• Journal of Communications and Networks
• /
• v.8 no.3
• /
• pp.297-305
• /
• 2006

High-rate wireless personal area network (HR-WPAN) has been standardized by the IEEE 802.15.3 task group (TG). To support multimedia services, the IEEE 802.15.3 TG adopts a time-slotted medium access control (MAC) protocol controlled by a central device. In the time division multiple access (TDMA)-based wireless packet networks, the packet scheduling algorithm plays a key role in quality of service (QoS) provisioning for multimedia services. In this paper, we propose an adaptive cross-layer packet scheduling method for the TDMA-based HR-WPAN. Physical channel conditions, MAC protocol, link layer status, random traffic arrival, and QoS requirement are taken into consideration by the proposed packet scheduling method. Performance evaluations are carried out through extensive simulations and significant performance enhancements are observed. Furthermore, the performance of the proposed scheme remains stable regardless of the variable system parameters such as the number of devices (DEVs) and delay bound.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.16 no.3
• /
• pp.401-406
• /
• 2021

Recently, mean-teacher models based on convolutional recurrent neural networks are popularly used in audio event detection. The mean-teacher model is an architecture that consists of two parallel CRNNs and it is possible to train them effectively on the weakly-labelled and unlabeled audio data by using the consistency learning metric at the output of the two neural networks. In this study, we tried to improve the performance of the mean-teacher model by using additional derivative features of the log-mel spectrum. In the audio event detection experiments using the training and test data from the Task 4 of the DCASE 2018/2019 Challenges, we could obtain maximally a 8.1% relative decrease in the ER(Error Rate) in the mean-teacher model using proposed derivative features.

• Journal of Biomedical Engineering Research
• /
• v.44 no.6
• /
• pp.363-376
• /
• 2023

According to the COVID-19, development of various medical software based on IoT(Internet of Things) was accelerated. Especially, interest in a central software system that can remotely monitor and control ventilators is increasing to solve problems related to the continuous increase in severe COVID-19 patients. Since medical device software is closely related to human life, this study aims to develop central monitoring system that can remotely monitor and control multiple ventilators in compliance with medical device software development standards and to verify performance of system. In addition, to ensure the safety and reliability of this central monitoring system, this study also specifies risk management requirements that can identify hazardous situations and evaluate potential hazards and confirms the implementation of cybersecurity to protect against potential cyber threats, which can have serious consequences for patient safety. As a result, we obtained medical device software manufacturing certificates from MFDS(Ministry of Food and Drug Safety) through technical documents about performance verification, risk management and cybersecurity application.The purpose of this study is to conduct a usability assessment to ensure that ergonomic design has been applied so that the ventilator central monitoring system can improve user satisfaction, efficiency, and safety. The rapid spread of COVID-19, which began in 2019, caused significant damage global medical system. In this situation, the need for a system to monitor multiple patients with ventilators was highlighted as a solution for various problems. Since medical device software is closely related to human life, ensuring their safety and satisfaction is important before their actual deployment in the field. In this study, a total of 21 participants consisting of respiratory staffs conducted usability test according to the use scenarios in the simulated use environment. Nine use scenarios were conducted to derive an average task success rate and opinions on user interface were collected through five-point Likert scale satisfaction evaluation and questionnaire. Participants conducted a total of nine use scenario tasks with an average success rate of 93% and five-point Likert scale satisfaction survey showed a high satisfaction result of 4.7 points on average. Users evaluated that the device would be useful for effectively managing multiple patients with ventilators. However, improvements are required for interfaces associated with task that do not exceed the threshold for task success rate. In addition, even medical devices with sufficient safety and efficiency cannot guarantee absolute safety, so it is suggested to continuously evaluate user feedback even after introducing them to the actual site.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.12
• /
• pp.485-496
• /
• 2018

So far, the deep learning, a field of artificial intelligence, has achieved remarkable results in solving problems from unstructured data. However, it is difficult to comprehensively judge situations like humans, and did not reach the level of intelligence that deduced their relations and predicted the next situation. Recently, deep neural networks show that artificial intelligence can possess powerful relational reasoning that is core intellectual ability of human being. In this paper, to analyze and observe the performance of Relation Networks (RN) among the neural networks for relational reasoning, two types of RN-based deep neural network models were constructed and compared with the baseline model. One is a visual question answering RN model using Sort-of-CLEVR and the other is a text-based question answering RN model using bAbI task. In order to maximize the performance of the RN-based model, various performance improvement experiments such as hyper parameters tuning have been proposed and performed. The effectiveness of the proposed performance improvement methods has been verified by applying to the visual QA RN model and the text-based QA RN model, and the new domain model using the dialogue-based LL dataset. As a result of the various experiments, it is found that the initial learning rate is a key factor in determining the performance of the model in both types of RN models. We have observed that the optimal initial learning rate setting found by the proposed random search method can improve the performance of the model up to 99.8%.

