• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.8 no.2
• /
• pp.483-503
• /
• 2014

In this paper, human action recognition using pyramid histograms of oriented gradients and collaborative multi-task learning is proposed. First, we accumulate global activities and construct motion history image (MHI) for both RGB and depth channels respectively to encode the dynamics of one action in different modalities, and then different action descriptors are extracted from depth and RGB MHI to represent global textual and structural characteristics of these actions. Specially, average value in hierarchical block, GIST and pyramid histograms of oriented gradients descriptors are employed to represent human motion. To demonstrate the superiority of the proposed method, we evaluate them by KNN, SVM with linear and RBF kernels, SRC and CRC models on DHA dataset, the well-known dataset for human action recognition. Large scale experimental results show our descriptors are robust, stable and efficient, and outperform the state-of-the-art methods. In addition, we investigate the performance of our descriptors further by combining these descriptors on DHA dataset, and observe that the performances of combined descriptors are much better than just using only sole descriptor. With multimodal features, we also propose a collaborative multi-task learning method for model learning and inference based on transfer learning theory. The main contributions lie in four aspects: 1) the proposed encoding the scheme can filter the stationary part of human body and reduce noise interference; 2) different kind of features and models are assessed, and the neighbor gradients information and pyramid layers are very helpful for representing these actions; 3) The proposed model can fuse the features from different modalities regardless of the sensor types, the ranges of the value, and the dimensions of different features; 4) The latent common knowledge among different modalities can be discovered by transfer learning to boost the performance.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.10
• /
• pp.221-230
• /
• 2020

Elevators, an important vertical means of transportation used daily, have enriched the city into a state-of-the-art giant skyscraper forest. The purpose of this study is to analyze problems about elevator safety and present engineering and legal improvement measures. The research method adopted the important engineering elements that make up elevator safety and the elevator safety law and the method of analyzing the papers already studied. Based on his experience as a policy advisor to the Korea Elevator Safety Authority, he analyzed the problems of the related statutes, focusing on the construction, electrical, electronic, information and communication, and the elevator safety law, which comprise the elevator safety system. As a result of the research, it was suggested that the so-called legal engineering system should be established through the convergence of laws and disciplines related to elevator safety. This study is expected to be an alternative to establishing an engineering and legal convergence system for elevator safety in the future.

• Bulletin of the Korean Chemical Society
• /
• v.34 no.7
• /
• pp.2023-2028
• /
• 2013

A kinetic study is reported for the nucleophilic substitution reactions of phenyl Y-substituted-phenyl carbonates (5a-5k) with piperidine in 80 mol % $H_2O$/20 mol % DMSO at $25.0{\pm}0.1^{\circ}C$. The plots of $k_{obsd}$ vs. [piperidine] for the reactions of substrates possessing a strong electron-withdrawing group (EWG) in the leaving group (i.e., 5a-5i) are linear and pass through the origin. In contrast, the plots for the reactions of substrates bearing a weak EWG or no substituent (i.e., 5j or 5k) curve upward, indicating that the electronic nature of the substituent Y in the leaving group governs the reaction mechanism. Thus, it has been suggested that the reactions of 5a-5i proceed through a stepwise mechanism with a zwitterionic tetrahedral intermediate (i.e., $T^{\pm}$) while those of 5j and 5k proceed through a stepwise mechanism with two intermediates (i.e., $T^{\pm}$ and its deprotonated form $T^-$). The slope of the Br${\o}$nsted-type plot for the second-order rate constants (i.e., $k_N$ or $Kk_2$) changes from -0.41 to -1.89 as the leaving-group basicity increases, indicating that a change in the rate-determining step (RDS) occurs. The reactions of 5a-5k with piperidine result in larger $k_1$ values than the corresponding reactions with ethylamine.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.39 no.2
• /
• pp.157-165
• /
• 2002

This paper presents the implementation of the speed control system for 3 phase induction motor using PD controller and neural networks. The PD controller is used to control the motor and to train neural networks at the first time. And neural networks are widely used as controllers because of a nonlinear mapping capability, we used feedforward neural networks(FNN) in order to simply design the speed control system of the 3 phase induction motor. Neural networks are tuned online using the speed reference, actual speed measured from an encoder and control input current to motor. PD controller and neural networks are applied to the speed control system for 3 phase induction motor, are compared with PI controller through computer simulation and experiment respectively. The results are illustrated that the output of the PD controller is decreased and feedforward neural networks act main controller, and the proposed hybrid controllers show better performance than the PI controller in abrupt load variation and the precise control is possible because the steady state error can be minimized by training neural networks.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.46 no.1
• /
• pp.59-67
• /
• 2009

