• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.1
• /
• pp.27-34
• /
• 2013

Background subtraction is one of the key techniques for automatic video content analysis, especially in the tasks of visual detection and tracking of moving object. In this paper, we present a new sample-based technique for background extraction that provides background image as well as background model. To handle both high-frequency and low-frequency events at the same time, multiple interval background models are adopted. The main innovation concerns the use of a confidence factor to select the best model from the multiple interval background models. To our knowledge, it is the first time that a confidence factor is used for merging several background models in the field of background extraction. Experimental results revealed that our approach based on multiple interval sampling works well in complicated situations containing various speed moving objects with environmental changes.

• Journal of Broadcast Engineering
• /
• v.25 no.2
• /
• pp.143-150
• /
• 2020

Huge amounts of contents are being uploaded every day on various streaming platforms. Among those videos, game and sports videos account for a great portion. The broadcasting companies sometimes create and provide highlight videos. However, these tasks are time-consuming and costly. In this paper, we propose models that automatically predict highlights in games and sports matches. While most previous approaches use visual information exclusively, our models use both audio and visual information, and present a way to understand short term and long term flows of videos. We also describe models that combine GAN to find better highlight features. The proposed models are evaluated on e-sports and baseball videos.

• Journal of Broadcast Engineering
• /
• v.24 no.4
• /
• pp.553-563
• /
• 2019

As the number of videos uploaded on live streaming platforms rapidly increases, the demand for providing highlight videos is increasing to promote viewer experiences. In this paper, we present novel methods for predicting highlights using chat logs and audio data in videos. The proposed models employ bi-directional LSTMs to understand the contextual flow of a video. We also propose to use the features over various time-intervals to understand the mid-to-long term flows. The proposed Our methods are demonstrated on e-Sports and baseball videos collected from personal broadcasting platforms such as Twitch and Kakao TV. The results show that the information from multiple time-intervals is useful in predicting video highlights.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.60 no.1
• /
• pp.142-149
• /
• 2011

This paper presents a new state of charge(SOC) estimation of large capacity of Li-ion battery (LIB) based on the multiple model adaptive estimation(MMAE) method. We first introduce an equivalent circuit model of LIB. The relationship between the terminal voltage and the open circuit voltage(OCV) is nonlinear and may vary depending on the changes of temperature and C-rate. In this paper, such behaviors are described as a set of multiple linear time invariant impedance models. Each model is identified at a temperature and a C-rate. These model set must be obtained a priori for a given LIB. It is shown that most of impedances can be modeled by first-order and second-order transfer functions. For the real time estimation, we transform the continuous time models into difference equations. Subsequently, we construct the model banks in the manner that each bank consists of four adjacent models. When an operating point of cell temperature and current is given, the corresponding model bank is directly determined so that it is included in the interval generated by four operating points of the model bank. The MMAE of SOC at an arbitrary operating point (T $^{\circ}C$, $I_{bat}$[A]) is performed by calculating a linear combination of voltage drops, which are obtained by four models of the selected model bank. The demonstration of the proposed method is shown through simulations using DUALFOIL.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.10 no.6
• /
• pp.137-143
• /
• 2010

A method for improving prediction accuracy through processing time series data has been studied in this research. We have designed techniques to model multiple similar time series data and avoided the shortcomings of single prediction model. We predicted the future changes by effective rules derived from these models. The methods for testing prediction accuracy consists of three types: fixed interval, sliding, and cumulative method. Among the three, cumulative method produced the highest accuracy.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.9
• /
• pp.351-360
• /
• 2018

The LWR (Locally Weighted Regression) model, which is traditionally a lazy learning model, is designed to obtain the solution of the prediction according to the input variable, the query point, and it is a kind of the regression equation in the short interval obtained as a result of the learning that gives a higher weight value closer to the query point. We study on an incremental ensemble learning approach for LWR, a form of lazy learning and memory-based learning. The proposed incremental ensemble learning method of LWR is to sequentially generate and integrate LWR models over time using a genetic algorithm to obtain a solution of a specific query point. The weaknesses of existing LWR models are that multiple LWR models can be generated based on the indicator function and data sample selection, and the quality of the predictions can also vary depending on this model. However, no research has been conducted to solve the problem of selection or combination of multiple LWR models. In this study, after generating the initial LWR model according to the indicator function and the sample data set, we iterate evolution learning process to obtain the proper indicator function and assess the LWR models applied to the other sample data sets to overcome the data set bias. We adopt Eager learning method to generate and store LWR model gradually when data is generated for all sections. In order to obtain a prediction solution at a specific point in time, an LWR model is generated based on newly generated data within a predetermined interval and then combined with existing LWR models in a section using a genetic algorithm. The proposed method shows better results than the method of selecting multiple LWR models using the simple average method. The results of this study are compared with the predicted results using multiple regression analysis by applying the real data such as the amount of traffic per hour in a specific area and hourly sales of a resting place of the highway, etc.

