arXiv subjects

Kenji Suzuki

Publications and source records attributed to Kenji Suzuki.

At least 37 records · Page 2Linked to original sources

arXiv2021

A Multi-task Mean Teacher for Semi-supervised Facial Affective Behavior Analysis

Affective Behavior Analysis is an important part in human-computer interaction. Existing multi-task affective behavior recognition methods suffer from the problem of incomplete labeled datasets. To tackle this problem, this paper presents a semi-supervised model with a mean teacher framework to leverage additional unlabeled data. To be specific, a multi-task model is proposed to learn three different kinds of facial affective representations simultaneously. After that, the proposed model is assigned to be student and teacher networks. When training with unlabeled data, the teacher network is employed to predict pseudo labels for student network training, which allows it to learn from unlabeled data. Experimental results showed that our proposed method achieved much better performance than baseline model and ranked 4th in both competition track 1 and track 2, and 6th in track 3, which verifies that the proposed network can effectively learn from incomplete datasets.

cs.CV↗

arXiv2021

Virtual Landmark-Based Control of Docking Support for Assistive Mobility Devices

This work proposes an autonomous docking control for nonholonomic constrained mobile robots and applies it to an intelligent mobility device or wheelchair for assisting the user in approaching resting furniture such as a chair or a bed. We defined a virtual landmark inferred from the target docking destination. Then, we solve the problem of keeping the targeted volume inside the field of view (FOV) of a tracking camera and docking to the virtual landmark through a novel definition that enables to control for the desired end-pose. In this article, we proposed a nonlinear feedback controller to perform the docking with the depth camera's FOV as a constraint. Then, a numerical method is proposed to find the feasible space of initial states where convergence could be guaranteed. Finally, the entire system was embedded for real-time operation on a standing wheelchair with the virtual landmark estimation by 3D object tracking with an RGB-D camera and we validated the effectiveness in simulation and experimental evaluations. The results show the guaranteed convergence for the feasible space depending on the virtual landmark location. In the implementation, the robot converges to the virtual landmark while respecting the FOV constraints.

cs.RO↗

arXiv2021

Neural Network Libraries: A Deep Learning Framework Designed from Engineers' Perspectives

While there exist a plethora of deep learning tools and frameworks, the fast-growing complexity of the field brings new demands and challenges, such as more flexible network design, speedy computation on distributed setting, and compatibility between different tools. In this paper, we introduce Neural Network Libraries (https://nnabla.org), a deep learning framework designed from engineer's perspective, with emphasis on usability and compatibility as its core design principles. We elaborate on each of our design principles and its merits, and validate our attempts via experiments.

cs.LG↗

arXiv2021

Data Cleansing for Deep Neural Networks with Storage-efficient Approximation of Influence Functions

Identifying the influence of training data for data cleansing can improve the accuracy of deep learning. An approach with stochastic gradient descent (SGD) called SGD-influence to calculate the influence scores was proposed, but, the calculation costs are expensive. It is necessary to temporally store the parameters of the model during training phase for inference phase to calculate influence sores. In close connection with the previous method, we propose a method to reduce cache files to store the parameters in training phase for calculating inference score. We only adopt the final parameters in last epoch for influence functions calculation. In our experiments on classification, the cache size of training using MNIST dataset with our approach is 1.236 MB. On the other hand, the previous method used cache size of 1.932 GB in last epoch. It means that cache size has been reduced to 1/1,563. We also observed the accuracy improvement by data cleansing with removal of negatively influential data using our approach as well as the previous method. Moreover, our simple and general proposed method to calculate influence scores is available on our auto ML tool without programing, Neural Network Console. The source code is also available.

cs.LG↗

arXiv2021

Passive Flow Control for Series Inflatable Actuators: Application on a Wearable Soft-Robot for Posture Assistance

This paper presents a passive control method for multiple degrees of freedom in a soft pneumatic robot through the combination of flow resistor tubes with series inflatable actuators. We designed and developed these 3D printed resistors based on the pressure drop principle of multiple capillary orifices, which allows a passive control of its sequential activation from a single source of pressure. Our design fits in standard tube connectors, making it easy to adopt it on any other type of actuator with pneumatic inlets. We present its characterization of pressure drop and evaluation of the activation sequence for series and parallel circuits of actuators. Moreover, we present an application for the assistance of postural transition from lying to sitting. We embedded it in a wearable garment robot-suit designed for infants with cerebral palsy. Then, we performed the test with a dummy baby for emulating the upper-body motion control. The results show a sequential motion control of the sitting and lying transitions validating the proposed system for flow control and its application on the robot-suit.