In this paper, we have proposed the structure of the IMADF(improved modified multiplication-free adaptive filter) to cancel the adaptive noise in biomedical signals. The IMADF structure use the one-step predicted filter in the multiplication-free adaptive digital filter(MADF) structure using the DPCM and Sign algorithm. And then we use the heart phantom model based on the magnetocardiographic (MCG) to test the biomedical signals and analyze the signal of it. Their functions of the heart phantom occur from the multidipole current source. This can play role the same in the real function of the human heart to study it. In the experimental results, the IMADF algorithm has reduced the computational complexity by use of only the addition operation without a multiplier. Also, under the condition of identical stationary-state error, it could obtain the stabled convergence characteristics that the IMADF algorithm is almost same as the sign algorithm, but is better than the MADF algorithm. Here, this algorithm has effective characteristics when the correlation of the input signal is highly.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.8
• /
• pp.126-135
• /
• 2014

The Euclidean distance between error probability density functions (EDEP) has been used as a performance criterion for supervised adaptive signal processing in impulsive noise environments. One of the drawbacks of the EDEP algorithm is a heavy computational complexity due to the double summation operations at each iteration time. In this paper, a recursive method to reduce its computational burden in the estimation of the EDEP and its gradient is proposed. For the data block size N, the computational complexity for the estimation of the EDEP and its gradient can be reduced to O(N) by the proposed method, while the conventional estimation method has $O(N^2)$. In the performance test, the proposed EDEP and its gradient estimation yield the same estimation results in the steady state as the conventional block-processing method. The simulation results indicates that the proposed method can be effective in practical adaptive signal processing.

• Journal of the Korean Society of Industry Convergence
• /
• v.20 no.4
• /
• pp.333-338
• /
• 2017

In recent years, there has been an increasing demand for healthcare services that can periodically monitor health status and maintain health by increasing the weight training population. However, injuries in the absence of trainer are increasing with the increase in the number of members in the fitness training center. Therefore, there is a need for a system that can periodically monitor the user's exercise state and assist in systematic and safe exercise even when the trainer is absent. In this study, we developed an auto training system that can effectively manage the exerciser while supporting the strength movement. The auto training system consists of a cable mount module, a control module, and a training information monitoring module. In order to evaluate performance of the developed system, the assistant force tests are carried out. Experimental results showed that the assistant force works well when the exerciser is out of power.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2016.02a
• /
• pp.170.1-170.1
• /
• 2016

$Bi_2Te_3$ has long been studied for its excellent thermoelectric characteristics. Recently, this material has been known as a topological insulator (TI). The surface states within the bulk band gap of a TI, which are protected by the time reversal symmetry, contribute to the conduction at the surface, while the bulk is in insulating state. In contrast to the bulk defects tuning the chemical potential to the Dirac energy, the native defects near the surface are expected not to change the shape of the Fermi surface and the related spin structure. Using scanning tunneling microscopy (STM), we have systematically characterized surface or near surface defects in p- and n- doped $Bi_2Te_3$, and identified their structure by first principles calculations. In addition, bias-polarity dependences of STM images revealed the electron donor/acceptor nature of each defect. A detailed theoretical study of the surface states near the Dirac energy reveals the robustness of the Dirac point, which verifies the effectiveness of the disturbance on the backscattering from various kinds of defects.

• Journal of Surface Science and Engineering
• /
• v.50 no.3
• /
• pp.219-224
• /
• 2017

The K gettering in $SiO_2/PSG/SiO_2/Al-1%Si$ multilevel thin films was investigated using SIMS(secondary ion mass spectrometry) and XPS(X-ray Photoelectron Spectroscopy) analysis. DC magnetron sputter techniques and APCVD(atmosphere pressure chemical vapor deposition) were utilized for the deposition of Al-1%Si thin films and $SiO_2/PSG/SiO_2$ passivations, respectively. Heat treatment was carried out at $400^{\circ}C$ for 5 h in air. SIMS depth profiling was used to determine the distribution of K, Al, Si, P and other elements throughout the $SiO_2/PSG/SiO_2/Al-1%Si$ multilevel thin films. XPS was used to analyze binding energies of Si and P elements in PSG passivation layers. K peaks were observed throughout the $PSG/SiO_2$ passivation layers on the Al-1%Si thin films and especially at the $PSG/SiO_2$ interfaces. K gettering in $SiO_2/PSG/SiO_2/Al-1%Si$ multilevel thin films is considered to be caused by a segregation type of gettering. The chemical state of Si and P elements in PSG passivation appears to be $SiO_2$ and $P_2O_5$, respectively

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.40 no.11
• /
• pp.705-715
• /
• 2016

In this study, the cycle performance of an air conditioner with multi-indoor units is analyzed and simulated. The cycle performance could be predicted through the integration of mathematical formulation for these devices. The condenser pressure is obtained by an iteration process to match the mass flow rates of the compressor and the expansion valve and the evaporator pressure is determined by an iteration process, in which the suction super heat is tracing the targeted super heat. The required software was developed by system programming. the software algorithm is extended to predict the cycle performance of an air conditioner system with multi-indoor units, and then the numerical results are compared with experimental results. This mathematical model is validated from the result of experiments conducted on 8.3kW air conditioner. The errors in capacity, electronic power, and COP are found to be within 10% in general.