Random walks on graphs are widely used for diverse applications in information science, such as data analysis, search, and network exploration; however, they often suffer from localized stagnation within densely connected subgraphs, leading to decreased exploration efficiency. To address this issue, we propose Coarse Walk, a new random walk method that probabilistically switches between the original graph G and a coarsened graph Gc. By traversing both G and Gc, Coarse Walk introduces non-local transitions without relying on global information about the entire graph, thus enabling the agent to escape local clusters. We evaluate Coarse Walk on six graph types (Random, Barábasi–Albert, Tree, Comb, Caveman, and Barbell) using six different random walk variants (SRW, BiasedRW, 3-History, NBRW, SARW, and VARW), measuring performance in terms of cover time and average hitting time. The results demonstrate that Coarse Walk significantly improves exploration efficiency for graphs with strong local substructures―such as Caveman and Barbell graphs―while providing only modest or negligible gains on graphs without prominent clusters, such as Random and Barábasi–Albert graphs.

Graph coarsening is a widely used technique for reducing the complexity of large-scale networks while preserving their essential structural properties. However, coarsening alters the degree distribution, a key characteristic of graphs, making it crucial to understand these transformations. In this study, we analyze the impact of various coarsening algorithms—including RM, COARSENET, MGC, LVN, LVE, kron, and HEM—on the degree distribution of undirected graphs. We first develop an analytical model describing how the degree distribution evolves under RM-based coarsening and validate our findings using numerical experiments on diverse graph types. Additionally, we investigate the feasibility of recovering the original degree distribution from the coarsened graph using both analytical methods and graph neural networks. Our results indicate that RM retains the degree distribution more accurately for graphs with low average degree, while COARSENET and MGC cause significant alterations. Although analytical reconstruction performs well for low-degree graphs, its accuracy declines for graphs with higher connectivity. In contrast, graph neural networks consistently achieve high reconstruction accuracy across all coarsening methods, with particularly strong performance when using LVE and kron.

TCP BBR (Bottleneck Bandwidth and Round-trip Propagation Time) has been proposed as an efficient congestion control mechanism that manages network congestion based on end-to-end measurements of bottleneck link bandwidth and round-trip time (RTT), rather than relying on the detection of packet loss events, as used by loss-based TCPs such as TCP CUBIC. Unlike loss-based congestion control mechanisms, TCP BBR estimates the available bandwidth of the bottleneck router and the network's round-trip propagation delay to adjust its congestion window, aiming to achieve the optimal operating point of the network. Numerous studies have examined the performance of TCP BBR through experiments and simulations, but analytical studies primarily focused on modeling its characteristics and behaviors during startup or steady-state phases, falling short in analyzing its dynamic behavior under changing network conditions. This paper presents a discrete-time fluid model to mathematically describe the dynamic interaction between TCP BBR flows and intermediate routers in an arbitrary network topology. The model captures the relationship among the time evolution of TCP BBR congestion window, bottleneck link bandwidth and RTT measurements, and packet queuing in routers, at the granularity of round-trip times. Through numerical examples, we demonstrate the effectiveness of our fluid model for TCP BBR and reveal optimal pacing gain settings analytically.

With the exponential growth of these networks, traditional packet-level simulations have become computationally prohibitive, making fluid-based simulations a more viable option due to their efficiency and scala-bility. We propose a novel fluid-based simulation technique that significantly accelerates the computation of intercon-nected, time-varying delay elements, which are critical in the numerical simulations of large-scale networks. Our approach leverages the inherent loops within fluid models of large-scale networks to reduce computational burden. To demonstrate the effectiveness of our technique, we introduce FLNET (Fluid-based Large-scale NETwork simulator), an efficient and scalable network simulator designed for large-scale TCP/IP networks adhering to TCP congestion control algorithms. Through rigorous experiments with FLNET, we reveal that our technique enables faster and more efficient network simulations, achieving a performance gain of approximately 3 to 5 times faster than the conventional fluid-based simulator. This paper contributes to the field by offering a scalable solution to the challenges of simulating large-scale networks, paving the way for more accurate and efficient network analysis and planning.

