• 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 Korea Society for Simulation Conference
• /
• 2001.10a
• /
• pp.313-317
• /
• 2001

The approach to predict time series without neglecting the fluctuation in a short period is tried by using a digital FIR filter and a neural network. The differential waveform of the Nikkei average closing price is filtered by the FIR band-pass filter of 101 length. It is filtered into the five frequency bands of 0-1Hz, 1-2Hz, 2-3Hz, 3-4Hz and 4-5Hz by setting the sampling frequency 10Hz. The each filtered waveform is learned and forecasted by the neural network. The neural network of the back propagation method is adopted in the learning the waveform. By inputting the data of 20 days in the past, the prediction of 10 days ahead is carried out. After learning the time series of each frequency band by the neural network, the predicted data far each frequency band are obtained. The predicted waveforms of each frequency band are synthesized to obtain a final forecast. The waveform can be forecasted well as a whole.

• Journal of Korea Water Resources Association
• /
• v.51 no.2
• /
• pp.165-174
• /
• 2018

The model using time series data can be considered as a flood forecasting model of a small river due to its efficiency for model development and the advantage of rapid simulation for securing predicted time when reliable data are obtained. Transfer Function Noise (TFN) model has been applied hourly flood forecast in Italy, and UK since 1970s, while it has mainly been used for long-term simulations in daily or monthly basis in Korea. Recently, accumulating hydrological data with good quality have made it possible to simulate hourly flood prediction. The purpose of this study is to assess the TFN model applicability that can reflect exogenous variables by combining dynamic system and error term to reduce prediction error for tributary rivers. TFN model with hourly data had better results than result from Storage Function Model (SFM), according to the flood events. And it is expected to expand to similar sized streams in the future.

• Journal of Korean Society for Quality Management
• /
• v.50 no.3
• /
• pp.459-471
• /
• 2022

Purpose: The purpose of this study was to increase prediction accuracy for an anomaly interval identified using an artificial intelligence-based time series anomaly detection technique by establishing a pre-processing process. Methods: Significant variables were extracted by applying feature selection techniques, and anomalies were derived using the TadGAN time series anomaly detection algorithm. After applying machine learning and deep learning methodologies using normal section data (excluding anomaly sections), the explanatory power of the anomaly sections was demonstrated through performance comparison. Results: The results of the machine learning methodology, the performance was the best when SHAP and TadGAN were applied, and the results in the deep learning, the performance was excellent when Chi-square Test and TadGAN were applied. Comparing each performance with the papers applied with a Conventional methodology using the same data, it can be seen that the performance of the MLR was significantly improved to 15%, Random Forest to 24%, XGBoost to 30%, Lasso Regression to 73%, LSTM to 17% and GRU to 19%. Conclusion: Based on the proposed process, when detecting unsupervised learning anomalies of data that are not actually labeled in various fields such as cyber security, financial sector, behavior pattern field, SNS. It is expected to prove the accuracy and explanation of the anomaly detection section and improve the performance of the model.

• KIPS Transactions on Software and Data Engineering
• /
• v.8 no.3
• /
• pp.123-128
• /
• 2019

This paper proposes the stock price prediction based on the artificial intelligence, where the model with recurrent convolution neural network (RCNN) layers is adopted. In the motivation of this prediction, long short-term memory model (LSTM)-based neural network can make the output of the time series prediction. On the other hand, the convolution neural network provides the data filtering, averaging, and augmentation. By combining the advantages mentioned above, the proposed technique predicts the estimated stock price of next day. In addition, in order to emphasize the recent time series, a custom weighted loss function is adopted. Moreover, stock data related to the stock price index are adopted to consider the market trends. In the experiments, the proposed stock price prediction reduces the test error by 3.19%, which is over other techniques by about 19%.

• Journal of KIISE:Computer Systems and Theory
• /
• v.37 no.6
• /
• pp.354-367
• /
• 2010

There are many methods for analyzing patterns in time series data. Although stock data represents a time series, there are few studies on stock pattern analysis and prediction. Since people believe that stock price changes randomly we cannot predict stock prices using a scientific method. In this paper, we measured the degree of the randomness of stock prices using Kolmogorov complexity, and we showed that there is a strong correlation between the degree and the accuracy of stock price prediction using our semi-global alignment method. We transformed the stock price data to quantized string sequences. Then we measured randomness of stock prices using Kolmogorov complexity of the string sequences. We use KOSPI 690 stock data during 28 years for our experiments and to evaluate our methodology. When a high Kolmogorov complexity, the stock price cannot be predicted, when a low complexity, the stock price can be predicted, but the prediction ratio of stock price changes of interest to investors, is 12% prediction ratio for short-term predictions and a 54% prediction ratio for long-term predictions.

• Journal of KIISE:Databases
• /
• v.34 no.6
• /
• pp.483-492
• /
• 2007

Time series of conventional prediction techniques uses the model which is generated from the training step. This model is applied to new input data without any change. If this model is applied directly to stream data, the rate of prediction accuracy will be decreased. This paper proposes an stream data prediction technique using sliding window and regression. This technique considers the characteristic of time series which may be changed over time. It is composed of two steps. The first step executes a fractional process for applying input data to the regression model. The second step updates the model by using its information as new data. Additionally, the model is maintained by only recent data in a queue. This approach has the following two advantages. It maintains the minimum information of the model by using a matrix, so space complexity is reduced. Moreover, it prevents the increment of error rate by updating the model over time. Accuracy rate of the proposed method is measured by RME(Relative Mean Error) and RMSE(Root Mean Square Error). The results of stream data prediction experiment are performed by the proposed technique IMQR(Incremental Multiple Quadratic Regression) is more efficient than those of MLR(Multiple Linear Regression) and SVR(Support Vector Regression).

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.20 no.6
• /
• pp.1-13
• /
• 2021

In this study, a time series analysis technique was applied to calibrate and predict traffic data for various purposes, such as planning, design, maintenance, and research. Existing algorithms have limitations in application to data such as traffic data because they show strong periodicity and seasonality or irregular data. To overcome and supplement these limitations, we applied the SARIMA model, an analytical technique that combines the autocorrelation model, the Seasonal Auto Regressive(SAR), and the seasonal Moving Average(SMA). According to the analysis, traffic volume prediction using the SARIMA(4,1,3)(4,0,3) 12 model, which is the optimal parameter combination, showed excellent performance of 85% on average. In addition to traffic data, this study is considered to be of great value in that it can contribute significantly to traffic correction and forecast improvement in the event of missing traffic data, and is also applicable to a variety of time series data recently collected.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2012.01a
• /
• pp.263-264
• /
• 2012

본 논문은 여러 개의 독립적인 시계열로 구성된 시계열 패널 자료를 이용하여 비선형 모형인 GRCA모형과 신경망을 이용하여 예측값을 구하여 서로 비교 분석하고자 한다. 먼저 GRCA모형에 대하여 연구하고 신경망의 구조와 예측값을 구하기 위한 여러 가지 변환함수를 유도한다. 단기 예측에서는 신경망 방법의 예측값이 더 좋았고, 장기예측에서는 비선형모형을 이용한 예측값이 더 좋은 것으로 나타났다.

• 한국데이터정보과학회:학술대회논문집
• /
• 2002.06a
• /
• pp.23-28
• /
• 2002

Support vector machine(SVM) has been very successful in classification, regression, time series prediction and density estimation. In this paper, we will propose SVM for fuzzy data classification.