cs.RO↗

arXiv2020

Facial movement synergies and Action Unit detection from distal wearable Electromyography and Computer Vision

Distal facial Electromyography (EMG) can be used to detect smiles and frowns with reasonable accuracy. It capitalizes on volume conduction to detect relevant muscle activity, even when the electrodes are not placed directly on the source muscle. The main advantage of this method is to prevent occlusion and obstruction of the facial expression production, whilst allowing EMG measurements. However, measuring EMG distally entails that the exact source of the facial movement is unknown. We propose a novel method to estimate specific Facial Action Units (AUs) from distal facial EMG and Computer Vision (CV). This method is based on Independent Component Analysis (ICA), Non-Negative Matrix Factorization (NNMF), and sorting of the resulting components to determine which is the most likely to correspond to each CV-labeled action unit (AU). Performance on the detection of AU06 (Orbicularis Oculi) and AU12 (Zygomaticus Major) was estimated by calculating the agreement with Human Coders. The results of our proposed algorithm showed an accuracy of 81% and a Cohen's Kappa of 0.49 for AU6; and accuracy of 82% and a Cohen's Kappa of 0.53 for AU12. This demonstrates the potential of distal EMG to detect individual facial movements. Using this multimodal method, several AU synergies were identified. We quantified the co-occurrence and timing of AU6 and AU12 in posed and spontaneous smiles using the human-coded labels, and for comparison, using the continuous CV-labels. The co-occurrence analysis was also performed on the EMG-based labels to uncover the relationship between muscle synergies and the kinematics of visible facial movement.

cs.HC↗

arXiv2020

Control Interface for Hands-free Navigation of Standing Mobility Vehicles based on Upper-Body Natural Movements

In this paper, we propose and evaluate a novel human-machine interface (HMI) for controlling a standing mobility vehicle or person carrier robot, aiming for a hands-free control through upper-body natural postures derived from gaze tracking while walking. We target users with lower-body impairment with remaining upper-body motion capabilities. The developed HMI bases on a sensing array for capturing body postures; an intent recognition algorithm for continuous mapping of body motions to robot control space; and a personalizing system for multiple body sizes and shapes. We performed two user studies: first, an analysis of the required body muscles involved in navigating with the proposed control; and second, an assessment of the HMI compared with a standard joystick through quantitative and qualitative metrics in a narrow circuit task. We concluded that the main user control contribution comes from Rectus Abdominis and Erector Spinae muscle groups at different levels. Finally, the comparative study showed that a joystick still outperforms the proposed HMI in usability perceptions and controllability metrics, however, the smoothness of user control was similar in jerk and fluency. Moreover, users' perceptions showed that hands-free control made it more anthropomorphic, animated, and even safer.

cs.RO↗

arXiv2018

LEO-to-ground optical communications using SOTA (Small Optical TrAnsponder) - Payload verification results and experiments on space quantum communications

Free-space optical communications have held the promise of revolutionizing space communications for a long time. The benefits of increasing the bitrate while reducing the volume, mass and energy of the space terminals have attracted the attention of many researchers for a long time. In the last few years, more and more technology demonstrations have been taking place with participants from both the public and the private sector. The National Institute of Information and Communications Technology (NICT) in Japan has a long experience in this field. SOTA (Small Optical TrAnsponder) was the last NICT space lasercom mission, designed to demonstrate the potential of this technology applied to microsatellites. Since the beginning of SOTA mission in 2014, NICT regularly established communication using the Optical Ground Stations (OGS) located in the Headquarters at Koganei (Tokyo) to receive the SOTA signals, with over one hundred successful links. All the goals of the SOTA mission were fulfilled, including up to 10-Mbit/s downlinks using two different wavelengths and apertures, coarse and fine tracking of the OGS beacon, space-to-ground transmission of the on-board-camera images, experiments with different error correcting codes, interoperability with other international OGS, and experiments on quantum communications. The SOTA mission ended on November 2016, more than doubling the designed lifetime of 1-year. In this paper, the SOTA characteristics and basic operation are explained, along with the most relevant technological demonstrations.

