• The Bulletin of The Korean Astronomical Society
• /
• v.38 no.2
• /
• pp.38.1-38.1
• /
• 2013

We present the Yonsei evolutionary population synthesis (YEPS) models based on the high-resolution empirical spectral energy distributions (SEDs). We have adopted the MILES library in the optical wavelength, and our new models based on the MILES library show good agreements with our previous models presented in the YEPS I. The effect of hot horizontal-branch (HB) stars on the integrated properties of simple stellar populations (SSPs) is again confirmed by our models based on empirical SEDs. In addition, we have extended our empirical models to the near-IR wavelength and predicted the strengths of the calcium II triplet (CaT) and the Paschen triplet (PaT) based on the INDO-US and the Cenarro library. We find that the effect of HB stars and the age of SSPs on the CaT is almost negligible. On the other hands, the PaT models are very sensitive to the existence of hot stars, e.g., HB stars and young turn-off stars, and show very similar results with Balmer lines. Interestingly, the CaT distribution of GCs in NGC 1407, which is at odds with the optical (B-I) color distribution, can be explained by the unique feature of the CaT-[Fe/H] relations that show almost the same equivalent widths in the metal-rich regime. We will also discuss the impact of the second-generation populations on the strength of the CaT.

• Smart Structures and Systems
• /
• v.22 no.2
• /
• pp.175-183
• /
• 2018

Time-series data often contain one of the most valuable pieces of information in many fields including manufacturing. Because time-series data are relatively cheap to acquire, they (e.g., vibration signals) have become a crucial part of big data even in manufacturing shop floors. Recently, deep-learning models have shown state-of-art performance for analyzing big data because of their sophisticated structures and considerable computational power. Traditional models for a machinery-monitoring system have highly relied on features selected by human experts. In addition, the representational power of such models fails as the data distribution becomes complicated. On the other hand, deep-learning models automatically select highly abstracted features during the optimization process, and their representational power is better than that of traditional neural network models. However, the applicability of deep-learning models to the field of prognostics and health management (PHM) has not been well investigated yet. This study integrates the "residual fitting" mechanism inherently embedded in the wavelet transform into the convolutional neural network deep-learning structure. As a result, the architecture combines a signal smoother and classification procedures into a single model. Validation results from rotor vibration data demonstrate that our model outperforms all other off-the-shelf feature-based models.

• Speech Sciences
• /
• v.10 no.4
• /
• pp.7-21
• /
• 2003

Sums-of-Products models were built for segment duration prediction of spoken Korean. An experiment for the modelling was carried out to apply the results to Korean text-to-speech synthesis systems. 670 read sentences were analyzed. trained and tested for the construction of the duration models. Traditional sequential rule systems were extended to simple additive, multiplicative and additive-multiplicative models based on Sums-of-Products modelling. The parameters used in the modelling include the properties of the target segment and its neighbors and the target segment's position in the prosodic structure. Two optimisation strategies were used: the downhill simplex method and the simulated annealing method. The performance of the models was measured by the correlation coefficient and the root mean squared prediction error (RMSE) between actual and predicted duration in the test data. The best performance was obtained when the data was trained and tested by ' additive-multiplicative models. ' The correlation for the vowel duration prediction was 0.69 and the RMSE. 31.80 ms. while the correlation for the consonant duration prediction was 0.54 and the RMSE. 29.02 ms. The results were not good enough to be applied to the real-time text-to-speech systems. Further investigation of feature interactions is required for the better performance of the Sums-of-Products models.

• Structural Engineering and Mechanics
• /
• v.84 no.5
• /
• pp.651-664
• /
• 2022

A precise prediction of the ultimate bond strength between rebar and surrounding concrete plays a major role in structural design, as it effects the load-carrying capacity and serviceability of a member significantly. In the present study, Gaussian models are employed for modelling bond strength of ribbed steel bars embedded in concrete. Gaussian models offer a non-parametric method based on Bayesian framework which is powerful, versatile, robust and accurate. Five different Gaussian models are explored in this paper-Gaussian Process (GP), Variational Heteroscedastic Gaussian Process (VHGP), Warped Gaussian Process (WGP), Sparse Spectrum Gaussian Process (SSGP), and Twin Gaussian Process (TGP). The effectiveness of the models is also evaluated in comparison to the numerous design formulae provided by the codes. The predictions from the Gaussian models are found to be closer to the experiments than those predicted using the design equations provided in various codes. The sensitivity of the models to various parameters, input feature space and sampling is also presented. It is found that GP, VHGP and SSGP are effective in prediction of the bond strength. For large data set, GP, VHGP, WGP and TGP can be computationally expensive. In such cases, SSGP can be utilized.

• Journal of information and communication convergence engineering
• /
• v.21 no.4
• /
• pp.346-350
• /
• 2023

Trajectory prediction is an essential element for driving autonomous vehicles, and various trajectory prediction models have emerged with the development of deep learning technology. Convolutional neural network (CNN) is the most commonly used neural network architecture for extracting the features of visual images, and the latest models exhibit high performances. This study was conducted to identify an efficient CNN backbone model among the components of deep learning models for trajectory prediction. We changed the existing CNN backbone network of multiple-trajectory prediction models used as feature extractors to various state-of-the-art CNN models. The experiment was conducted using nuScenes, which is a dataset used for the development of autonomous vehicles. The results of each model were compared using frequently used evaluation metrics for trajectory prediction. Analyzing the impact of the backbone can improve the performance of the trajectory prediction task. Investigating the influence of the backbone on multiple deep learning models can be a future challenge.

