images. In this study, we propose a novel bit-depth expansion method using a Swin Transformer-based network with Channel Attention Layers (CALs). This network achieves high-performance bit-depth expansion by utilizing not only spatial features, which is an advantage of the Swin Transformer-based network, but also the correlation between channels obtained by CALs. Experimental results show that the proposed method outperforms conventional methods.

In this paper, we propose a novel graph structure with excluded seed regions and expanded adjacent nodes for image segmentation via Graph Cut. Graph Cut is one of the most effective image segmentation methods. The computational complexity of Graph Cut is known as O(n²log n). In the conventional method, since n is the total number of input image pixels, the larger the image size, the longer the computation time. In our previous study, accelerated Graph Cut by excluding seed region from graph structure was proposed. Our previous study has reduced the number of n and accelerated computation time. However, the challenge of accuracy remains. In this study, we expanded adjacent nodes to improve accuracy. According to the simulations, the proposed method achieved accurate results compared to conventional method.

Utilizing computer vision for animal behavior monitoring provides significant benefits by minimizing direct handling and capturing diverse traits through a single sensor. However, accurately identifying animals remains a challenge. To address this, this study introduces an innovative approach to monitor black cattle in dynamic agricultural environments to ensure their health welfare. By integrating advanced techniques like DETIC for automated labeling and YOLOv8 for real-time detection, the research emphasizes improving accuracy and robustness in tracking black cattle tracking within complex open ranch environments. Moreover, the customized ByteTrack model tailored for ranch scenarios significantly enhances cattle tracking across intricate landscapes. Achieving a mean Average Precision (mAP) of 0.901 and a Multi-Object Tracking Accuracy (MOTA) of average accuracy 92.185% of four videos, this approach appears to offer a viable resolution for conducting individual cattle behavior analysis experiments through the application of computer vision.

This study proposes an advanced camera-based monitoring system for individual black cows in closed farms. By leveraging computer vision and deep learning, the system identifies five key cow actions: eating, drinking, sitting, standing, and walking. A multi-stage approach classifies actions first as static (eating, drinking, sitting, standing) or dynamic (walking) categories based on Kalman Filter velocity information. Further classification distinguishes among four static actions. A Convolutional Neural Network (CNN) refines especially for sitting and standing. On the other hand, cow head regions and specific zone locations help distinguish eating and drinking. The system achieves an overall accuracy of 80% in long data sequences, demonstrating its potential for precision livestock farming.

We propose an inhibitory control system for downhill turning by using a pre-installed friction brake mounted on a manual wheelchair. The braking force of the wheel is controlled by an added DC geared motor that pulls the brake wire in accordance with the ground angle. The proposed automatic braking system can reduce unexpected turns on downhill slopes. The proposed control system is also designed to be installed as an add-on device for manual wheelchairs. The effectiveness of the proposed system is verified using a wheelchair through a slope running experiments.

In recent years, collective human behavior has been investigated by using small wireless sensors in several fields including medical and educational settings, where individuals' body movements and their proximity are observed and analyzed. These data have been mainly collected in several educational and medical settings, and few data observed in other situations are thus far available. We here investigate the collective human behavior of audience members during a live music concert by observing their body movements. We examine how different pieces of music to which audience members listen affect their body movements, using the time series data of the magnitude of body movements and the similarities in the body movements of the audience members. The results suggest that fast-tempo music can lead to an increase in the magnitude of body movements of audience members. Moreover, slow-tempo music might cause an increase in similarity between the body movements of the audience members.