• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2005.04a
• /
• pp.301-304
• /
• 2005

In this paper, we propose a new fuzzy set-based polynomial neuron (FSPN) involving the information granule, and new fuzzy-neural networks - Fuzzy Set based Polynomial Neural Networks (FSPNN). We have developed a design methodology (genetic optimization using Genetic Algorithms) to find the optimal structure for fuzzy-neural networks that expanded from Group Method of Data Handling (GMDH). It is the number of input variables, the order of the polynomial, the number of membership functions, and a collection of the specific subset of input variables that are the parameters of FSPNN fixed by aid of genetic optimization that has search capability to find the optimal solution on the solution space. We have been interested in the architecture of fuzzy rules that mimic the real world, namely sub-model (node) composing the fuzzy-neural networks. We adopt fuzzy set-based fuzzy rules as substitute for fuzzy relation-based fuzzy rules and apply the concept of Information Granulation to the proposed fuzzy set-based rules.

• Journal of the Institute of Convergence Signal Processing
• /
• v.7 no.2
• /
• pp.53-59
• /
• 2006

Dynamic neural networks have been applied to diverse fields requiring temporal signal processing such as system identification and signal prediction. This paper proposes the gamma neural network(GAM), which uses gamma memory kernel in the hidden layer of feedforward multilayered network, to improve dynamics of networks and then describes nonlinear adaptive prediction using the proposed network as an adaptive filter. The proposed network is evaluated in nonlinear signal prediction and compared with feedforword(FNN) and recurrent neural networks(RNN) for the relative comparison of prediction performance. Simulation results show that the GAM network performs better with respect to the convergence speed and prediction accuracy, indicating that it can be a more effective prediction model than conventional multilayered networks in nonlinear prediction for nonstationary signals.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.63 no.4
• /
• pp.534-540
• /
• 2014

In this paper, we propose a fuzzy combined Polynomial Neural Network(PNN) for pattern classification. The fuzzy combined PNN comes from the generic TSK fuzzy model with several linear polynomial as the consequent part and is the expanded version of the fuzzy model. The proposed pattern classifier has the polynomial neural networks as the consequent part, instead of the general linear polynomial. PNNs are implemented by stacking the simple polynomials dynamically. To implement one layer of PNNs, the various types of simple polynomials are used so that PNNs have flexibility and versatility. Although the structural complexity of the implemented PNNs is high, the PNNs become a high order-multi input polynomial finally. To estimate the coefficients of a polynomial neuron, The weighted linear discriminant analysis. The output of fuzzy rule system with PNNs as the consequent part is the linear combination of the output of several PNNs. To evaluate the classification ability of the proposed pattern classifier, we make some experiments with several machine learning data sets.

• Communications for Statistical Applications and Methods
• /
• v.27 no.3
• /
• pp.313-326
• /
• 2020

Deep Learning is the most important key to the development of Artificial Intelligence (AI). There are several distinguishable architectures of neural networks such as MLP, CNN, and RNN. Among them, we try to understand one of the main architectures called Recurrent Neural Network (RNN) that differs from other networks in handling sequential data, including time series and texts. As one of the main tasks recently in Natural Language Processing (NLP), we consider Neural Machine Translation (NMT) using RNNs. We also summarize fundamental structures of the recurrent networks, and some topics of representing natural words to reasonable numeric vectors. We organize topics to understand estimation procedures from representing input source sequences to predict target translated sequences. In addition, we apply multiple translation models with Gated Recurrent Unites (GRUs) in Keras on English-Korean sentences that contain about 26,000 pairwise sequences in total from two different corpora, colloquialism and news. We verified some crucial factors that influence the quality of training. We found that loss decreases with more recurrent dimensions and using bidirectional RNN in the encoder when dealing with short sequences. We also computed BLEU scores which are the main measures of the translation performance, and compared them with the score from Google Translate using the same test sentences. We sum up some difficulties when training a proper translation model as well as dealing with Korean language. The use of Keras in Python for overall tasks from processing raw texts to evaluating the translation model also allows us to include some useful functions and vocabulary libraries as well.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2005.05b
• /
• pp.766-770
• /
• 2005

Cascade-Correlation Neural Networks Model(CCNNM) is used to estimate daily evaporation using limited climatical variables such as atmospheric temperature, dewpoint temperature, relative humidity, wind speed, sunshine duration and radiation. DeBruln equation is applied to estimate daily free-surface evaporation. It is converted into pan evaporation using pan coefficient. The results of CCNNM shows better than those of Debruin equation. This research represents that the strong nonlinear relationship such as evaporation modeling can be generalized by the CCNNM ; a special type of Backpropagation algorithm Neural Networks Model.

