Keon-Jun Park

Design of FCM-Based Fuzzy Neural Networks and Its Optimization for Pattern Recognition

In this paper, we introduce a new category of fuzzy neural network with multi-output based on fuzzy c-means clustering algorithm (FCM-based FNNm). The premise part of the rules of the proposed network is realized with the aid of the scatter partition of input space generated by FCM clustering algorithm. The number of the partition of input space is composed of the number of clusters and the individual partitioned spaces describe the fuzzy rules. Due to these characteristics, we may alleviate the problem of the curse of dimensionality. The consequence part of the rule is represented by polynomial functions with multi-output. And the coefficients of the polynomial functions are learned by BP algorithm. To optimize the parameters of FCM-based FNNm we consider real-coded genetic algorithms. The proposed network is evaluated with the use of numerical experimentation.

Design of Optimized Pattern Recognizer by Means of Fuzzy Neural Networks Based on Individual Input Space

In this paper, we introduce the fuzzy neural network based on the individual input space to design the pattern recognizer. The proposed networks configure the network by individually dividing each input space. The premise part of the networks is independently composed of the fuzzy partition of individual input spaces and the consequence part of the networks is represented by polynomial functions. The learning of fuzzy neural networks is realized by adjusting connection weights of the neurons in the consequent part of the fuzzy rules and it follows a back-propagation algorithm. In addition, in order to optimize the parameters of the proposed network, we use real-coded genetic algorithms. Finally, we design the optimized pattern recognizer using the experimental data for pattern recognition.

A Study on the LED VLC Trans-Receiver Module for Use of Ubiquitous-Sensor Network and Its Efficiency

This study aims to implement a VLC (Visible Light Communication) trans-receiver module media-transfer system using LED (Light Emitting Diode) white lighting based on PC modules, while analyzing its efficiency of transfer technology. LED lighting can control different wavelength lights depending on materials and components, and it is possible to communicate by transferring data through the light of LED lighting. To implement an LED VLC, this study intends to use 12-LED light emitting displays for the transmission part and widely-known infrared sensors for the receiver part. For the trans-receiver developed for LED VLC, a visible media transfer system was established with the initial distance value set as over 30cm, and the entire system transmission speed set as 125kbps. To analyze the performance of an implemented module, this study installed LED and infrared sensors on the PC module and showed prediction and communication distances, while attempting to confirm its application methods and possibility. To increase the overall efficiency of an LED module, this study measured its different performance with lens equipped and unequipped respectively and found out its efficiency increased by about 20%.

Design of a LED Emotional Lighting System for Indoor Exercise and Resting Situations using Fuzzy Inference

In this study, an LED emotional lighting algorithm optimized to the situation was implemented using fuzzy inference for users in exercise and resting situations in indoor environment. The fuzzy theory was used instead of the conventional simple color temperature control in order to control the colors and color temperatures of LED light sources in line with user environment under complex conditions. An LED emotional lighting system based on fuzzy theory was designed through a combination of colors according to the color and temperature of emotional language based on the user behavior. As a result, the color and color temperature can give a good effect on user`s emotions for an emotional lighting that is effective for resting and exercise.

LBG-Based Non-fuzzy Inference System for Nonlinear Process

We introduce a non-fuzzy inference system based on Linde-Buzo-Gray (LBG) Algorithm to construct model for nonlinear process. In grid partition, the generation of fuzzy rules has the problem that the number of fuzzy rules exponentially increases. To solve this problem, we generate the subspaces using the scatter partition of input space. The rules of non-fuzzy inference systems are generated by partitioning the input space in the scatter form using LBG algorithm. The consequence part of the rules is represented in the form of polynomial functions. The data widely used in nonlinear process is used to evaluate the performance of the proposed model.

Optimization of Non-fuzzy Neural Networks Based on Crisp Rules in Scatter Partition

We introduce a design of non-fuzzy neural networks that have crisp rules in scatter partition. To generate the crisp rules and construct the networks, we use hard c-means clustering algorithm. The partitioned local spaces indicate the crisp rules of the proposed networks. The consequence part of the rule is represented by polynomial functions. The coefficients of the polynomial functions are learned using back-propagation algorithm. In order to optimize the parameters of the proposed networks we use particle swarm optimization techniques. The proposed networks are evaluated with the example for nonlinear process.

Hard partition-based non-fuzzy inference system for nonlinear process

We introduec a non-fuzzy inference system based on hard partition to contruct model for nonlinear process. In fuzzy modeling, the generation of fuzzy rules has the problem that the number of fuzzy rules exponentially increases. To solve this problem, the rules of non-fuzzy inference systems are generated by partitioning the input space in the scatter form using HCM clustering algorithm. The premise parameters of the rules are determined by membership matrix by means of HCM clustering algorithm. The consequence part of the rules is represented in the form of polynomial functions. The proposed model is evaluated with the performance using the data widely used in nonlinear process.

Design of Interval Type-2 Fuzzy Relation-Based Neuro-Fuzzy Networks for Nonlinear Process

In this paper, we introduce the design of interval type-2 fuzzy relation-based neuro-fuzzy networks (IT2FRNFN) for modeling nonlinear process. IT2FRNFN is the network of combination between the neuro-fuzzy network (NFN) and interval type-2 fuzzy set with uncertainty. The premise part of the network is composed of the fuzzy relation division of input space and the consequence part of the network is represented by polynomial functions with interval set. And we also consider genetic algorithms to determine the structure and estimate the values of the parameters. The proposed network is evaluated with the nonlinear process.

Design of Interval Type-2 FCM-Based FNN and Genetic Optimization for Pattern Recognition

A new category of fuzzy neural networks with multiple outputs based on an interval type-2 fuzzy c-means clustering algorithm (IT2FCM-based FNNm) for pattern recognition is proposed in this paper. The premise part of the rules of the proposed network is realized with the aid of the scatter partition of the input space generated by the IT2FCM clustering algorithm. The number of the partition of input space equals the number of clusters, and the individual partitioned spaces describe the fuzzy rules. The consequence part of the rules is represented by polynomial functions with an interval set along with multiple outputs. The coefficients of the polynomial functions are learned by the back-propagation (BP) algorithm. To optimize the parameters of the IT2FCM-based FNNm, we consider real-coded genetic algorithms. The proposed network is evaluated with the use of numerical experimentation for pattern recognition.

A comparative study of information granulation-based fuzzy inference systems developed by means of space optimization algorithm (SOA) and genetic algorithms

In this study, we introduce a hybrid identification method of fuzzy inference systems based on the SOA and information granulation. The underlying idea of SOA comes from an analysis of the solution space. When dealing with the optimization of the fuzzy models, SOA leads not only to better search performance to find a global minimum but is also more computationally effective in comparison with the standard evolutionary algorithms (e.g. genetic algorithm). Information granulation realized with the aid of K-Means clustering is used to determine the initial values of the apex parameters of membership function of the fuzzy model. This method is used to carry out parameter estimation of the fuzzy models as well as to realize structure determination. The overall hybrid optimization of fuzzy inference systems comes in the form of two optimization mechanisms: a structural optimization and a parametric optimization. The structural optimization is supported by the SOA and K-Means while the parametric optimization is realized via SOA and a standard least squares method. The evaluation of the performance of the proposed model is presented by using a series of examples such as a Three-variable nonlinear function and Boston housing data. A comparative analysis demonstrates that SOA results in the improved performance both in terms of the quality of the model and the required computing time. Experimental results show that the proposed model leads to superior performance in comparison with the performance of some other fuzzy models reported in the literature.