physics.space-ph↗

arXiv2017

A Systematic Literature Review of Experiments in Socially Assistive Robotics using Humanoid Robots

We perform a Systematic Literature Review to discover how Humanoid robots are being applied in Socially Assistive Robotics experiments. Our search returned 24 papers, from which 16 were included for closer analysis. To do this analysis we used a conceptual framework inspired by Behavior-based Robotics. We were interested in finding out which robot was used (most use the robot NAO), what the goals of the application were (teaching, assisting, playing, instructing), how the robot was controlled (manually in most of the experiments), what kind of behaviors the robot exhibited (reacting to touch, pointing at body parts, singing a song, dancing, among others), what kind of actuators the robot used (always motors, sometimes speakers, hardly ever any other type of actuator) and what kind of sensors the robot used (in many studies the robot did not use any sensors at all, in others the robot frequently used camera and/or microphone). The results of this study can be used for designing software frameworks targeting Humanoid Socially Assistive Robotics, especially in the context of Software Product Line Engineering projects.

cs.RO↗

arXiv2017

Baryon number current in holographic noncommutative QCD

We consider the noncommutative deformation of the finite temperature holographic QCD (Sakai--Sugimoto) model in external electric and magnetic field and evaluate the effect of the noncommutaivity on the properties of the conductor-insulator phase transition associated with a baryon number current. Although the noncommutative deformation of the gauge theory does not change the phase structure with respect to the baryon number current, the transition temperature $T_{c}$, the transition electric field $e_{c}$ and magnetic field $b_{c}$ in the conductor-insurator phase transition depend on the noncommutativity parameter $θ$. Namely, the noncommutativity of space coordinates has an influence on the shape of the phase diagram for the conductor-insurator phase transition. On the other hand, the allowed range of the noncommutativity parameter can be restricted by the reality condition of the constants of motion.

hep-th↗

arXiv2014

Recursion Method for Deriving Energy-Independent Effective Interaction

The effective-interaction theory has been one of the useful and practical methods for solving nuclear many-body problems based on the shell model. Various approaches have been proposed which are constructed in terms of the so-called $\widehat{Q}$ box and its energy derivatives introduced by Kuo {\it et al}. In order to find out a method of calculating them we make decomposition of a full Hilbert space into subspaces (the Krylov subspaces) and transform a Hamiltonian to a block-tridiagonal form. This transformation brings about much simplification of the calculation of the $\widehat{Q}$ box. In the previous work a recursion method has been derived for calculating the $\widehat{Q}$ box analytically on the basis of such transformation of the Hamiltonian. In the present study, by extending the recursion method for the $\widehat{Q}$ box, we derive another recursion relation to calculate the derivatives of the $\widehat{Q}$ box of arbitrary order. With the $\widehat{Q}$ box and its derivatives thus determined we apply them to the calculation of the $E$-independent effective interaction given in the so-called Lee-Suzuki (LS) method for a system with a degenerate unperturbed energy. We show that the recursion method can also be applied to the generalized LS scheme for a system with non-degenerate unperturbed energies. If the Hilbert space is taken to be sufficiently large, the theory provides an exact way of calculating the $\widehat{Q}$ box and its derivatives. This approach enables us to perform recursive calculations for the effective interaction to arbitrary order for both systems with degenerate and non-degenerate unperturbed energies.

nucl-th↗

arXiv2013

The spectrum of low spin mesons at finite temperature in holographic noncommutative QCD

We have constructed a noncommutative deformation of the holographic QCD (Sakai-Sugimoto) model and evaluated the mass spectrum of low spin vector mesons at finite temperature. The masses of light vector- and pseudovector-meson in the noncommutative holographic QCD model reduces over the whole area in the intermediate-temperature regime compared to the commutative case. However, the space noncommutativity does not change the properties of temperature dependence for the mass spectrum of low spin mesons. The masses of meson also decrease with increasing temperature in noncommutative case.

hep-th↗

arXiv2013

Formulation of effective interaction in terms of renormalized vertices and propagators

