• International journal of advanced smart convergence
• /
• v.12 no.4
• /
• pp.171-176
• /
• 2023

This study is about a speech recognition error correction system designed to detect and correct speech recognition errors before natural language processing to increase the success rate of intent analysis in natural language processing with optimal efficiency in various service domains. An encoder is constructed to embedded the correct speech token and one or more error speech tokens corresponding to the correct speech token so that they are all located in a dense vector space for each correct token with similar vector values. One or more utterance tokens within a preset Manhattan distance based on the correct utterance token in the dense vector space for each embedded correct utterance token are detected through an error detector, and the correct answer closest to the detected error utterance token is based on the Manhattan distance. Errors are corrected by extracting the utterance token as the correct answer.

• International journal of advanced smart convergence
• /
• v.13 no.2
• /
• pp.16-24
• /
• 2024

With an increase in the relevance of next-generation integrated networking environments, the need to effectively utilize advanced networking techniques also increases. Specifically, integrating Software-Defined Networking (SDN) with Multi-access Edge Computing (MEC) is critical for enhancing network flexibility and addressing challenges such as security vulnerabilities and complex network management. SDN enhances operational flexibility by separating the control and data planes, introducing management complexities. This paper proposes a reinforcement learning-based network path optimization strategy within SDN environments to maximize performance, minimize latency, and optimize resource usage in MEC settings. The proposed Enhanced Proximal Policy Optimization (PPO)-based scheme effectively selects optimal routing paths in dynamic conditions, reducing average delay times to about 60 ms and lowering energy consumption. As the proposed method outperforms conventional schemes, it poses significant practical applications.

• International journal of advanced smart convergence
• /
• v.13 no.2
• /
• pp.80-87
• /
• 2024

This paper focuses on improving accuracy in constrained computing settings by employing the ReLU (Rectified Linear Unit) activation function. The research conducted involves modifying parameters of the ReLU function and comparing performance in terms of accuracy and computational time. This paper specifically focuses on optimizing ReLU in the context of a Multilayer Perceptron (MLP) by determining the ideal values for features such as the dimensions of the linear layers and the learning rate (Ir). In order to optimize performance, the paper experiments with adjusting parameters like the size dimensions of linear layers and Ir values to induce the best performance outcomes. The experimental results show that using ReLU alone yielded the highest accuracy of 96.7% when the dimension sizes were 30 - 10 and the Ir value was 1. When combining ReLU with the Adam optimizer, the optimal model configuration had dimension sizes of 60 - 40 - 10, and an Ir value of 0.001, which resulted in the highest accuracy of 97.07%.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.9 no.5
• /
• pp.139-146
• /
• 2009

Parametric array for audio applications is analyzed by numerical modeling and analytical approximation. The nonlinear wave equations are used to provide design guidelines for the audio parametric array. A time domain finite difference code that accurately solves the KZK (Khokhlov-Zabolotskaya-Kuznetsov) nonlinear parabolic wave equation is used to predict the response of the parametric array. The time domain code relates the source size and the carrier frequency to the audible signal response including the output level and beamwidth to considering the implementation issues for audio applications of the parametric array, the emphasis is given to the frequency response and distortion. We use the time domain code to find out the optimal parameters that will help produce the parametric array with highest achievable output in terms of the average power within the demodulated signal. Parameters such as primary input frequency, audio source radius and the modulation method are given utmost importance. The output effect of those parameters are demonstrated through the numerical simulation.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.16 no.4
• /
• pp.89-96
• /
• 2016

In this paper, we study mobile data collector (MDC) based data-gathering schemes in wireless sensor networks. In Such networks, MDCs are used to collect data from the environment and transfer them to the sink. The majority of existing data-gathering schemes suffer from high data-gathering latency because they use only a single MDC. Although some schemes use multiple MDCs, they focus on maximizing network lifetime rather than minimizing packet delay. In order to address the limitations of existing schemes, this paper focuses on minimizing packet delay for given number of MDCs and minimizing the number of MDCs for a given delay bound of packets. To achieve the minimum packet delay and minimum number of MDCs, two optimization problems are formulated, and traveling distance and traveling time of MDCs are estimated. The interior-point algorithm is used to obtain the optimal solution for each optimization problem. Numerical results and analysis are presented to validate the proposed method.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.15 no.4
• /
• pp.185-190
• /
• 2015

