• 대한임베디드공학회논문지
• /
• 제18권6호
• /
• pp.319-325
• /
• 2023

In this study, we proposed a method for hyperparameter optimization in the building and training of a deep learning model designed to process point cloud data collected by a millimeter-wave radar system. The primary aim of this study is to facilitate the deployment of a baseline model in resource-constrained IoT devices. We evaluated a RadHAR baseline deep learning model trained on a public dataset composed of point clouds representing five distinct human activities. Additionally, we introduced a coarse-to-fine hyperparameter optimization procedure, showing substantial potential to enhance model efficiency without compromising predictive performance. Experimental results show the feasibility of significantly reducing model size without adversely impacting performance. Specifically, the optimized model demonstrated a 3.3% improvement in classification accuracy despite a 16.8% reduction in number of parameters compared th the baseline model. In conclusion, this research offers valuable insights for the development of deep learning models for resource-constrained IoT devices, underscoring the potential of hyperparameter optimization and model size reduction strategies. This work contributes to enhancing the practicality and usability of deep learning models in real-world environments, where high levels of accuracy and efficiency in data processing and classification tasks are required.

• 산업공학
• /
• 제3권1호
• /
• pp.13-20
• /
• 1990

The application of human-factor concept in developing a complex weapon system is important for system engineer to determine the system performance and reliability. This paper describes the evaluation procedure of human factors in the X-system wherein the evaluation result provides a better performance than the previous model in operatability and maintainability. The criteria used for the evaluation of a design adequacy of the X-system are based on the military human-factor standards.

• 산업경영시스템학회지
• /
• 제33권4호
• /
• pp.85-99
• /
• 2010

This study is to identify the present condition and level of human resource administration in domestic profit and non-profit organizations, and to investigate on the relation between human resource management and management performance in enterprise. This study is to make items related to the human resource management of American Malcolm Baldrige Award's level and management performance researching questions, investigates seven patterns of organs surveyed positively. The results are as follows: Firstly, the profit organs like manufacturing companies shows more positive than non-profit in comparison with human resource management of each organs. And a local government and public enterprise which is non-profit show to recognize human resource management is insufficient on th whole. Secondly, perception level of a medical institution, educational institution and service industry about aromaticity of human resource management by global standard shows to be positive, but the local government negative. Thirdly, the profit organ in the recognition of human resource management about details practice too, shows to be positive, while non-profit negative. Fourth, the result which compares and analyzes management performance between the similar industry overall show to recognize positive, but public enterprise negative about product, service performance and human resource management. Fifthly, the details practice of human resource administration shows to influence meaningly to all management performance. Therefore, all organizations will positively confront human resource management, make the circumstance of organization through systematic program, and promote the management performance of the organization.

• 한국시뮬레이션학회논문지
• /
• 제31권1호
• /
• pp.11-18
• /
• 2022

최근 공공장소 및 시설에서 범죄예방 및 시설 안전을 목적으로 영상정보 기반의 인체의 행위를 분류하는 연구가 활발히 진행되고 있다. 이러한 인체 행위분류의 성능을 향상하기 위해서 대부분의 연구는 전이학습 기반의 딥러닝을 적용하고 있다. 그러나 딥러닝의 기반이 되는 중추 네트워크 모델(Backbone Network Model)의 수가 증가하고 아키텍처가 다양해짐에도 불구하고, 소수의 모델만 사용하는 분위기 때문에 운용목적에 적합한 중추 네트워크 모델을 찾는 연구는 미흡한 실정이다. 본 연구는 영상정보를 기초로 인체 행위를 분류하는 인공지능 모델을 개발하기 위해 최근에 개발된 5가지의 딥러닝 중추 네트워크 모델을 대상으로 전이학습을 적용하고 각 모델의 정확도 및 학습효율 측면에서 비교 및 분석하여 가장 효율이 높은 모델을 제안하였다. 이를 위해, 기본적인 인체 행위가 아닌 운동 종목 기반의 활동적이고 신체접촉이 높은 12가지의 인체 활동을 선정하고 관련된 7,200개의 이미지를 수집하였으며, 5가지의 중추 네트워크 모델에 총 20회의 전이학습을 균등하게 적용하고 학습과정과 결과성능을 통해 인체 행위를 분류하는데 적합한 중추 네트워크 모델을 정량적으로 비교 및 분석하였다. 그 결과 XceptionNet 모델이 학습 및 검증 정확도에서 0.99 및 0.91로, Top 2 및 평균 정밀도에서 0.96 및 0.91로 나타났으며 학습 소요시간은 1,566초, 모델용량의 크기는 260.4MB로 정확도와 학습효율 측면에서 다른 모델보다 높은 성능이 나타남을 확인할 수 있었다. 이러한 결과는 전이학습을 적용하여 인체 행위분류를 진행하는 다양한 연구 분야에 활용되기를 기대한다.

