• International Journal of Control, Automation, and Systems
• /
• v.2 no.4
• /
• pp.423-434
• /
• 2004

In this paper, we introduce a new topology of Self-Organizing Polynomial Neural Networks (SOPNN) based on genetically optimized Multi-Layer Perceptron (MLP) and discuss its comprehensive design methodology involving mechanisms of genetic optimization. Let us recall that the design of the 'conventional' SOPNN uses the extended Group Method of Data Handling (GMDH) technique to exploit polynomials as well as to consider a fixed number of input nodes at polynomial neurons (or nodes) located in each layer. However, this design process does not guarantee that the conventional SOPNN generated through learning results in optimal network architecture. The design procedure applied in the construction of each layer of the SOPNN deals with its structural optimization involving the selection of preferred nodes (or PNs) with specific local characteristics (such as the number of input variables, the order of the polynomials, and input variables) and addresses specific aspects of parametric optimization. An aggregate performance index with a weighting factor is proposed in order to achieve a sound balance between the approximation and generalization (predictive) abilities of the model. To evaluate the performance of the GA-based SOPNN, the model is experimented using pH neutralization process data as well as sewage treatment process data. A comparative analysis indicates that the proposed SOPNN is the model having higher accuracy as well as more superb predictive capability than other intelligent models presented previously.reviously.

• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.53 no.8
• /
• pp.551-560
• /
• 2004

In this paper, we propose a new architecture of Self-Organizing Fuzzy Polynomial Neural Networks (SOFPNN) that is based on a genetically optimized multilayer perceptron with fuzzy polynomial neurons (FPNs) and discuss its comprehensive design methodology involving mechanisms of genetic optimization, especially genetic algorithms (GAs). The proposed SOFPNN gives rise to a structurally optimized structure and comes with a substantial level of flexibility in comparison to the one we encounter in conventional SOFPNNs. The design procedure applied in the construction of each layer of a SOFPNN deals with its structural optimization involving the selection of preferred nodes (or FPNs) with specific local characteristics (such as the number of input variables, the order of the polynomial of the consequent part of fuzzy rules, and a collection of the specific subset of input variables) and addresses specific aspects of parametric optimization. Through the consecutive process of such structural and parametric optimization, an optimized and flexible fuzzy neural network is generated in a dynamic fashion. To evaluate the performance of the genetically optimized SOFPNN, the model is experimented with using two time series data(gas furnace and chaotic time series), A comparative analysis reveals that the proposed SOFPNN exhibits higher accuracy and superb predictive capability in comparison to some previous models available in the literatures.

• International journal of advanced smart convergence
• /
• v.13 no.2
• /
• pp.61-68
• /
• 2024

In the quest for advancing diabetes diagnosis, this study introduces a novel two-step machine learning approach that synergizes the probabilistic predictions of Logistic Regression with the classification prowess of Random Forest. Diabetes, a pervasive chronic disease impacting millions globally, necessitates precise and early detection to mitigate long-term complications. Traditional diagnostic methods, while effective, often entail invasive testing and may not fully leverage the patterns hidden in patient data. Addressing this gap, our research harnesses the predictive capability of Logistic Regression to estimate the likelihood of diabetes presence, followed by employing Random Forest to classify individuals into diabetic, pre-diabetic or nondiabetic categories based on the computed probabilities. This methodology not only capitalizes on the strengths of both algorithms-Logistic Regression's proficiency in estimating nuanced probabilities and Random Forest's robustness in classification-but also introduces a refined mechanism to enhance diagnostic accuracy. Through the application of this model to a comprehensive diabetes dataset, we demonstrate a marked improvement in diagnostic precision, as evidenced by superior performance metrics when compared to other machine learning approaches. Our findings underscore the potential of integrating diverse machine learning models to improve clinical decision-making processes, offering a promising avenue for the early and accurate diagnosis of diabetes and potentially other complex diseases.

