In the viewpoint of the Bayesian brain hypothesis, Bayesian network model of cerebral cortex is promissing not only for computational modeling of brain, but also for an efficient brain- like artificial intelligence. A norious drawback in a Bayesian network is, however, the number of parameters that grows exponentially against the number of parent variables for a random variable. Restriction of the model may be a solution to this problem. Inspired by the biological plausibility, we previously proposed to use the combination of the

-OR and -AND gates, whose numbers of parameters grow linearly with the number of parent random variables. Although we showed that this model can have translation invariance in a small-scale setting, it was difficult to enlarge the scale because of the hidden variables. In this study, we extend the previous attempt by employing a variational learning method to overcome the intractability of the estimation of the massive hidden variables. We can scale the model up to learn the hand-written digit data.

抄録全体を表示

　The peripheral vestibular organs sense angular and linear accelerations of the head, and help to maintain gaze and posture during head movements through the vestibulo-ocular and vestibulo-spinal reflexes. Bilateral vestibular dysfunction causes persistent imbalance and oscillopsia while moving the body or the head, and so far no effective treatments have been found for this condition except vestibular rehabilitation. We used

galvanic vestibular stimulation ( GVS), which was delivered as zero-mean current noise, of an imperceptible amplitude to improve postural performance in patients with bilateral vestibular dysfunction. We demonstrated that the application of GVS for 30 sec significantly improved postural performance as measured by stabilometry in patients with bilateral vestibulopathy. We then examined the effect of GVS on locomotion, and showed that it significantly increased the gait velocity and stride length, but significantly decreased the stride time in patients with vestibulopathy. Furthermore, to examine the effect of GVS on vestibular function, we examined the effects of GVS on ocular vestibular evoked myogenic potentials (oVEMP) in response to bone-conducted vibration (BCV). The GVS increased the amplitude of oVEMP responses without affecting the latencies of oVEMP responses, suggesting that GVS improves static and dynamic postural stability by enhancing the function of the vestibular afferents. Finally, we tested the long-term effects of GVS on postural stability in elderly adults, and showed that GVS can lead to a postural stability improvement that lasts for several hours after the cessation of the stimulus, probably via neuroplasticity. This newly discovered effect could contribute to an increase in the application of GVS in appropriate patients.

抄録全体を表示

By means of galvanic vestibular stimulation (GVS), it might be possible to ameliorate the blunted responsiveness of degenerated neuronal circuits in patients with degenerative neurological diseases. We evaluated the effects of 24-hour GVS on the long-term heart rate dynamics in patients with multiple system atrophy and on the daytime trunk activity dynamics in patients with either levodopa-responsive Parkinson's disease or levodopa-unresponsive parkinsonism patients. Patients were also examined for cognitive performance by means of a continuous performance test. Short-range or high-frequency fluctuations of the heart rate were significantly increased by the GVS as compared with that by sham stimulation, suggestive of improved autonomic, especially parasympathetic, responsiveness. The long-range anti-persistency of trunk activity patterns probed by an autocorrelation measure was significantly increased by the GVS, suggestive of quickening of bradykinesic rest-to-active transitions. The mean reaction time in the continuous performance test was also significantly decreased by the GVS, without significant changes in either the omission or commission error ratios, which is suggestive of improved motor execution during the cognitive task. Thus, GVS improved the motor and autonomic responsiveness and is effective for ameliorating the symptoms in patients with multiple system atrophy or Parkinson's disease.