Passengers of level 3-5 autonomous personal mobility vehicles (APMV) can perform non-driving tasks, such as reading books and smartphones, while driving. It has been pointed out that such activities may increase motion sickness, especially when frequently avoiding pedestrians or obstacles in shared spaces. Many studies have been conducted to build countermeasures, of which various computational motion sickness models have been developed. Among them, models based on subjective vertical conflict (SVC) theory, which describes vertical changes in direction sensed by human sensory organs v.s. those expected by the central nervous system, have been actively developed. To model motion sickness due to conflict between visual vertical information and vestibular sensation, we proposed a 6 DoF SVC-VV model which added a visually perceived vertical block into a conventional 6 DoF SVC model to predict visual vertical directions from image data simulating the visual input of a human. In a driving experiment, 27 participants rode on the APMV and experienced slalom driving with two visual conditions: looking ahead (LAD) and working with a tablet device (WAD). We verified that passengers got motion sickness while riding the APMV, and the symptoms were severer when especially working on it, by simulating the frequent pedestrian avoidance scenarios of the APMV in the experiment. In addition, the results of the experiment demonstrated that the proposed 6 DoF SVC-VV model could describe the increased motion sickness experienced when the visual vertical and gravitational acceleration directions were different.

Information-Centric Networking (ICN) has gained attention as one of the next-generation internet architectures that focuses on the data being transmitted rather than the hosts transmitting it. Due to the differences between ICN and TCP/IP networks, it is not possible to evaluate the performance of ICN using network simulators designed for TCP/IP. A number of studies have been conducted to develop ICN network simulators. However, further acceleration of ICN network simulators is expected to enable large-scale ICN network performance evaluation. In this paper, we analyze the performance bottleneck of the flow-level ICN simulator called FICNSIM (Fluid-based ICNSIMulator) by profiling its performance using the Julia language source code. Specifically, we identify the processing that is causing the performance bottleneck of FICNSIM and investigate the scalability of FICNSIM with respect to network scale.

Passengers (drivers) of level 3-5 autonomous personal mobility vehicles (APMV) and cars can perform non-driving tasks, such as reading books and smartphones, while driving. It has been pointed out that such activities may increase motion sickness. Many studies have been conducted to build countermeasures, of which various computational motion sickness models have been developed. Many of these are based on subjective vertical conflict (SVC) theory, which describes vertical changes in direction sensed by human sensory organs vs. those expected by the central nervous system. Such models are expected to be applied to autonomous driving scenarios. However, no current computational model can integrate visual vertical information with vestibular sensations. We proposed a 6 DoF SVC-VV model which add a visually perceived vertical block into a conventional six-degrees-of freedom SVC model to predict VV directions from image data simulating the visual input of a human. Hence, a simple image-based VV estimation method is proposed. As the validation of the proposed model, this paper focuses on describing the fact that the motion sickness increases as a passenger reads a book while using an AMPV, assuming that visual vertical (VV) plays an important role. In the static experiment, it is demonstrated that the estimated VV by the proposed method accurately described the gravitational acceleration direction with a low mean absolute deviation. In addition, the results of the driving experiment using an APMV demonstrated that the proposed 6 DoF SVC-VV model could describe that the increased motion sickness experienced when the VV and gravitational acceleration directions were different.

In this paper, we propose SepNet, a deep neural network (DNN) designed to predict separation matrices from multichannel observations. One well-known approach to blind source separation (BSS) involves independent component analysis (ICA). A recently developed method called independent low-rank matrix analysis (ILRMA) is one of its powerful variants. These methods allow the estimation of separation matrices based on deterministic iterative algorithms. Specifically, ILRMA is designed to update the separation matrix according to an update rule derived based on the majorization-minimization principle. Although ILRMA performs reasonably well under some conditions, there is still room for improvement in terms of both separation accuracy and computation time, especially for large-scale microphone arrays. The existence of a deterministic iterative algorithm that can find one of the stationary points of the BSS problem implies that a DNN can also play that role if designed and trained properly. Motivated by this, we propose introducing a DNN that learns to convert a predefined input (e.g., an identity matrix) into a true separation matrix in accordance with a multichannel observation. To enable it to find one of the multiple solutions corresponding to different permutations of the source indices, we further propose adopting a permutation invariant training strategy to train the network. By using a fully convolutional architecture, we can design the network so that the forward propagation can be computed efficiently. The experimental results revealed that SepNet was able to find separation matrices faster and with better separation accuracy than ILRMA for mixtures of two sources.