• Design Convergence Study
• /
• v.16 no.5
• /
• pp.177-190
• /
• 2017

This research aims to suggest a multimodal non-touch gesture interface design to improve the usability of 3D sculpting task. The task and procedure of design sculpting of users were analyzed across multiple circumstances from the physical sculpting to computer software. The optimal body posture, design process, work environment, gesture-task relationship, the combination of natural hand gesture and arm movement of designers were defined. The preliminary non-touch 3D S/W were also observed and natural gesture interaction, visual metaphor of UI and affordance for behavior guide were also designed. The prototype of gesture based 3D sculpting system were developed for validation of intuitiveness and learnability in comparison to the current S/W. The suggested gestures were proved with higher performance as a result in terms of understandability, memorability and error rate. Result of the research showed that the gesture interface design for productivity system should reflect the natural experience of users in previous work domain and provide appropriate visual - behavioral metaphor.

• Proceedings of the ESK Conference
• /
• 1998.04a
• /
• pp.167-172
• /
• 1998

Methods of evaluating the physiological activity of the living body are EEG, EOG, Heart rate, and Rectal temperature, etc. In the study of Hagiwara and Araki(1993), they found positive correlations among performance test, physiological measurement, and subjective feeling measurement. Human alertness of pilot directly influences on the flight performance that accomplishes a lot of vigilance task and procedure execution in flight. Accordingly, this paper deals with the quantitative and objective performance test based on tracking error and reaction time by means of the new computer test program into which the perception-motion system of human beings is applies. Throughout this experiment using performance thst, the results suggest that performance capability in state of sleep deprivation 2 hours and alcoholic 0.05 .apaprox. 0.06% in blood were more impaired than one in a normal state, and they further showed statistically significant differences between them, which were influenced by impairment factors of body regulation and pilot's grade. We also obtained the prediction value and the 95% confidence interval of tracking error and reaction time at the normal state for the purpose of distinguishing performance capability between the normal state and the abnormal state. And it is ecpected that the evaluation of human alertness using performance test will be applied to the quantitative assessment of an each pilot's realistic consciousness/attention, and will lead a flight commander to the accurate decision of mission approval prior to a flight.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.2
• /
• pp.139-147
• /
• 2021

This paper introduces and implements a Sound Event Detection (SED) system based on weakly-supervised learning where only part of the data is labeled, and analyzes the effect of parameters. The SED system estimates the classes and onset/offset times of events in the acoustic signal. In order to train the model, all information on the event class and onset/offset times must be provided. Unfortunately, the onset/offset times are hard to be labeled exactly. Therefore, in the weakly-supervised task, the SED model is trained by "strongly labeled data" including the event class and activations, "weakly labeled data" including the event class, and "unlabeled data" without any label. Recently, the SED systems using the mean-teacher model are widely used for the task with several parameters. These parameters should be chosen carefully because they may affect the performance. In this paper, performance analysis was performed on parameters, such as the feature, moving average parameter, weight of the consistency cost function, ramp-up length, and maximum learning rate, using the data of DCASE 2020 Task 4. Effects and the optimal values of the parameters were discussed.

• The Journal of the Acoustical Society of Korea
• /
• v.38 no.5
• /
• pp.593-600
• /
• 2019

The similarity in tones between speakers can lower the performance of speaker verification. To improve the performance of speaker verification systems, we propose a multi-task learning technique using deep neural network to learn speaker information and age information. Multi-task learning can improve generalization performances, because it helps deep neural networks to prevent hidden layers from overfitting into one task. However, we found in experiments that learning of age information does not work well in the process of learning the deep neural network. In order to improve the learning, we propose a method to dynamically change the objective function weights of speaker identification and age estimation in the learning process. Results show the equal error rate based on RSR2015 evaluation data set, 6.91 % for the speaker verification system without using age information, 6.77 % using age information only, and 4.73 % using age information when weight change technique was applied.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.32 no.3A
• /
• pp.230-236
• /
• 2007

This paper presents a resource management mechanism for the mesh networking in IEEE 802.15.3 High-rate WPAN. IEEE 802.15 TGS is standardizing the MAC and PHY for mese networking. This task group researches the mechanism that are extension of network coverage without increasing transmit power of receive sensitivity, and studies the enhanced reliability via route redundancy. In this paper we propose the distributed resource management scheme that is fairly using the channel resource in the piconet without centralized piconet coordinator. Each DEV reserves the channel time and broadcasts its information. This scheme has unfairness for later associated DEV because of preoccupation of earlier associated DEVs. This paper presents the method that fairly allocates the channel time in MAC layer. And we evaluate the performance enhancement using simple simulations.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.30B no.2
• /
• pp.69-77
• /
• 1993

Automatic image recognition system is essential for a better man-to machine interaction. Because of the noise and deformation due to the sensor operation, it is not simple to build an image recognition system even for the fixed images. In this paper neural network which has been reported to be adequate for pattern recognition task is applied to the fixed and variational(rotation, size, position variation for the fixed image)recognition with a hope that the problems of conventional pattern recognition techniques are overcome. At fixed image recognition system. ART model is trained with face images obtained by camera. When recognizing an matching score. In the test when wigilance level 0.6 - 0.8 the system has achievel 100% correct face recognition rate. In the variational image recognition system, 65 invariant moment features sets are taken from thirteen persons. 39 data are taken to train multi-layer perceptron and other 26 data used for testing. The result shows 92.5% recognition rate.