抄録全体を表示

This paper intends to report on investigations made on effects of street noise upon the learning of pupils in school rooms. 4. 357 pupils among 6 primary, 3 junior and 2 senior schools in Osaka City were selected. The situations of these schools, in which the subjects were learning, were . The following investigations were performed: (1) The psychic and somatic response of pupils on the noise caused by traffic, whistle etc. by means of questionnaire method. (2) Kraepelin's tests and syllable articulation tests in and quiet school rooms. The following results were obtained: (1) The sound levels encountered in school rooms (91 rooms) ranged 40-65 phons and in about one quarter of these rooms, the levels were 55 phons or higher while the sound levels in quiet school rooms (investigated as control) ranged 40-45 phons. (2) Among the replies by the subjects who studied in school rooms with the sound level higher than 55 phons, more than 50 % filled out "Yes (measuring, difficulty)" regarding to the items concerned with emotion and perception of speech while same answer was given by 31 % of the subjects regarding to mental works and 24 % regarding to physical conditions. (3) The statistical study of Kraepelin's tests revealed that there were no significant differences in the work amount, the percentage of errors the percentage of rest efficiency between the tests in and quiet school rooms. (4) In school rooms, with noise level higher than 55 phons, articulation of syllable decreased especially at the rear seats. (5) The noise level in school rooms, which are exposed to street noise, seems necessary to be lower than 55 phons for satisfactory learning.

抄録全体を表示

Objectives:We examined speech recognition ability of elementary school‐aged Japanese children with unilateral hearing loss under

environments and then examined the effects of the FM system fitted into their normal hearing ear on their speech recognition ability. Methods:Twelve children with severe‐to‐profound sensorineural hearing loss and ten age‐matched children with bilateral normal hearing received speech recognition test in quiet and environments. Other nine children with severe‐to‐profound sensorineural hearing loss received speech recognition test with or without the FM receiver fitted to the normal hearing ear in quiet and environments. Results:Signal‐to‐noise ratio (SNR) in Japanese elementary school classrooms was estimated to be ‐6.9 dB at a preferential seat. In environment of ‐5 dB SNR similar to working classrooms, the correct rates of speech recognition test in children with unilateral hearing loss were significantly lower, compared with those in children with bilateral normal hearing. In the environment, the correct rates in children aided by the FM system with unilateral hearing loss were significantly better, compared with unaided ones. Conclusion:The results suggested that the FM system is recommended as an audiological management for improvement of speech recognition of children with unilateral hearing loss in classrooms. J. Med. Invest. 65:216‐220, August, 2018

抄録全体を表示

This study investigated the effects of fundamental frequency (F0) separation on children's ability to recognize a target melody (the first six notes of the ‘Happy Birthday’ song) whose notes were interleaved with those of a background (distractor) melody. Children aged four, five, six, nine, ten and young adults participated in the study. Participants heard two pairs of interleaved melodies. One pair contained the correct version of the target melody; in the other pair, the 3rd or 4th notes of the target melody were shifted upwards or downwards by two semitones. Listeners were asked to select which of two pairs of interleaved melodies contained the correct version of the ‘Happy Birthday’ song, after which they received visual feedback about their response. Results showed that greater F0 separation improved performance, and that there was a small but statistically significant trend of increasing performance as age increased.

抄録全体を表示

The subjective feeling of people exposed to environmental noise is an important indicator of the assessment of noise environment. The subjective feeling under noise exposure is ordinarily surveyed on both the disturbance for health conditions and the degree of subjective noisiness or annoyance. The degree of noisiness expressed in the words for rating. If the psychophysical grade of the rating words is not measured, the interpretation on the results of survey becomes obscure. From the above mentioned viewpoint, in this paper, the psychophysical values of several rating words in the Japanese language were measured on the scale.
The scales measuring the degree or the quiet degree were divided into 10 equal parts. Nine rated adverbs were measured on 410 female students and 70 male students aged 19 to 22 years old.
These rating words are nearly identical with the following English words: extremely, very, hgghly, fairly, passably, relatively, rather, slightly, and barely.
The results were as follows;
1) The psychophysical values given to each of the rating words on or quiet scale are given in Fig., In Fig., the modes and the ranges from 25% ile to 75% ile in the distribution are shown. The intervals between each of the values given to 9 words were unequal.
2) Each of the rating words gave similar values on the scale and on the quiet scale.
3) No sexual differences were seen in the values of each word.
4) In 5 rating words, the values on the hedonic scale reported in the reference by another author were smaller than the values of 25% ile in the distribution on scale.