• Journal of KIISE
• /
• v.44 no.12
• /
• pp.1290-1295
• /
• 2017

Document classification, a task of classifying the category of each document based on text, is one of the fundamental areas for natural language processing. Document classification may be used in various fields such as topic classification and sentiment classification. Neural network models for document classification can be divided into two categories: word-level models and character-level models that treat words and characters as basic units respectively. In this study, we propose a neural network model that combines character-level and word-level models to improve performance of document classification. The proposed model extracts the feature vector of each word by combining information obtained from a word embedding matrix and information encoded by a character-level neural network. Based on feature vectors of words, the model classifies documents with a hierarchical structure wherein recurrent neural networks with attention mechanisms are used for both the word and the sentence levels. Experiments on real life datasets demonstrate effectiveness of our proposed model.

• Journal of the Korea Society of Computer and Information
• /
• v.20 no.4
• /
• pp.41-48
• /
• 2015

Recently, the spread of 3D printers has been increasing the demand for 3D models. However, the creation of 3D models should have a trained specialist using specialized softwares. This paper is about an algorithm to produce a 3D model from a single sheet of two-dimensional front face photograph, so that ordinary people can easily create 3D models. The background and the foreground are separated from a photo and predetermined constant number vertices are placed on the seperated foreground 2D image at a same interval. The arranged vertex location are extended in three dimensions by using the gray level of the pixel on the vertex and the characteristics of eyebrows and nose of the nomal human face. The separating method of the foreground and the background uses the edge information of the silhouette. The AdaBoost algorithm using the Haar-like feature is also employed to find the location of the eyes and nose. The 3D models obtained by using this algorithm are good enough to use for 3D printing even though some manual treatment might be required a little bit. The algorithm will be useful for providing 3D contents in conjunction with the spread of 3D printers.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.10C
• /
• pp.1007-1016
• /
• 2005

This paper presents a feature extraction method for shape based retrieval of 3D models. Since the feature descriptor of 3D model should be invariant to translation, rotation and scaling, it is necessary to preprocess the 3D models to represent them in a canonical coordinate system. We use the PCA(Principal Component Analysis) method to preprocess the 3D models. Also, we apply that to make a MBR(Minimum Boundary Rectangle) and a circumsphere. The proposed algorithm is as follows. We generate a circumsphere around 3D models, where radius equals 1(r=1) and locate each model in the center of the circumsphere. We produce the concentric spheres with a different radius($r_i=i/n,\;i=1,2,{\ldots},n$). After looking for meshes intersected with the concentric spheres, we compute the curvature of the meshes. We use these curvatures as the model descriptor. Experimental results numerically show the performance improvement of proposed algorithm from min. 0.1 to max. 0.6 in comparison with conventional methods by ANMRR, although our method uses .relatively small bins. This paper uses $R{^*}-tree$ as the indexing.

• Journal of Korea Multimedia Society
• /
• v.22 no.5
• /
• pp.535-546
• /
• 2019

Recently, there have been a lot of researches on the whole face replacement system, but it is not easy to obtain stable results due to various attitudes, angles and facial diversity. To produce a natural synthesis result when replacing the face shown in the video image, technologies such as face area detection, feature extraction, face alignment, face area segmentation, 3D attitude adjustment and facial transposition should all operate at a precise level. And each technology must be able to be interdependently combined. The results of our analysis show that the difficulty of implementing the technology and contribution to the system in facial replacement technology has increased in facial feature point extraction and facial alignment technology. On the other hand, the difficulty of the facial transposition technique and the three-dimensional posture adjustment technique were low, but showed the need for development. In this paper, we propose four facial replacement models such as 2-D Faceswap, OpenPose, Deekfake, and Cycle GAN, which are suitable for the Cox platform. These models have the following features; i.e. these models include a suitable model for front face pose image conversion, face pose image with active body movement, and face movement with right and left side by 15 degrees, Generative Adversarial Network.

• Journal of Internet Computing and Services
• /
• v.9 no.6
• /
• pp.27-35
• /
• 2008

With the proliferation of wireless devices, mobile ad-hoc networking (MANETS) has become a very exciting and important technology. However, MANET is more vulnerable than wired networking. Existing security mechanisms designed for wired networks have to be redesigned in this new environment. In this paper, we discuss the problem of anomaly detection in MANET. The focus of our research is on techniques for automatically constructing anomaly detection models that are capable of detecting new or unseen attacks. We propose a new anomaly detection method for MANETs. The proposed method performs cross-feature analysis on the basis of Rough sets to capture the inter-feature correlation patterns in normal traffic. The performance of the proposed method is evaluated through a simulation. The results show that the performance of the proposed method is superior to the performance of Huang method that uses cross-feature based on the probability of feature attribute value. Accordingly, we know that the proposed method effectively detects anomalies.