• Journal of Institute of Control, Robotics and Systems
• /
• v.11 no.2
• /
• pp.127-136
• /
• 2005

In this paper, a new architecture and comprehensive design methodology of genetically optimized Rule-based Fuzzy Polynomial Neural Networks(gRFPNN) are introduced and a series of numeric experiments are carried out. The architecture of the resulting gRFPNN results from asynergistic usage of the hybrid system generated by combining rule-based Fuzzy Neural Networks(FNN) with polynomial neural networks (PNN). FNN contributes to the formation of the premise part of the overall rule-based structure of the gRFPNN. The consequence part of the gRFPNN is designed using PNNs. At the premise part of the gRFPNN, FNN exploits fuzzy set based approach designed by using space partitioning in terms of individual variables and comes in two fuzzy inference forms: simplified and linear. As the consequence part of the gRFPNN, the development of the genetically optimized PNN dwells on two general optimization mechanism: the structural optimization is realized via GAs whereas in case of the parametric optimization we proceed with a standard least square method-based learning. To evaluate the performance of the gRFPNN, the models are experimented with the use of several representative numerical examples. A comparative analysis shows that the proposed gRFPNN are models with higher accuracy as well as more superb predictive capability than other intelligent models presented previously.

• Journal of Korean Society for Atmospheric Environment
• /
• v.15 no.6
• /
• pp.779-789
• /
• 1999

To support daily ground-level $O_3$ forecasting in Seoul, a transfer function model(TFM) has been developed by using surface meteorological data and pollutant data(previous-day [$O_3$] and [$NO_2$]) from 1 May to 31 August in 1997. The forecast performance of the TFM was evaluated by statistical comparison with $O_3$ concentration observed during September it is shown that correlation coefficient(R), root mean squared error(RMSE), normalized mean squared error(NMSE) and mean relative error(MRE) were 0.73, 15.64, 0.006 and 0.101, respectively. The TFM appeared to have some difficulty forecasting very high $O_3$ concentrations. To compare with this model, multiple regression model(MRM) was developed for the same period. According to statistical comparison between the TFM and MRM. two models had similar predictive capability but TFM based on $O_3$ concentration higher than 60 ppb provided more accurate forecast than MRM. It was concluded that statistical model based on TFM can be useful for improving the accuracy of local $O_3$ forecast.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.2 no.1
• /
• pp.43-48
• /
• 2002

This paper discusses the application of fuzzy-ARTMAP neural network to compensate the nonlinearity of satellite communication channel. The fuzzy-ARTMAP is the class of ART(adaptive resonance theory) architectures designed fur supervised loaming. It has capabilities not fecund in other neural network approaches, that includes a small number of parameters, no requirements fur the choice of initial weights, automatic increase of hidden units, and capability of adding new data without retraining previously trained data. By a match tracking process with vigilance parameter, fuzzy-ARTMAP neural network achieves a minimax teaming rule that minimizes predictive error and maximizes generalization. Thus, the system automatically leans a minimal number of recognition categories, or hidden units, to meet accuracy criteria. As a input-converting process for implementing fuzzy-ARTMAP equalizer, the sigmoid function is chosen to convert actual channel output to the proper input values of fuzzy-ARTMAP. Simulation studies are performed over satellite nonlinear channels. QPSK signals with Gaussian noise are generated at random from Volterra model. The performance of proposed fuzzy-ARTMAP equalizer is compared with MLP equalizer.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.60 no.1
• /
• pp.184-192
• /
• 2011

In this study, we introduce a identification methodology for FCM-based fuzzy model. The two underlying design mechanisms of such networks involve Fuzzy C-Means (FCM) clustering method and Particle Swarm Optimization(PSO). The proposed algorithm is based on FCM clustering method for efficient processing of data and the optimization of model was carried out using PSO. The premise part of fuzzy rules does not construct as any fixed membership functions such as triangular, gaussian, ellipsoidal because we build up the premise part of fuzzy rules using FCM. As a result, the proposed model can lead to the compact architecture of network. In this study, as the consequence part of fuzzy rules, we are able to use four types of polynomials such as simplified, linear, quadratic, modified quadratic. In addition, a Weighted Least Square Estimation to estimate the coefficients of polynomials, which are the consequent parts of fuzzy model, can decouple each fuzzy rule from the other fuzzy rules. Therefore, a local learning capability and an interpretability of the proposed fuzzy model are improved. Also, the parameters of the proposed fuzzy model such as a fuzzification coefficient of FCM clustering, the number of clusters of FCM clustering, and the polynomial type of the consequent part of fuzzy rules are adjusted using PSO. The proposed model is illustrated with the use of Automobile Miles per Gallon(MPG) and Boston housing called Machine Learning dataset. A comparative analysis reveals that the proposed FCM-based fuzzy model exhibits higher accuracy and superb predictive capability in comparison to some previous models available in the literature.