抄録全体を表示

The authors previously proposed a method for blindly estimating the speech transmission index (STI) based on the concept of the modulation transfer function (MTF). This method, however, over- or under-estimates STIs from observed signals in real environments due to the effect of background noise. In this paper, the proposed method was developed from the previous method to resolve the problem by simultaneously estimating the effects of inverse MTFs in and reverberant environments. Simulations were carried out to verify that the proposed method can correctly estimate STIs in reverberant environments by using reverberant AM signals. The results revealed that the proposed approach could be used to effectively estimate STIs from reverberant signals in various room acoustics.

抄録全体を表示

This paper proposes a dependable visual control method for a small-scale helicopter with a wireless camera. The camera sometimes captures images, and we cannot control the helicopter by directly using the captured images. The proposed method decides whether each captured image is or noise-free in real-time. images are not used to make the control signal.

抄録全体を表示

This paper focuses on the generalization of classifiers in

problems and aims at construction learning classifier system (LCS) that can acquire the optimal classifier subset by dynamically determining the classifier generalization criteria. In this paper, an accuracy-based LCS (XCS) that uses the mean of the reward (XCS-MR) is introduced, which can correctly identify classifiers as either accurate or inaccurate for problems, and investigates its effectiveness when used for several problems. Applying XCS and an XCS based on the variance of reward (XCS-VR) as the conventional LCSs, along with XCS-MR, to 11-multiplexer problems where the reward value changes according to a Gaussian distribution, Cauchy distribution, and lognormal distribution revealed the following: (1) XCS-VR and XCS-MR could select the correct action for every type of reward distribution; (2) XCS-MR could appropriately generalize the classifiers with the smallest amount of data; and (3) XCS-MR could acquire the optimal classifier subset in every trial for every type of reward distribution.

抄録全体を表示

The effects of FM system fitted into the normal hearing ear (NHE) or a cartilage conduction hearing aid (CCHA) fitted into the affected ear (AE) on the speech recognition ability in noise were examined in children with unilateral congenital aural atresia (UCAA). In children with bilateral normal hearing (BNH), speech recognition score (SRS) was significantly decreased in the

environment of -5 dB signal-to-noise ratio (SNR), compared with those in quiet. In children with UCAA, SRS was significantly decreased in environments of 0 and -5 dB SNR, compared with those in quiet. In environments of 0 and -5 dB SNR, SRS in children with UCAA was significantly decreased, compared those in children with BNH. In the environment of -5 dB SNR, SRS in UCAA children aided by FM system fitted into NHE was significantly better than those in unaided children in the same group. In the environment of 0 dB SNR, SRS in UCAA children aided by CCHA into AE tended to be higher than those in unaided children in the same group. FM system and CCHA can be recommended as an audiological management for the improvement of speech recognition in children with UCHL in classrooms. J. Med. Invest. 67 : 134-138, February, 2020

抄録全体を表示

Infants with cerebral palsy often have breathing, which is considered to be due to stenosis of the upper airway caused by tonic contraction of the muscles at the level of the tongue base or mesopharynx.
Recently we treated two infants with cerebral palsy, stridor and dyspnea. The first had hypertrophied adenoids; adenoidectomy was curative. The cause of the second patient's stridor and dyspnea was laryngomalacia. Tracheostomy was necessary in this case.
These two cases show that such organic diseases are often present in patients with cerebral palsy and can cause breathing and dyspnea. For the exact diagnosis, flexible fiberoscopy is very useful.

抄録全体を表示

General methods for constructing reliable electromagnetic response functions from interrupted and recordings of the naturally occurring electromagnetic variations are described. The aim is to be able to effectively utilize the 11 years of hourly mean electric and magnetic data (1932-42) from Tucson, Arizona, despite the many gaps and large extraneous values. This long-time series is extremely useful in helping one to focus on techniques that yield stable and reproducible response functions. The ultimate goal is to obtain accurate response functions in order to construct reliable earth electrical conductivity models and to determine the ocean fluid-induced part of the electromagnetic field. The Tucson electric field data is found to have diurnal and semidiurnal tidal frequencies most likely of oceanographic origin and a 6 d^-1 peak of unknown origin.

