• Journal of the Korean Data and Information Science Society
• /
• 제17권4호
• /
• pp.1151-1160
• /
• 2006

The similarity weight, the pearson's correlation coefficient, which is used in the recommender system has a weak point that it cannot predict all of the prediction value. The similarity weight, the vector similarity, has a weak point of the high MAE although the prediction coverage using the vector similarity is higher than that using the pearson's correlation coefficient. The purpose of this study is to suggest how to raise the prediction coverage. Also, the MAE using the suggested method in this study was compared both with the MAE using the pearson's correlation coefficient and with the MAE using the vector similarity, so was the prediction coverage. As a result, it was found that the low of the MAE in the case of using the suggested method was higher than that using the pearson's correlation coefficient. However, it was also shown that it was lower than that using the vector similarity. In terms of the prediction coverage, when the suggested method was compared with two similarity weights as I mentioned above, it was found that its prediction coverage was higher than that pearson's correlation coefficient as well as vector similarity.

• 대한음성학회지:말소리
• /
• 제57호
• /
• pp.139-152
• /
• 2006

In this paper, we present a phrase break prediction method using CRF(Conditional Random Fields), which has good performance at classification problems. The phrase break prediction problem was mapped into a classification problem in our research. We trained the CRF using the various linguistic features which was extracted from POS(Part Of Speech) tag, lexicon, length of word, and location of word in the sentences. Combined linguistic features were used in the experiments, and we could collect some linguistic features which generate good performance in the phrase break prediction. From the results of experiments, we can see that the proposed method shows improved performance on previous methods. Additionally, because the linguistic features are independent of each other in our research, the proposed method has higher flexibility than other methods.

• 기술사
• /
• 제18권2호
• /
• pp.1-5
• /
• 1985

In the flood prediction research, it is pointed out that the difficulty of flood prediction is the frequently experienced overestimation of flood peak. That is caused by the rainfall prediction difficulty and the nonlinearity of hydrological phenomena. Even though the former reason will remain still unsolved, but the latter one can be possibly resolved the method of the AMRA (Auto Regressive Moving Average) model for each runoff component as developed by Dr. Hino and Dr. Hasebe. The principle of the method consists of separating though the numerical filters the total runoff time series into long-term, intermediate and short-term components, or ground water flow, interflow, and surface flow components. As a total system, a hydrological system is a non-linear one. However, once it is separated into two or three subsystems, each subsystem may be treated as a linear system. Also the rainfall components into each subsystem a estimated inversely from the runoff component which is separated from the observed flood. That is why flood prediction can be done without rainfall data. In the prediction of surface flow, the Kalman filter will be applicable but this paper shows only impulse function method.

• 한국컴퓨터정보학회논문지
• /
• 제21권4호
• /
• pp.73-80
• /
• 2016

Most fault prediction models have class imbalance problems because training data usually contains much more non-fault class modules than fault class ones. This imbalanced distribution makes it difficult for the models to learn the minor class module data. Data imbalance is much higher when severity-based fault prediction is used. This is because high severity fault modules is a smaller subset of the fault modules. In this paper, we propose severity-based models to solve these problems using the three sampling methods, Resample, SpreadSubSample and SMOTE. Empirical results show that Resample method has typical over-fit problems, and SpreadSubSample method cannot enhance the prediction performance of the models. Unlike two methods, SMOTE method shows good performance in terms of AUC and FNR values. Especially J48 decision tree model using SMOTE outperforms other prediction models.

• IEIE Transactions on Smart Processing and Computing
• /
• 제5권5호
• /
• pp.323-326
• /
• 2016

High Efficiency Video Coding (HEVC) adopts intra transform skip mode, in which a residual block is directly quantized in the pixel domain without transforming the block into the frequency domain. Intra transform skip mode provides a significant coding gain for screen content. However, when intra-prediction errors are not transformed, the errors are often correlated along the intra-prediction direction. This paper introduces a residual differential pulse code modulation (DPCM) method for the intra-predicted and transform-skipped blocks to remove redundancy. The proposed method performs pixel-by-pixel residual prediction along the intra-prediction direction to reduce the dynamic range of intra-prediction errors. Experimental results show that the transform skip mode's Bjøntegaard delta rate (BD-rate) is improved by 12.8% for vertically intra-predicted blocks. Overall, the proposed method shows an average 1.2% reduction in BD-rate, relative to HEVC, with negligible computational complexity.