• International Journal of Internet, Broadcasting and Communication
• /
• v.12 no.1
• /
• pp.67-72
• /
• 2020

The fifth generation (5G) mobile communication has an impact on the human life over the whole world, nowadays, through the artificial intelligence (AI) and the internet of things (IoT). The low latency of the 5G new radio (NR) access is implemented by the state-of-the art technologies, such as non-orthogonal multiple access (NOMA). This paper investigates a practical issue that in NOMA, for the practical channel models, such as fading channel environments, the successive interference cancellation (SIC) should be performed on the stronger channel users with low power allocation. Only if the SIC is performed on the user with the stronger channel gain, NOMA performs better than orthogonal multiple access (OMA). Otherwise, NOMA performs worse than OMA. Such the superiority requirement can be easily implemented for the channel being static or slow varying, compared to the block interval time. However, most mobile channels experience fading. And symbol by symbol channel estimations and in turn each symbol time, selections of the SIC-performing user look infeasible in the practical environments. Then practically the block of symbols uses the single channel estimation, which is obtained by the training sequence at the head of the block. In this case, not all the symbol times the SIC is performed on the stronger channel user. Sometimes, we do perform the SIC on the weaker channel user; such cases, NOMA performs worse than OMA. Thus, we can say that by what percent NOMA is better than OMA. This paper calculates analytically the percentage by which NOMA performs better than OMA in the practical mobile communication systems. We show analytically that the percentage for NOMA being better than OMA is only the function of the ratio of the stronger channel gain variance to weaker. In result, not always, but almost time, NOMA could perform better than OMA.

• Journal of Korean Society of Transportation
• /
• v.18 no.1
• /
• pp.47-59
• /
• 2000

The purpose of this study is to develop travel-time estimation model using neural networks and prediction model using neural networks and kalman-filtering technique. The data used in this study are travel speed collected from inductive loop vehicle detection systems(VDS) and travel time collected from the toll collection system (TCS) between Seoul and Osan toll Plaza on the Seoul-Pusan Expressway. Two models, one for travel-time estimation and the other for travel-time Prediction were developed. Application cases of each model were divided into two cases, so-called, a single-region and a multiple-region. because of the different characteristics of travel behavior shown on each region. For the evaluation of the travel time estimation and Prediction models, two Parameters. i.e. mode and mean were compared using five-minute interval data sets. The test results show that mode was superior to mean in representing the relationship between speed and travel time. It is, however shown that mean value gives better results in case of insufficient data. It should be noted that the estimation and the Prediction of travel times based on the VDS data have been improved by using neural networks, because the waiting time at exit toll gates can be included for the estimation of travel time based on the VDS data by considering differences between VDS and TCS travel time Patterns in the models. In conclusion, the results show that the developed models decrease estimation and prediction errors. As a result of comparing the developed model with the existing model using the observed data, the equality coefficients of the developed model was average 88% and the existing model was average 68%. Thus, the developed model was improved minimum 17% and maximum 23% rather then existing model .

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.6
• /
• pp.1957-1980
• /
• 2021

The significant aim of this research has always been to evaluate the mechanism for efficient and inherently aware usage of vitality in-home devices, thus improving the information of smart metering systems with regard to the usage of selected homes and the time of use. Advances in information processing are commonly used to quantify gigantic building activity data steps to boost the activity efficiency of the building energy systems. Here, some smart data mining models are offered to measure, and predict the time series for energy in order to expose different ephemeral principles for using energy. Such considerations illustrate the use of machines in relation to time, such as day hour, time of day, week, month and year relationships within a family unit, which are key components in gathering and separating the effect of consumers behaviors in the use of energy and their pattern of energy prediction. It is necessary to determine the multiple relations through the usage of different appliances from simultaneous information flows. In comparison, specific relations among interval-based instances where multiple appliances use continue for certain duration are difficult to determine. In order to resolve these difficulties, an unsupervised energy time-series data clustering and a frequent pattern mining study as well as a deep learning technique for estimating energy use were presented. A broad test using true data sets that are rich in smart meter data were conducted. The exact results of the appliance designs that were recognized by the proposed model were filled out by Deep Convolutional Neural Networks (CNN) and Recurrent Neural Networks (LSTM and GRU) at each stage, with consolidated accuracy of 94.79%, 97.99%, 99.61%, for 25%, 50%, and 75%, respectively.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.33 no.4A
• /
• pp.352-360
• /
• 2008

Recently Task Group (TG) 4 of the Institute of Electrical and Electronics Engineers (IEEE) 802.15a has been recommended a system with ranging capability in existence of multiple Simultaneous operating piconets (SOPs) as well as low-cost, low-power. According to the ranging service, coherent and non-coherent based ranging schemes using ternary code have been adopted as a standard. However it is hard to estimate an accurate time of arrival (TOA) in case of using direct sequence based TOA estimation method because pulse repetition interval (PRI) offered by TG is more limited than the maximum excess delay (MED) of channel. To mitigate inter pulse interference (IPI) problem, this paper proposes a non-coherent TOA estimation scheme using non-periodic transmission (NPT) pattern. The proposed receiver is based on a non-coherent energy detection considering with motivation of low rate wireless personal area network (LR-WPAN). TOA information is estimated via proper comparison with a prescribed threshold after the sliding correlation and search back window (SBW) process for reducing TOA error. To verify the performance of proposed ranging scheme, two distinct channel models approved by IEEE 802.15.4a TG are considered. According to the simulation results, we could conclude that the proposed scheme have performed better performance than the conventional method on the existence of multiple SOPs.