• 한국경영과학회:학술대회논문집
• /
• 대한산업공학회/한국경영과학회 1995년도 춘계공동학술대회논문집; 전남대학교; 28-29 Apr. 1995
• /
• pp.787-796
• /
• 1995

A man model is a useful design tool for the evaluation of man machine systems and products. An arm reach trajectory prediction for such a model will be specifically useful to present human activities and, consequently, could increase the accuracy and reality of the evaluation. In this study, a three-dimensional reach trajectory prediction model was developed using an inverse kinematics technique. The upper body was modeled as a four link open kinematic chain with seven degrees of freedom. The Resolved Motion Method used for the robot kinematics problem was used to predict the joint movements. The cost function of the perceived discomfort developed using the central composite design was also used as a performance function. This model predicts the posture by moving the joints to minimize the discomfort on the constraint of the end effector velocity directed to a target point. The results of the pairwise t-test showed that all the joint coordinates except the shoulder joint's showed statistically no differences at .alpha. = 0.01. The reach trajectory prediction model developed in this study was found to accurately simulate human arm reach trajectory and the model will help understand the human arm reach movement.

• 반도체디스플레이기술학회지
• /
• 제18권1호
• /
• pp.97-101
• /
• 2019

Understanding and classification of the human's emotion play an important tasks in interacting with human and machine communication systems. This paper proposes a novel emotion recognition method by extracting facial keypoints, which is able to understand and classify the human emotion, using active Appearance Model and the proposed classification model of the facial features. The existing appearance model scheme takes an expression of variations, which is calculated by the proposed classification model according to the change of human facial expression. The proposed method classifies four basic emotions (normal, happy, sad and angry). To evaluate the performance of the proposed method, we assess the ratio of success with common datasets, and we achieve the best 93% accuracy, average 82.2% in facial emotion recognition. The results show that the proposed method effectively performed well over the emotion recognition, compared to the existing schemes.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제5권3호
• /
• pp.246-252
• /
• 2005

Intelligent Space is a space where many sensors and intelligent devices are distributed. Mobile robots exist in this space as physical agents, which provide human with services. To realize this, human and mobile robots have to approach each other as much as possible. Moreover, it is necessary for them to perform interactions naturally. It is desirable for a mobile robot to carry out human affinitive movement. In this research, a mobile robot is controlled by the Intelligent Space through its resources. The mobile robot is controlled to follow walking human as stably and precisely as possible. In order to follow a human, control law is derived from the assumption that a human and a mobile robot are connected with a virtual spring model. Input velocity to a mobile robot is generated on the basis of the elastic force from the virtual spring in this model. And its performance is verified by the computer simulation and the experiment.

• The Journal of Asian Finance, Economics and Business
• /
• 제7권10호
• /
• pp.391-399
• /
• 2020

This study aims to analyze the relationship of e-HRM implementation to employee performance both directly and indirectly through the intervening of the Human Resource service Quality variable, both practically and theoretically. This study uses variance-based structural equation modeling (SEM) techniques with partial least square (PLS) statistical testing tools to test the direct relationship of e-HRM and the performance and relationship moderated by Human Resources service quality tested on 200 civil servants in five offices under the coordination of the Government of the South Sulawesi Province of Indonesia. The data collection model in this study uses an online survey. The data analysis stages through the explanatory concept consist of, first, the interpretation of the distribution of the average frequency of respondents' answers; second, outer-loading; third, determination of the validity and reliability; fourth, the coefficient of determination test and partial test; fifth, the GoF model; sixth, validity test; and seventh, hypothesis testing. This study explores four hypotheses in a comprehensive fashion; the results of this study show that all hypotheses have positive and significant effects both through direct and intervening relationships. Among the three direct relationships, the relationship of e-HRM variables on HR Service Quality is greatest and most dominant.

• Journal of Information Processing Systems
• /
• 제14권1호
• /
• pp.150-161
• /
• 2018

We propose a novel deep neural network model for detecting human activities in untrimmed videos. The process of human activity detection in a video involves two steps: a step to extract features that are effective in recognizing human activities in a long untrimmed video, followed by a step to detect human activities from those extracted features. To extract the rich features from video segments that could express unique patterns for each activity, we employ two different convolutional neural network models, C3D and I-ResNet. For detecting human activities from the sequence of extracted feature vectors, we use BLSTM, a bi-directional recurrent neural network model. By conducting experiments with ActivityNet 200, a large-scale benchmark dataset, we show the high performance of the proposed DeepAct model.

• 대한산업공학회지
• /
• 제30권1호
• /
• pp.17-26
• /
• 2004

In this study, the usability evaluation of a menu-structure was performed using spreading activation model with respect to human's memory retrieval. Spreading Activation Model is effectively used to understand the process of information retrieval, so it can be used as a theoretical background for modeling of the process of human's information retrieval. For spreading activation test (SAT), subjects were presented with 67 pairs of menu titles, which consist of a menu title in the high level menu item and a menu title for the next lower level menu item, from Korea University's web site. For performance tests, three scenarios were developed with longer reaction times and ambiguous associations found in the SAT to reflect the existing problems of the website. As a result, the SAT was found to bean effective tool to enhance the website usability because the SAT could bea substitute for the performance test with a high correlation $({\rho}=0.735,\;{\alpha}=0.05)$. After remaining menu titles with slow reaction times and ambiguous associations found in SAT, the website usability was significantly improved with faster reaction times and less ambiguous associations proven with smaller number of web-page visits. Therefore, the SAT could be used as a methodology to design and evaluate the user-centered menu structure related to human's association structure.