• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.13 no.3
• /
• pp.186-199
• /
• 2013

We pose pattern classification as a density estimation problem where we consider mixtures of generative models under partially labeled data setups. Unlike traditional approaches that estimate density everywhere in data space, we focus on the density along the decision boundary that can yield more discriminative models with superior classification performance. We extend our earlier work on the recursive estimation method for discriminative mixture models to semi-supervised learning setups where some of the data points lack class labels. Our model exploits the mixture structure in the functional gradient framework: it searches for the base mixture component model in a greedy fashion, maximizing the conditional class likelihoods for the labeled data and at the same time minimizing the uncertainty of class label prediction for unlabeled data points. The objective can be effectively imposed as individual mixture component learning on weighted data, hence our mixture learning typically becomes highly efficient for popular base generative models like Gaussians or hidden Markov models. Moreover, apart from the expectation-maximization algorithm, the proposed recursive estimation has several advantages including the lack of need for a pre-determined mixture order and robustness to the choice of initial parameters. We demonstrate the benefits of the proposed approach on a comprehensive set of evaluations consisting of diverse time-series classification problems in semi-supervised scenarios.

• Proceedings of the KSRS Conference
• /
• 2007.10a
• /
• pp.337-340
• /
• 2007

Many researches conducted to investigate the relationship between surface emissivity and surface temperature in the past two decades and pointed out that the emissivity play a key role in applying remote sensing data to retrieve surface temperature. The task of surface temperature estimation is so important in many research fields, such as earth energy budgets, evapotranspiration, drought, global change and heat island effect. Therefore, it is indispensable to develop an effective and accurate technique to estimate the emissivity for accurate surface temperature estimations. This study developed an improved emissivity estimation technique for the use of surface temperature retrievals with MODIS data. The result of applying this improved technique using Band 31 of MODIS shows that the accuracy of estimated surface temperatures will be improved. This study also uses MODIS data observed in 2005 to establish the relationship between the surface emissivity correction factor and NDVI. Through the use of these correction factors, the land surface temperature can be retrieved more accurate with MODIS data.

• Journal of Korean Society of Transportation
• /
• v.19 no.6
• /
• pp.171-182
• /
• 2001

Travel Time is an important characteristic of traffic conditions in a road network. Currently, there are so many road users to get a unsatisfactory traffic information that is provided by existing collection systems such as, Detector, Probe car, CCTV and Anecdotal Report. This paper presents the results achieved with Data Fusion Model, Hybrid Neuro Fuzzy System for on - line estimation of travel times using RTMS(Remote Traffic Microwave Sensor) and Probe Data in the signalized arterial road. Data Fusion is the most important process to compose the various of data which can present real value for traffic situation and is also the one of the major process part in the TIC(Traffic Information Center) for analyzing and processing data. On-line travel time estimation methods(FALEM) on the basis of detector data has been evaluated by real value under KangNam Test Area.

• Journal of Advanced Research in Ocean Engineering
• /
• v.3 no.1
• /
• pp.32-40
• /
• 2017

Recently, the amount of data to be processed and the complexity thereof have been increasing due to the development of information and communication technology, and industry's interest in such big data is increasing day by day. In the shipbuilding and offshore industry also, there is growing interest in the effective utilization of data, since various and vast amounts of data are being generated in the process of design, production, and operation. In order to effectively utilize big data in the shipbuilding and offshore industry, it is necessary to store and process large amounts of data. In this study, it was considered efficient to apply Hadoop and R, which are mostly used in big data related research. Hadoop is a framework for storing and processing big data. It provides the Hadoop Distributed File System (HDFS) for storing big data, and the MapReduce function for processing. Meanwhile, R provides various data analysis techniques through the language and environment for statistical calculation and graphics. While Hadoop makes it is easy to handle big data, it is difficult to finely process data; and although R has advanced analysis capability, it is difficult to use to process large data. This study proposes a big data platform based on Hadoop for applications in the shipbuilding and offshore industry. The proposed platform includes the existing data of the shipyard, and makes it possible to manage and process the data. To check the applicability of the platform, it is applied to estimate the weights of offshore structure topsides. In this study, we store data of existing FPSOs in Hadoop-based Hortonworks Data Platform (HDP), and perform regression analysis using RHadoop. We evaluate the effectiveness of large data processing by RHadoop by comparing the results of regression analysis and the processing time, with the results of using the conventional weight estimation program.

