• ETRI Journal
• /
• v.32 no.1
• /
• pp.120-128
• /
• 2010

In this paper, we describe a modified fixed-threshold sequential minimal optimization (FSMO) for 1-slack structural support vector machine (SVM) problems. Because the modified FSMO uses the fact that the formulation of 1-slack structural SVMs has no bias, it breaks down the quadratic programming (QP) problems of 1-slack structural SVMs into a series of smallest QP problems, each involving only one variable. For various test sets, the modified FSMO is as accurate as existing structural SVM implementations (n-slack and 1-slack SVM-struct) but is faster on large data sets.

• International journal of advanced smart convergence
• /
• v.12 no.4
• /
• pp.202-207
• /
• 2023

In this paper, we discuss the design and implementation of a recommendation platform actually built in the field. We survey deep learning-based recommendation models that are effective in reflecting individual user characteristics. The recently proposed RNN-based sequential recommendation models reflect individual user characteristics well. The recommendation platform we proposed has an architecture that can collect, store, and process big data from a company's commercial services. Our recommendation platform provides service providers with intuitive tools to evaluate and apply timely optimized recommendation models. In the model evaluation we performed, RNN-based sequential recommendation models showed high scores.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.24 no.3
• /
• pp.295-301
• /
• 2018

This study compares optimization algorithms for efficient estimations of ship's hydrodynamic coefficients. Two constrained algorithms, the interior point and the sequential quadratic programming, are compared for the estimation. Mathematical optimization is designed to get optimal hydrodynamic coefficients for modelling a ship, and benchmark data are collected from sea trials of a training ship. A calibration for environmental influence and a sensitivity analysis for efficiency are carried out prior to implementing the optimization. The optimization is composed of three steps considering correlation between coefficients and manoeuvre characteristics. Manoeuvre characteristics of simulation results for both sets of optimized coefficients are close to each other, and they are also fit to the benchmark data. However, this similarity interferes with the comparison, and it is supposed that optimization conditions, such as designed variables and constraints, are not sufficient to compare them strictly. An enhanced optimization with additional sea trial measurement data should be carried out in future studies.

• Journal of the Korea Safety Management & Science
• /
• v.9 no.3
• /
• pp.135-142
• /
• 2007

The authors used association rules and patterns in sequential of data mining in order to raise the efficiency of engineering changes. The association rule can reduce the number of engineering changes since it can estimate the parts to be changed. The patterns in sequential can perform engineering changes effectively by estimating the parts to be changed from sequence estimation. According to this result, unnecessary engineering changes are eliminated and the number of engineering changes decrease. This method can be used for improving design quality and productivity in company managing engineering changes and related information.

• Spatial Information Research
• /
• v.2 no.2
• /
• pp.175-188
• /
• 1994

A competitive learning network was proposed as unsupervised training method of remote sensing data, Its performance and computational re¬quirements were compared with conventional clustering techniques such as Se¬quential and K - Means. An airborne remote sensing data set was used to study the performance of these classifiers. The proposed algorithm required a little more computational time than the conventional techniques. However, the perform¬ance of competitive learning network algorithm was found to be slightly more than those of Sequential and K - Means clustering techniques.

• Korean Journal of Hydrosciences
• /
• v.8
• /
• pp.125-136
• /
• 1997

The simulation techniques of hydrologic data series have develped for the purposes of the design of water resources system, the optimization of reservoir operation, and the design of flood control of reservoir, etc. While the stochastic models are usually used in most analysis of water resources fields for the generation of data sequences, the indexed sequential modeling (ISM) method based on generation of a series of overlapping short-term flow sequences directly from the historical record has been used for the data generation in the western USA since the early of 1980s. It was reported that the reliable results by ISM were obtained in practical applications. In this study, we generate annual inflow series at a location of Hong Cheon Dam site by using ISM method and autoregressive, order-1 model (AR(1)), and estimate the drought characteristics for the comparison aim between ISM and AR(1).

• Journal of the Korean Data and Information Science Society
• /
• v.20 no.6
• /
• pp.1009-1014
• /
• 2009

The Internet has changed the daily lives of human being in Korea and elsewhere in the world. It has changed the paradigms of traditional commercial activities and created immense opportunities for new business models. Recently, there has been much attention to the internet shopping mall as a means of commercial transaction. To make internet shopping mall competitive, effective customer satisfaction service should be provided and it is necessary to dynamic analysis method for customers' purchasing pattern. In this paper we apply the sequential analysis to comparison of two kinds of sales through the analysis of customers' purchasing pattern.

• Journal of Astronomy and Space Sciences
• /
• v.28 no.1
• /
• pp.63-70
• /
• 2011

The needs for satellite formation flying are gradually increasing to perform the advanced space missions in remote sensing and observation of the space or Earth. Formation flying in low Earth orbit can perform the scientific missions that cannot be realized with a single spacecraft. One of the various techniques of satellite formation flying is the determination of the precise baselines between the satellites within the formation, which has to be in company with the precision validation. In this paper, the baseline of Gravity Recovery and Climate Experiment (GRACE) A and B was determined with the real global positioning system (GPS) measurements of GRACE satellites. And baseline precision was validated with the batch and sequential processing methods using K/Ka-band ranging system (KBR) biased range measurements. Because the proposed sequential method validate the baseline precision, removing the KBR bias with the epoch difference instead of its estimation, the validating data (KBR biased range) are independent of the data validated (GPS-baseline) and this method can be applied to the real-time precision validation. The result of sequential precision validation was 1.5~3.0 mm which is similar to the batch precision validation.

• Communications for Statistical Applications and Methods
• /
• v.23 no.2
• /
• pp.131-146
• /
• 2016

In research on behavioral studies, significant attention has been paid to the stage-sequential process for longitudinal data. Latent class profile analysis (LCPA) is an useful method to study sequential patterns of the behavioral development by the two-step identification process: identifying a small number of latent classes at each measurement occasion and two or more homogeneous subgroups in which individuals exhibit a similar sequence of latent class membership over time. Maximum likelihood (ML) estimates for LCPA are easily obtained by expectation-maximization (EM) algorithm, and Bayesian inference can be implemented via Markov chain Monte Carlo (MCMC). However, unusual properties in the likelihood of LCPA can cause difficulties in ML and Bayesian inference as well as estimation in small samples. This article describes and addresses erratic problems that involve conventional ML and Bayesian estimates for LCPA with small samples. We argue that these problems can be alleviated with a small amount of prior input. This study evaluates the performance of likelihood and MCMC-based estimates with the proposed prior in drawing inference over repeated sampling. Our simulation shows that estimates from the proposed methods perform better than those from the conventional ML and Bayesian method.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.11 no.9
• /
• pp.827-832
• /
• 2001

In this paper, we propose a sequentially optimization method for fuzzy inference system using fuzzy equalization and linguistic hedge. The fuzzy equalization does not require the usual learning step for generating fuzy rules. However, it is too sensitive for the given input-output data set. So, we adopt a sequential scheme which sequentially optimizes the fuzzy inference system. Here, the parameters of fuzzy membership function obtained from the fuzzy equalization are optimized by the genetic algorithm, and then they are also modified to increase the performance index using the linguistic hedge. Finally, we applied it to rice taste data and got better results than previous ones.