In this paper we characterized and designed the waveguide antipodal finline transition at 57-65GHz frequency band in V-band for 5G mobile communication systems. Especially, we proposed the design method of spline taper for finline tapers by means of increasing curvature from linear taper. We could perform optimization more effectively by excluding improper regions for optimal performance from optimization using the method. Return losses and insertion losses of antipodal finline transitions were mainly affected by the taper shape of the finline. The resonances in the structure of the finline transition were the strongest enemies who deteriorate the performance of the transition. And we alleviated the resonances using semicircle shaped patch. The designed antipodal finline transition showed good performance as it showed less than -24.2dB of return loss and -0.24dB of insertion loss in the band(57-65GHz) which we suppose to use.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.11 no.2
• /
• pp.213-218
• /
• 2011

In this paper, we have developed a tongue diagnosis system using fuzzy rules. A healthy person's tongue is red in color and has less tongue coating. However, when a person suffers from a disease, the color of their tongue changes from red to white, blue, or black. Therefore, it can analyze patient's health if analyze color and coated tongue of tongue. Medical diagnosis system can automatically determines the symptoms of the disease of a patient and their and calculate the optimal acupuncture time on the basis of the patient's physical conditions, illness conditions, and age from any place and at any time. The computer simulation results have shown that electro-acupuncture administered by using the medical diagnosis system developed in this study is more effective than the conventional method.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.15 no.1
• /
• pp.147-155
• /
• 2015

In this paper, we consider a mobile sensor network for monitoring a polluted area where human beings cannot access. Due to the limited sensing range of individual unmanned vehicles, they need to cooperate to achieve an effective sensing coverage and move to a more polluted region. In order to address the limitations of sensing and communication ranges, we propose a hazardous substance monitoring network based on virtual force algorithms, and develop a testbed. In the considered monitoring network, each unmanned vehicle achieves an optimal coverage and move to the highest interest area based on neighboring nodes sensing values and locations. By using experiments based on the developed testbed, we show that the proposed monitoring network can autonomously move toward a more polluted area and obtain a high weighted coverage.

• International Journal of Internet, Broadcasting and Communication
• /
• v.9 no.2
• /
• pp.70-77
• /
• 2017

Where there are all the flowers, the songs of all kinds of insects and birds are put in, the sunshine and shadows flicker The forest through which the water sound flows is an optimum resting space. All living creatures in these spaces will awaken the five senses of humans and perhaps turn the sensibility index (EQ). The forest meditation in the forest, which can be an optimal shelter for the people who need it, needs to feel the reverence of nature, to refine emotions, to be a self-reflection, to have a mind to respect, Have an important meaning. In this paper, we tried to consider the cause of the influence of forest sounds on human hearing from the acoustical aspect. The type of sound source of forest was divided into four seasons of spring, summer, autumn, winter. And the change in the duration of the sound during the four seasons, so that the general characteristics of the sounds of the four seasons are as follows: It can be seen that the change in the ratio of sub-band energy is almost equal to the change in dB in frequency of the equal-light curve. To compare this phenomenon, the criterion for changing the sound duration of each forest is natural The main forms of the luminance curve, such as the change in the duration of the white signal in the sound, are determined by the minimum, maximum audible frequency and the most sensitive frequency band, and the auditory characteristics of the other three inflection points Determines the overall shape of the equal-light curve.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.10 no.5
• /
• pp.265-270
• /
• 2010

Reinforcement learning performs learning through interacting with trial-and-error in dynamic environment. Therefore, in dynamic environment, reinforcement learning method like TD-learning and TD(${\lambda}$)-learning are faster in learning than the conventional stochastic learning method. However, because many of the proposed reinforcement learning algorithms are given the reinforcement value only when the learning agent has reached its goal state, most of the reinforcement algorithms converge to the optimal solution too slowly. In this paper, we present GDRLS algorithm for finding the shortest path faster in a maze environment. GDRLS is select the candidate states that can guide the shortest path in maze environment, and learn only the candidate states to find the shortest path. Through experiments, we can see that GDRLS can search the shortest path faster than TD-learning and TD(${\lambda}$)-learning in maze environment.