• Proceedings of the Korea Society of Information Technology Applications Conference
• /
• 2005.11a
• /
• pp.255-258
• /
• 2005

A blind-mode integer frequency offset estimation algorithm is proposed for an OFDM system. Imperfect integer frequency offset estimation causes ambiguity in the data sub-carrier position. Morelli's blind integer frequency offiet estimation algorithm exploits the likelihood function by comparing the power in sub-carriers. It, however, shows performance degradation when there is the fractional frequency offset. The proposed algorithm solves this by using interpolation in the frequency domain. In this algorithm, it is exploited that the effect of the frequency offset is shown as a shift of power spectrum. By calculating the covariance of over-sampled samples, most approximate samples to integer point are obtained. It enables integer frequency offset estimation less affected by fractional frequency offset.

• Journal of Korea Multimedia Society
• /
• v.20 no.11
• /
• pp.1842-1848
• /
• 2017

Recently, user motion estimation methods using various wearable devices have been actively studied. In this paper, we propose a motion estimation system using Myo, which is one of the wearable devices, using two Myo and their dependency relations. The estimated motion is used as a command for remotely controlling the robot. Myo's Orientation and EMG signals are used for motion estimation. These two type data sets are used complementarily to increase the accuracy of motion estimation. We design and implement the system according to the proposed method and analyze the results through experiments. As a result of comparison with previous studies, the accuracy of motion estimation can be improved by about 12.3%.

• Journal of Korea Society of Digital Industry and Information Management
• /
• v.7 no.3
• /
• pp.107-113
• /
• 2011

An OFDMA(Orthogonal Frequency Division Multiple Access) includes a MIMO(Multi-Input Multi-Output) scheme for improving spectral efficiency and data throughput. Recognizing that the performance of MIMO system is heavily dependent upon the accuracy of channel estimation, we propose a novel channel estimation for the MIMO scheme based on OFDMA. Conventional interpolation-based channel estimation suffers from poor estimation error at specific subcarriers. Proposed scheme makes use of a planar interpolation instead of linear interpolation for those subcarriers of bad accuracy. Simulation results show that the proposed scheme improves the performance of MIMO system by improving the accuracy in channel estimation especially for the adverse subcarrier positions. It is observed that the proposed scheme outperforms the conventional method by about 2dB in terms of both mean squared error and overall bit error rate with a reasonable computational complexity.

• Journal of Korean Institute of Industrial Engineers
• /
• v.42 no.4
• /
• pp.256-262
• /
• 2016

The Weibull is widely used in reliability analysis, and several studies have attempted to improve estimation of the distribution's parameters. least squares estimation (LSE) or Maximum likelihood estimation (MLE) are often used to estimate distribution parameters. However, it has been proven that Bayesian methods are more suitable for small sample sizes than LSE and MLE. In this work, the Weibull parameter estimation accuracy of LSE, MLE, and Bayesian method are compared for sample sets with 3 to 30 data points. The Bayesian method was most accurate for sample sizes under 25, and the accuracy of the Bayesian method was similar to LSE and MLE as the sample size increased.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.15 no.4
• /
• pp.435-441
• /
• 2012

This paper provides simulation-based results of effect analysis of sample size and sampling periods on accuracy of reliability estimation methods using multiple comparisons with analysis of variance. Sum of squared errors in estimated reliability measures were evaluated through applying seven estimation methods for one-shot systems to simulated quantal-response data. Analysis of variance was implemented to investigate change in these errors according to variations of sample size and sampling periods for each estimation method, and then the effect analysis on accuracy in reliability estimation was performed using multiple comparisons based on sample size and sampling periods. An efficient way to allocate both sample size and sampling periods for reliability estimation tests of one-shot systems is proposed in this paper from the effect analysis results.

• Journal of Positioning, Navigation, and Timing
• /
• v.11 no.2
• /
• pp.91-98
• /
• 2022

In the next generation wireless communication system, the beamforming technique based on a massive antenna is one of core technologies for transmitting and receiving huge amounts of data, efficiently and accurately. For highly performed and highly reliable beamforming, it is required to accurately estimate the Angle of Arrival (AOA) for the desired signal incident to an antenna. Employing the massive antenna with a large number of elements, although the accuracy of the AOA estimation is enhanced, its computational complexity is dramatically increased so much that real-time communication is difficult. In order to improve this problem, AOA estimation algorithms based on the massive antenna with the low computational complexity have been actively studied. In this paper, we compute and analyze the computational complexity of the cascade AOA estimation algorithm based on the Flexible Massive Concentric Circular Array (FMCCA). In addition, its computational complexity is compared to conventional AOA estimation techniques such as the Multiple Signal Classification (MUSIC) algorithm with the high resolution and the Only Beamspace MUSIC (OBM) algorithm.