In this paper, we propose a low-latency online extension of wave-U-net for single-channel speech enhancement, which utilizes teacher-student learning to reduce the system latency while keeping the enhancement performance high. Wave-U-net is a recently proposed end-to-end source separation method, which achieved remarkable performance in singing voice separation and speech enhancement tasks. Since the enhancement is performed in the time domain, wave-U-net can efficiently model phase information and address the domain transformation limitation, where the time-frequency domain is normally adopted. In this paper, we apply wave-U-net to face-to-face applications such as hearing aids and in-car communication systems, where a strictly low-latency of less than 10 ms is required. To this end, we investigate online versions of wave-U-net and propose the use of teacher-student learning to prevent the performance degradation caused by the reduction in input segment length such that the system delay in a CPU is less than 10 ms. The experimental results revealed that the proposed model could perform in real-time with low-latency and high performance, achieving a signal-to-distortion ratio improvement of about 8.73 dB.

Despite the excellent performance of neural-network-based audio source separation methods and their wide range of applications, their robustness against intentional attacks has been largely neglected. In this work, we reformulate various adversarial attack methods for the audio source separation problem and intensively investigate them under different attack conditions and target models. We further propose a simple yet effective regularization method to obtain imperceptible adversarial noise while maximizing the impact on separation quality with low computational complexity. Experimental results show that it is possible to largely degrade the separation quality by adding imperceptibly small noise when the noise is crafted for the target model. We also show the robustness of source separation models against a black-box attack. This study provides potentially useful insights for developing content protection methods against the abuse of separated signals and improving the separation performance and robustness.

In this paper, we deal with a multichannel source separation problem under a highly reverberant condition. The multichannel variational autoencoder (MVAE) is a recently proposed source separation method that employs the decoder distribution of a conditional VAE (CVAE) as the generative model for the complex spectrograms of the underlying source signals. Although MVAE is notable in that it can significantly improve the source separation performance compared with conventional methods, its capability to separate highly reverberant mixtures is still limited since MVAE uses an instantaneous mixture model. To overcome this limitation, in this paper we propose extending MVAE to simultaneously solve source separation and dereverberation problems by formulating the separation system as a frequency-domain convolutive mixture model. A convergence-guaranteed algorithm based on the coordinate descent method is derived for the optimiza- tion. Experimental results revealed that the proposed method outperformed the conventional methods in terms of all the source separation criteria in highly reverberant environments.

Although the performance of an RF rectifier is commonly evaluated using power conversion efficiency (PCE), that of various existing RF rectifiers cannot directly be compared. This is because PCEs are evaluated individually under different operating conditions. In this paper, an absolute figure of merit (FoM) using a reference rectifier circuit is proposed. The ideal factor, n-value, of a pn junction diode in a reference rectifier exhibiting the same PCE for operating conditions similar to that of a target rectifier under evaluation is taken as an FoM. It is demonstrated with a few examples that the proposed FoM is independent of the operation condition of the rectifier and can be used to evaluate various rectifiers objectively including those having a multistage configuration.

This letter proposes a multichannel source separation technique, the multichannel variational autoencoder (MVAE) method, which uses a conditional VAE (CVAE) to model and estimate the power spectrograms of the sources in a mixture. By training the CVAE using the spectrograms of training examples with source-class labels, we can use the trained decoder distribution as a universal generative model capable of generating spectrograms conditioned on a specified class index. By treating the latent space variables and the class index as the unknown parameters of this generative model, we can develop a convergence-guaranteed algorithm for supervised determined source separation that consists of iteratively estimating the power spectrograms of the underlying sources, as well as the separation matrices. In experimental evaluations, our MVAE produced better separation performance than a baseline method.

Shadow is significant for an effective projection of real objects and has an important role in three-dimensional effects and reality of virtual space. Therefore, an avatar's shadow should have interactive effects with the avatar's self in a virtual space. In our previous research, we reported that an auto-generated interactive avatar's motion that is different from the talker's own motion does not always support human communication. However, using an avatar-shadow, talkers can communicate effectively, without inconsistency between their own motions and their avatar's motions in a virtual space in which the avatar-shadow's motions are expressed by combining the talker's own motions with auto-generated interactive motions. Accordingly, we propose an embodied entrainment avatar-shadow system. This system is composed of the talkers' avatars and their avatar-shadows based on their own motions and auto-generated entrained motions. In addition, we develop a prototype from three viewpoints for a more effective use of the avatar-shadow. In this paper, we perform two conversation experiments to study the effects of the avatar-shadow using sensory evaluation. The first experiment is a role-play experiment in which two subjects play roles of a speaker and a listener. The second one is a free conversation experiment in which two subjects talk about daily life. The results of the sensory evaluation experiments allow us to demonstrate the effectiveness of the avatar-shadow's existence and the auto-generated nodding response.