One of the useful and practical methods for solving quantum-mechanical many-body systems is to recast the full problem into a form of the effective interaction acting within a model space of tractable size. Many of the effective-interaction theories in nuclear physics have been formulated by use of the so called $\hatQ$ box introduced by Kuo et.al. It has been one of the central problems how to calculate the $\hatQ$ box accurately and efficiently. We first show that, introducing new basis states, the Hamiltonian is transformed to a block-tridiagonal form in terms of submatrices with small dimension. With this transformed Hamiltonian, we next prove that the $\hatQ$ box can be expressed in two ways: One is a form of continued fraction and the other is a simple series expansion up to second order with respect to renormalized vertices and propagators. This procedure ensures to derive an exact $\hatQ$ box, if the calculation converges as the dimension of the Hilbert space tends to infinity. The $\hatQ$ box given in this study corresponds to a non-perturbative solution for the energy-dependent effective interaction which is often referred to as the Bloch-Horowitz or the Feshbach form. By applying the $\hatZ$-box approach based on the $\hatQ$ box proposed previously, we introduce a graphical method for solving the eigenvalue problem of the Hamiltonian. The present approach has a possibility of resolving many of the difficulties encountered in the effective-interaction theory.

nucl-th↗

arXiv2011

Chiral Symmetry Restoration in Holographic Noncommutative QCD

We consider the noncommutative deformation of the Sakai--Sugimoto model at finite temperature and finite baryon chemical potential. The space noncommutativity is possible to have an influence on the flavor dynamics of the QCD. The critical temperature and critical value of the chemical potential are modified by the space noncommutativity. The influence of the space noncommutativity on the flavor dynamics of the QCD is caused by the Wess--Zumino term in the effective action of the D8-branes. The intermediate temperature phase, in which the gluons deconfine but the chiral symmetry remains broken, is easy to be realized in some region of the noncommutativity parameter.

hep-th↗

arXiv2005

Glueball mass spectra for supergravity duals of noncommutative gauge theories

We derive the glueball masses in noncommutative super Yang--Mills theories in four dimensions via the dual supergravity description. The spectrum of glueball masses is discrete due to the noncommutativity and the glueball masses are proportional to the noncommutativity parameter with dimension of length. The mass spectrum in the WKB approximation closely agrees with the mass spectrum in finite temperature Yang--Mills theory.

hep-th↗

arXiv2005

Glueball Masses from Supergravity with Flavor

We study the glueballs in a four-dimensional ${\cal N}=2$ super Yang-Mills theory with fundamental matters in terms of the supergravity dual. The supergravity background is constructed by $N$ D3 brane with a probe D7 brane. We numerically compute the glueball masses for $0^{++}$ and $1^{--}$ in the background. We find that the mass ratio of $0^{++}$ glueballs is mostly in agreement with the lattice calculations. We compare the glueball masses with the meson masses. The mass ratio $M_{GB}/M_{meson}$ is calculated with 1.4. If the mass $M_{GB}$ is set to $1.6 GeV$, the meson mass $M_{meson}$ is given with $1.1 GeV$.

hep-th↗

arXiv2002

D1/D5 system and Wilson Loops in (Non-)commutative Gauge Theories

We study the behavior of the Wilson loop in the (5+1)-dimensional supersymmetric Yang-Mills theory with the presence of the solitonic object. Using the dual string description of the Yang-Mills theory that is given by the D1/D5 system, we estimate the Wilson loops both in the temporal and spatial cases. For the case of the temporal loop, we obtain the velocity dependent potential. For the spatial loop, we find that the area law is emerged due to the effect of the D1-branes. Further, we consider D1/D5 system in the presence of the constant B field. It is found that the Wilson loop obeys the area law for the effect of the noncommutativity.

hep-th↗

arXiv2001

D0-D4 system and QCD_{3+1}

We consider a $(3+1)$-dimensional QCD model using a dual supergravity description with a non-extremal $D0$-$D4$ brane background. We calculate the spectrum of glueball masses and Wilson loops in the background. The mass spectrum is shown to coincide with one in non-extremal $D4$-brane systems, and an area low of spatial Wilson loops is established. We show that there is a region that Kaluza-Klein modes of the Euclidean time direction are decoupled without decoupling glueball masses.

hep-th↗