• Geomechanics and Engineering
• /
• 제18권4호
• /
• pp.373-389
• /
• 2019

It is important to study the problem of durability for tunnel structures. As a main influence on the durability of tunnel structures, carbonation-induced corrosion is studied. For the complicated environment of tunnel structures, based on the data samples from real engineering examples, the intelligent method (genetic programming) is used to construct the service life prediction model of tunnel structures. Based on the model, the prediction of service life for tunnel structures in carbonation environments is studied. Using the data samples from some tunnel engineering examples in China under carbonation environment, the proposed method is verified. In addition, the performance of the proposed prediction model is compared with that of the artificial neural network method. Finally, the effect of two main controlling parameters, the population size and sample size, on the performance of the prediction model by genetic programming is analyzed in detail.

• Journal of information and communication convergence engineering
• /
• 제21권1호
• /
• pp.82-89
• /
• 2023

Various machine-learning models may yield high predictive power for massive time series for time series prediction. However, these models are prone to instability in terms of computational cost because of the high dimensionality of the feature space and nonoptimized hyperparameter settings. Considering the potential risk that model training with a high-dimensional feature set can be time-consuming, we evaluate a feature-importance-based feature selection method to derive a tradeoff between predictive power and computational cost for time series prediction. We used two machine learning techniques for performance evaluation to generate prediction models from a retail sales dataset. First, we ranked the features using impurity- and Local Interpretable Model-agnostic Explanations (LIME) -based feature importance measures in the prediction models. Then, the recursive feature elimination method was applied to eliminate unimportant features sequentially. Consequently, we obtained a subset of features that could lead to reduced model training time while preserving acceptable model performance.

• 한국지반공학회논문집
• /
• 제24권10호
• /
• pp.83-91
• /
• 2008

현장계측 자료를 이용하여 연약지반의 향후 침하거동을 예측하는 기존의 방법들은 모두 즉시재하 조건을 가정하고 개발된 방법으로써 실제로는 연약지반의 안정성 등을 고려하여 점증재하가 이루어지는 현장에 적용하기에는 많은 제약이 있다. 본 연구에서는 연약층의 두께, 성토하중 크기, 선행압밀하중, 배수거리, 성토속도 등의 다양한 영향인자를 고려하였으며 점증재하가 완료된 이후의 지반개량 기간에도 지속적으로 예측 정확도를 높일 수 있는 계측기반 침하거동 예측기법을 개발하였다. 점증재하 과정에서의 예측방법과 성토완료 이후의 예측방법이 개발되었으며, 성토 완료 이후의 예측방법은 기하학적 보정을 이용한 정확도 향상기법과 확률론적 보정을 이용한 정확도 향상기법 두 가지를 제안하였다. 대형압밀시험 결과를 이용한 예측기법의 적용성 검증 결과, 기존의 예측기법을 적용할 수 없는 점증재하 초기에도 비교적 적은 데이터를 이용하여 상당히 높은 정확도를 가지고 침하거동을 예측할 수 있었다. 또한, 성토완료이후에도 기존 예측기법과 제안된 방법의 비교, 분석 결과 최종침하량과 RMSE에서 모두 제안된 방법이 기존의 예측기법에 비하여 우수한 예측결과를 보였다.

• 전자공학회논문지S
• /
• 제35S권8호
• /
• pp.86-95
• /
• 1998

시간 방향 예측을 사용할 수 없는 Intra-프레임 부호화는 Inter-프레임 부호화에 비해 상당히 높은 비트율을 필요로 한다. 이러한 Intra-프레임의 비트율 감축을 위하여 최근 표준화가 진행중인 MPEG-4에서는 DCT영역에서의 화면내 예측 부호화 방식에 관한 연구를 진행해 왔다. 본 논문에서는 기존 방식들에 비해 약간의 복잡도가 추가된 DCT 영역에서의 화면내 예측 부호화 방식들을 제안한다. 제안 방식은 DC 계수 예측 방식과 2가지의 AC 계수 예측 방식들로 구성된다. DC 계수 예측 방식으로는, 이웃 블록들의 DC 계수간 기 울기를 이용하여 부가 정보없이 선택적으로 예측 방향을 결정하는 방식을 제안한다. AC 계수 예측 방식으로는, 현재 블록의 DC 계수를 고려하여 MPEG-4의 AC 계수 예측 방향을 개선한 방식을 제안하며, 또한, AC 계수 단위로 예측 여부를 결정하는 방식을 제안한다. 실험을 통하여 제안 방식에 의해 상당한 비트율 감축을 얻을 수 있음을 보인다.

• Journal of the Korean Data and Information Science Society
• /
• 제18권4호
• /
• pp.945-953
• /
• 2007

In this paper, we propose a smoothing and thus noise-reducing method of data sequences for stock price prediction with hidden Markov models, HMMs. The suggested method just uses simple moving average. A proper average size is obtained from forecasting experiments with stock prices of bank sector of Korean Exchange. Forecasting method with HMM and moving average smoothing is compared with a conventional method.