• KIISE Transactions on Computing Practices
• /
• v.22 no.3
• /
• pp.127-132
• /
• 2016

Several methods have been proposed to apply artificial neural networks to text compression in the past. However, the networks and targets are both limited to the small size due to hardware capability in the past. Modern GPUs have much better calculation capability than CPUs in an order of magnitude now, even though CPUs have become faster. It becomes possible now to train greater and complex neural networks in a shorter time. This paper proposed a method to transform the distribution of original data with a probabilistic neural predictor. Experiments were performed on a feedforward neural network and a recurrent neural network with gated-recurrent units. The recurrent neural network model outperformed feedforward network in compression rate and prediction accuracy.

• Journal of Distribution Science
• /
• v.17 no.11
• /
• pp.73-79
• /
• 2019

Purpose: The purpose of this study was evaluated to compare the predictive power of distress prediction models by using discriminant analysis method and logit analysis method for food service franchise industry in Korea. Research design, data and methodology: Forty-six food service franchise industry with high sales volume in the 2017 were selected as the sample food service franchise industry for analysis. The fourteen financial ratios for analysis were calculated from the data in the 2017 statement of financial position and income statement of forty-six food service franchise industry in Korea. The fourteen financial ratios were used as sample data and analyzed by t-test. As a result seven statistically significant independent variables were chosen. The analysis method of the distress prediction model was performed by logit analysis and multiple discriminant analysis. Results: The difference between the average value of fourteen financial ratios of forty-six food service franchise industry was tested through t-test in order to extract variables that are classified as top-leveled and failure food service franchise industry among the financial ratios. As a result of the univariate test appears that the variables which differentiate the top-leveled food service franchise industry to failure food service industry are income to stockholders' equity, operating income to sales, current ratio, net income to assets, cash flows from operating activities, growth rate of operating income, and total assets turnover. The statistical significances of the seven financial ratio independent variables were also confirmed by logit analysis and discriminant analysis. Conclusions: The analysis results of the prediction accuracy of each distress prediction model in this study showed that the forecast accuracy of the prediction model by the discriminant analysis method was 84.8% and 89.1% by the logit analysis method, indicating that the logit analysis method has higher distress predictability than the discriminant analysis method. Comparing the previous distress prediction capability, which ranges from 75% to 85% by discriminant analysis and logit analysis, this study's prediction capacity, which is 84.8% in the discriminant analysis, and 89.1% in logit analysis, is found to belong to the range of previous study's prediction capacity range and is considered high number.

• Journal of Environmental Science International
• /
• v.18 no.2
• /
• pp.129-139
• /
• 2009

Development of an artificial neural network model was presented to predict the daily maximum $SO_2$ concentration in the urban-industrial area of Ulsan. The network model was trained during April through September for 2000-2005 using $SO_2$ potential parameters estimated from meteorological and air quality data which are closely related to daily maximum $SO_2$ concentrations. Meteorological data were obtained from regional modeling results, upper air soundings and surface field measurements and were then used to create the $SO_2$ potential parameters such as synoptic conditions, mixing heights, atmospheric stabilities, and surface conditions. In particular, two-stage clustering techniques were used to identify potential index representing major synoptic conditions associated with high $SO_2$ concentration. Two neural network models were developed and tested in different conditions for prediction: the first model was set up to predict daily maximum $SO_2$ at 5 PM on the previous day, and the second was 10 AM for a given forecast day using an additional potential factors related with urban emissions in the early morning. The results showed that the developed models can predict the daily maximum $SO_2$ concentrations with good simulation accuracy of 87% and 96% for the first and second model. respectively, but the limitation of predictive capability was found at a higher or lower concentrations. The increased accuracy for the second model demonstrates that improvements can be made by utilizing more recent air quality data for initialization of the model.