• Journal of Institute of Control, Robotics and Systems
• /
• v.6 no.3
• /
• pp.273-283
• /
• 2000

In this paper we suggest an optimal design method of Fuzzy-Neural Networks(FNN) model for complex and nonlinear systems. The FNNs use the simplified inference as fuzzy inference method and Error Back Propagation Algorithm as learning rule. And we use a HCM(Hard C-Means) Clustering Algorithm to find initial parameters of the membership function. The parameters such as parameters of membership functions learning rates and momentum weighted value is proposed to achieve a sound balance between approximation and generalization abilities of the model. According to selection and adjustment of a weighting factor of an aggregate objective function which depends on the number of data and a certain degree of nonlinearity (distribution of I/O data we show that it is available and effective to design and optimal FNN model structure with a mutual balance and dependency between approximation and generalization abilities. This methodology sheds light on the role and impact of different parameters of the model on its performance (especially the mapping and predicting capabilities of the rule based computing). To evaluate the performance of the proposed model we use the time series data for gas furnace the data of sewage treatment process and traffic route choice process.

• Structural Engineering and Mechanics
• /
• v.60 no.4
• /
• pp.655-665
• /
• 2016

This paper aims to develop models to accurately predict the behavior of fresh concrete exposed to vibration using artificial neural networks (ANNs) model and regression model (RM). For this purpose, behavior of a full scale precast concrete mold was investigated experimentally and numerically. Experiment was performed under vibration with the use of a computer-based data acquisition system. Transducers were used to measure time-dependent lateral displacements at some points on mold while both mold is empty and full of fresh concrete. Modeling of empty and full mold was made using both ANNs and RM. For the modeling of ANNs: Experimental data were divided randomly into two parts. One of them was used for training of the ANNs and the remaining part was used for testing the ANNs. For the modeling of RM: Sinusoidal regression model equation was determined and the predicted data was compared with measured data. Finally, both models were compared with each other. The comparisons of both models show that the measured and testing results are compatible. Regression analysis is a traditional method that can be used for modeling with simple methods. However, this study also showed that ANN modeling can be used as an alternative method for behavior of fresh concrete exposed to vibration in precast concrete structures.

• Bulletin of the Korean Chemical Society
• /
• v.26 no.1
• /
• pp.139-145
• /
• 2005

Artificial neural networks (ANNs), for a first time, were successfully developed for the modeling and prediction of solvent effects on rate constant of [2+2] cycloaddition reaction of diethyl azodicarboxylate with ethyl vinyl ether in various solvents with diverse chemical structures using quantitative structure-activity relationship. The most positive charge of hydrogen atom (q$^+$), dipole moment ($\mu$), the Hildebrand solubility parameter (${\delta}_H^2$) and total charges in molecule (q$_t$) are inputs and output of ANN is log k$_2$ . For evaluation of the predictive power of the generated ANN, the optimized network with 68 various solvents as training set was used to predict log k$_2$ of the reaction in 16 solvents in the prediction set. The results obtained using ANN was compared with the experimental values as well as with those obtained using multi-parameter linear regression (MLR) model and showed superiority of the ANN model over the regression model. Mean square error (MSE) of 0.0806 for the prediction set by MLR model should be compared with the value of 0.0275 for ANN model. These improvements are due to the fact that the reaction rate constant shows non-linear correlations with the descriptors.

• KIPS Transactions on Computer and Communication Systems
• /
• v.10 no.3
• /
• pp.71-80
• /
• 2021

The selection of an appropriate neural network algorithm is an important step for accurate data prediction in machine learning. Many algorithms based on basic artificial neural networks have been devised to efficiently predict future data. These networks include deep neural networks (DNNs), recurrent neural networks (RNNs), long short-term memory (LSTM) networks, and gated recurrent unit (GRU) neural networks. Developers face difficulties when choosing among these networks because sufficient information on their performance is unavailable. To alleviate this difficulty, we evaluated the performance of each algorithm by comparing their errors and processing times. Each neural network model was trained using a tax dataset, and the trained model was used for data prediction to compare accuracies among the various algorithms. Furthermore, the effects of activation functions and various optimizers on the performance of the models were analyzed The experimental results show that the GRU and LSTM algorithms yields the lowest prediction error with an average RMSE of 0.12 and an average R2 score of 0.78 and 0.75 respectively, and the basic DNN model achieves the lowest processing time but highest average RMSE of 0.163. Furthermore, the Adam optimizer yields the best performance (with DNN, GRU, and LSTM) in terms of error and the worst performance in terms of processing time. The findings of this study are thus expected to be useful for scientists and developers.

• Journal of Institute of Control, Robotics and Systems
• /
• v.9 no.4
• /
• pp.329-339
• /
• 2003

A neural network based adaptive controller design method is proposed for reconfigurable flight control systems in the presence of variations in aerodynamic coefficients or control effectiveness decrease caused by control surface damage. The neural network based adaptive nonlinear controller is developed by making use of the backstepping technique for command following of the angle of attack, sideslip angle, and bank angle. On-line teaming neural networks are implemented to guarantee reconfigurability and robustness to the uncertainties caused by aerodynamic coefficients variations. The main feature of the proposed controller is that the adaptive controller is designed with assumption that not any of the nonlinear functions of the system is known accurately, whereas most of the previous works assume that only some of the nonlinear functions are unknown. Neural networks loam through the weight update rules that are derived from the Lyapunov control theory. The closed-loop stability of the error states is also investigated according to the Lyapunov theory. A nonlinear dynamic model of an F-16 aircraft is used to demonstrate the effectiveness of the proposed control law.