抄録全体を表示

An improved fully convolutional network based on post-processing with global variance (GV) equalization and noise-aware training (PN-FCN) for speech enhancement model is proposed. It aims at reducing the complexity of the speech improvement system, and it solves overly smooth speech signal spectrogram problem and poor generalization capability. The PN-FCN is fed with the

speech samples augmented with an estimate of the noise. In this way, the PN-FCN uses additional online noise information to better predict the clean speech. Besides, PN-FCN uses the global variance information, which improve the subjective score in a voice conversion task. Finally, the proposed framework adopts FCN, and the number of parameters is one-seventh of deep neural network (DNN). Results of experiments on the Valentini-Botinhaos dataset demonstrate that the proposed framework achieves improvements in both denoising effect and model training speed.

抄録全体を表示

　われわれは, 難治性の前庭障害患者の体平衡を改善する新たな治療の一つとして, ノイズ前庭電気刺激 (ノイズ GVS) を利用したバランス改善機器の開発を進めており, 2つの自主臨床試験を経て, 現在 AMED の委託研究 (医療機器開発研究推進事業) として, 重度のふらつきを有する難治性の前庭障害患者を対象に, ノイズ GVS のバランス改善効果と安全性を検証する医師主導治験を進めているところである.

　ノイズ GVS は, 耳後部に貼付した電極より直流電流を流すことで前庭神経を刺激する方法で, 微弱な入力信号に対する非線形の応答がノイズ様の相動的な刺激を与えることによって増強される, 確率共振現象を利用している. ノイズ GVS は, 痛みや不快感などの副作用を伴わない微弱な刺激でパフォーマンスを改善することが可能であり, 外科的侵襲を伴わず, 耳後部に貼付する表面電極と携帯型の刺激装置があれば, どこでも簡単に使用することができる, などの利点を有する.

　われわれはこれまでの研究で, ノイズ GVS は, ① 短期刺激において, 刺激中に健常者および両側前庭障害患者のバランス機能を改善すること, ② 歩行速度を増大させる作用のあること, ③ 30分以上の長期刺激においては, 刺激中のみならず刺激終了後もバランス機能の改善効果が維持される, 持越し効果があることを示してきた. 現在進めている治験は, 前庭障害に基づく重度のふらつきを有する患者を対象とした, 二重盲検ランダム化プラセボ対照試験であり, ノイズ GVS のバランス効果の改善効果を検証する pivotal study である. 本稿では, これまでのノイズ GVS の開発の経緯について概説する.

抄録全体を表示

The restoration of bone-conducted speech is a very important issue that enables robust speech communication in extremely environments. We proposed a method of blind restoration in our previous studies based on a scheme of linear prediction with a method of training and prediction based on the simple recurrent neural network. However, prediction based on neural networks is not suitable for training with large corpora, which is necessary for real applications. The over-training problem with simple recurrent neural networks makes it difficult to train various kinds of bone-conducted speech in one session. In addition, it is difficult to adapt the neural network model to bone-conducted speech in unknown environments to build an open dataset restoration of bone-conducted speech. Thus, a method of training and prediction based on the Gaussian mixture model was used in this research, instead of a neural network. A method of re-estimating the residual ratio in the scheme of linear prediction is also proposed. We also investigated how the proposed method works to restore bone-conducted speech in extremely environments. Objective and subjective evaluations were carried out to evaluate the improvements in sound quality and the intelligibility of restored speech. The results revealed that our proposed method outperformed previous methods in both human hearing and automatic speech recognition systems even in extremely environments.

抄録全体を表示

This paper presents a new relay feedback method to guarantee the accuracy of the enhanced relay feedback method. The new method uses the integrals of a process input and output to tune Proportional-Integral-Derivative (PID) controllers automatically even if the process input and output are contaminated by measurement noises and disturbances. This feedback method uses a new disturbance estimator and a new noise magnitude estimator to remove the effects of noises and disturbances on estimates of frequency response data. Simulation and application to a real system to control liquid level demonstrate that the proposed methods provide fairly accurate ultimate data of the process and that the PID controller tuned by the proposed method achieves control faster and with less overshoot than the conventional auto-tuning strategy based on the conventional relay feedback method.

抄録全体を表示