• Journal of the Korean Institute of Intelligent Systems
• /
• v.7 no.3
• /
• pp.34-42
• /
• 1997

This paper presents new fuzzy learning algorithms and their applications to time series prediction. During generating fuzzy rules from numerical data, there is a tendency to produce conflicting rules which have same premise but different consequence. To resolve the problem, we propose MCM(Modified Center Method) which is proven to reduce the error in the prediction. We have applied MCM to the analysis of Mackey-Glass time series and Gas Furnace da.ta to verify its efficiency.

• Journal of Information Processing Systems
• /
• v.18 no.1
• /
• pp.115-129
• /
• 2022

Adopting Internet of Things (IoT)-based technologies in smart homes helps users analyze home appliances electricity consumption for better overall cost monitoring. The IoT application like smart home system (SHS) could suffer from large missing values gaps due to several factors such as security attacks, sensor faults, or connection errors. In this paper, a novel framework has been proposed to predict large gaps of missing values from the SHS home appliances electricity consumption time-series datasets. The framework follows a series of steps to detect, predict and reconstruct the input time-series datasets of missing values. A hybrid convolutional neural network-long short term memory (CNN-LSTM) neural network used to forecast large missing values gaps. A comparative experiment has been conducted to evaluate the performance of hybrid CNN-LSTM with its single variant CNN and LSTM in forecasting missing values. The experimental results indicate a performance superiority of the CNN-LSTM model over the single CNN and LSTM neural networks.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.59 no.6
• /
• pp.1159-1166
• /
• 2010

To improve prediction quality of a nonlinear prediction system, the system's capability for uncertainty of nonlinear data should be satisfactory. This paper presents a TSK fuzzy prediction system that can consider and deal with the uncertainty of nonlinear data sufficiently. In the design procedures of the proposed system, HCBKA(Hierarchical Correlationship-Based K-means clustering Algorithm) was used to generate the accurate fuzzy rule base that can control output according to input efficiently, and the first-order difference method was applied to reflect various characteristics of the nonlinear data. Also, multiple prediction systems were designed to analyze the prediction tendencies of each difference data generated by the difference method. In addition, to enhance the prediction quality of the proposed system, an error compensation method was proposed and it compensated the prediction error of the systems suitably. Finally, the prediction performance of the proposed system was verified by simulating two typical time series examples.

• 제어로봇시스템학회:학술대회논문집
• /
• 2001.10a
• /
• pp.146.4-146
• /
• 2001

Various methods have been proposed for the time series prediction. Most of the conventional methods only optimize parameters of mathematical models, but to construct an appropriate functional form of the model is more difficult in the first place. We employ the Genetic Programming (GP) to construct the functional form of prediction models. Our method is distinguished because the model parameters are optimized by using Back-Propagation (BP)-like method and the prediction model includes discontinuous functions, such as if and max, as node functions for describing complicated phenomena. The above-mentioned functions are non-differentiable, but the BP method requires derivative. To solve this problem, we develop ...

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 1996.04a
• /
• pp.67-70
• /
• 1996

The values measured at different time and enumerated sequentially by homogenous interval is called time series. Its goal is to predict values in future by analysing the measured values in past. The stastical approach to time series prediction tend to be by a neural approach with difficulties in expressing the reationship among past data. In neural approach, the preblem is the acquisting of the enough training data in advance. The goal of this paper is that such problem is solved by generating another term as virtual term between terms in time series.

• Journal of The Korean Society of Agricultural Engineers
• /
• v.65 no.2
• /
• pp.1-11
• /
• 2023

This study analyzed the impact of greenhouse cultivation area and groundwater level changes due to the water curtain cultivation in the greenhouse complexes. The groundwater observation data in the Miryang study area were used and classified into greenhouse and field cultivation areas to compare the groundwater impact of water curtain cultivation in the greenhouse complex. We identified the characteristics of the groundwater time series data by the terrain of the study area and selected the optimal model through time series analysis. We analyzed the time series data for each terrain's two representative groundwater observation wells. The Seasonal ARIMA model was chosen as the optimal model for riverside well, and for plain and mountain well, the ARIMA model and Seasonal ARIMA model were selected as the optimal model. A suitable prediction model is not limited to one model due to a change in a groundwater level fluctuation pattern caused by a surrounding environment change but may change over time. Therefore, it is necessary to periodically check and revise the optimal model rather than continuously applying one selected ARIMA model. Groundwater forecasting results through time series analysis can be used for sustainable groundwater resource management.

• The Korean Journal of Applied Statistics
• /
• v.32 no.5
• /
• pp.671-681
• /
• 2019

Studies on predicting arrhythmia using machine learning have been actively conducted with increasing number of arrhythmia patients. Existing studies have predicted arrhythmia based on multivariate data of feature variables extracted from RR interval data at a specific time point. In this study, we consider that the pattern of the heart state changes with time can be important information for the arrhythmia prediction. Therefore, we investigate the usefulness of predicting the arrhythmia with multivariate time series data obtained by extracting and accumulating the multivariate vectors of the feature variables at various time points. When considering 1-nearest neighbor classification method and its ensemble for comparison, it is confirmed that the multivariate time series data based method can have better classification performance than the multivariate data based method if we select an appropriate time series distance function.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.10 no.2
• /
• pp.113-121
• /
• 2000

This paper presents a parallel-structure fuzzy system(PSFS) for prediction of time series data. The PSFS consists of a multiple number of fuzzy systems connected in parallel. Each component fuzzy system in the PSFS predicts the same future data independently based on its past time series data with different embedding dimension and time delay. The component fuzzy systems are characterized by multiple-input singleoutput( MIS0) Sugeno-type fuzzy rules modeled by clustering input-output product space data. The optimal embedding dimension for each component fuzzy system is chosen to have superior prediction performance for a given value of time delay. The PSFS determines the final prediction result by averaging the outputs of all the component fuzzy systems excluding the predicted data with the minimum and the maximum values in order to reduce error accumulation effect.

• Journal of Korean Institute of Industrial Engineers
• /
• v.28 no.2
• /
• pp.171-177
• /
• 2002

In this paper, we propose a new data fusion method to improve the performance of individual prediction models for time series data. Individual models used are ARIMA and neural network and their results are combined based on the weight reflecting the inverse of EWMA of squared prediction error of each individual model. Monte Carlo simulation is used to identify the situation where the proposed approach can take a vintage point over typical fusion methods which utilize MSE for weight. Study results indicate the following: EWMA performs better than MSE fusion when the data size is large with a relatively big amplitude, which is often observed in intra-cranial pressure data. Additionally, EWMA turns out to be a best choice among MSE fusion and the two individual prediction models when the data size is large with relatively small random noises, often appearing in tax revenue data.

• Journal of Advanced Marine Engineering and Technology
• /
• v.21 no.1
• /
• pp.66-70
• /
• 1997

In this paper a new learning algorithm for curvature smoothing and improved generalization for multi-layer neural networks is proposed. To enhance the generalization ability a constraint term of hidden neuron activations is added to the conventional output error, which gives the curvature smoothing characteristics to multi-layer neural networks. When the total cost consisted of the output error and hidden error is minimized by gradient-descent methods, the additional descent term gives not only the Hebbian learning but also the synaptic weight decay. Therefore it incorporates error back-propagation, Hebbian, and weight decay, and additional computational requirements to the standard error back-propagation is negligible. From the computer simulation of the time series prediction with Santafe competition data it is shown that the proposed learning algorithm gives much better generalization performance.