• ETRI Journal
• /
• v.15 no.2
• /
• pp.35-51
• /
• 1993

This paper presents function approximation based on nonparametric estimation. As an estimation model of function approximation, a three layered network composed of input, hidden and output layers is considered. The input and output layers have linear activation units while the hidden layer has nonlinear activation units or kernel functions which have the characteristics of bounds and locality. Using this type of network, a many-to-one function is synthesized over the domain of the input space by a number of kernel functions. In this network, we have to estimate the necessary number of kernel functions as well as the parameters associated with kernel functions. For this purpose, a new method of parameter estimation in which linear learning rule is applied between hidden and output layers while nonlinear (piecewise-linear) learning rule is applied between input and hidden layers, is considered. The linear learning rule updates the output weights between hidden and output layers based on the Linear Minimization of Mean Square Error (LMMSE) sense in the space of kernel functions while the nonlinear learning rule updates the parameters of kernel functions based on the gradient of the actual output of network with respect to the parameters (especially, the shape) of kernel functions. This approach of parameter adaptation provides near optimal values of the parameters associated with kernel functions in the sense of minimizing mean square error. As a result, the suggested nonparametric estimation provides an efficient way of function approximation from the view point of the number of kernel functions as well as learning speed.

• Journal of Information Technology Services
• /
• v.17 no.3
• /
• pp.171-189
• /
• 2018

To remove software defects and improve performance of software, many developers perform code inspections and use static analysis tools. A code inspection is an activity that is performed manually to detect software defects in the developed source. However, there is no clear criterion which source codes are inspected. A static analysis tool can automatically detect software defects by analyzing the source codes without running the source codes. However, it has disadvantage that analyzes only the codes in the functions without analyzing the relations among source functions. The functions in the source codes are interconnected and formed a social network. Functions that occupy critical locations in a network can be important enough to affect the overall quality. Whereas, a static analysis tool merely suggests which functions were called several times. In this study, the core functions will be elicited by using social network analysis and DEA (Data Envelopment Analysis) for CUBRID open database sources. In addition, we will suggest clear criteria for selecting the target sources for code inspection and will suggest ways to find core functions to minimize defects and improve performance.

• Knowledge Management Research
• /
• v.17 no.3
• /
• pp.23-44
• /
• 2016

This study examined the cause-and-effect relationships among developmental network characteristics, developmental functions, and career identity. The study analyzed the data of employees with more than two years of experience and a college degree or over qualifications, from domestic companies. For analysis, the structural equation model was applied. Most of all, it was substantiated that the structural relationships among the developmental network characteristics, developmental functions, and career identity. The specific results were as follows. First, tie strength had a positive effect on all of developmental functions but psychosocial support mediated the relationship between the tie strength and career identity. Second, an indirect relationship between indicators of network diversity and identity was not found or the indirect relationship was very weak.

• Journal of the Korea Society of Computer and Information
• /
• v.23 no.12
• /
• pp.49-55
• /
• 2018

In this paper, we propose schemes of blockchain utilization in tactical communication environment. The military tactical communication environment has similar characteristics with blockchain network such as distributed architecture, decentralization, and the need for data integrity. A communication node constituting a tactical communication network is constituted by a system capable of configuring and connecting a network for each node. When a communication node, having such capabilities, is configured as a node of blockchain network, various functions could be performed. In this paper, we propose utilization schemes of authentication, integrity, record management, and privilege control based blockchain technology. Functions for authentication, integrity verification, and record management need to ensure the stored data and could track history. The requirement of function's characteristics are matched to blockchain which is storing data sequentially and difficult to hack data, so that it could perform functionally and sufficiently well. Functions for authority control should be able to assign different privileges according to the state of the requestor. Smart contract will function when certain conditions are satisfied and it will be able to perform its functions by using it. In this paper, we will look over functions and utilization schemes of blockchain technology which could reliably share and synchronize data in a tactical communication environment composed of distributed network environment.

• Transactions on Control, Automation and Systems Engineering
• /
• v.2 no.3
• /
• pp.169-174
• /
• 2000

In this paper, Haar wavelet-based neural network is described for the identification and control of discrete-time nonlinear dynamical systems. Wavelets are suited to depict functions with local nonlinearities and fast variations because of their intrinsic properties of finite support and self-similarity. Due to the orthonormal properties of Haar wavelet functions, wavelet neural networks result in a greatly simplified training problem. This wavelet-based scheme performs adaptively both the identification of nonlinear functions and the control of the overall system, while the multilayer neural network is applied to the control system just after its sufficient learning of the unknown functions. Simulation shows that the wavelet network can be a good alternative to a multilayer neural network with backpropagation.

• Journal of Applied and Pure Mathematics
• /
• v.4 no.5_6
• /
• pp.269-286
• /
• 2022

In this article we exhibit univariate and multivariate quantitative approximation by Kantorovich-Choquet type quasi-interpolation neural network operators with respect to supremum norm. This is done with rates using the first univariate and multivariate moduli of continuity. We approximate continuous and bounded functions on ℝ^N , N ∈ ℕ. When they are also uniformly continuous we have pointwise and uniform convergences. Our activation functions are induced by the arctangent, algebraic, Gudermannian and generalized symmetrical sigmoid functions.

• Proceedings of the KIEE Conference
• /
• 1997.11a
• /
• pp.142-145
• /
• 1997

This paper presents an modified orthogonal neural network(MONN) based on orthogonal functions and applies the network to nonlinear system control. The accuracy of orthogonal neural network is essentially dependent on the choice of basic orthogonal functions. Modified orthogonal neural network is modified model of orthogonal neural network with input transformation to adapt its basic orthogonal functions. The results show that the modified orthogonal neural network has the excellent performance of approximating and controlling nonlinear systems and the input transformation make the ability of modified orthogoneural neural network better than one of orthogonal neural network.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.4 no.2
• /
• pp.211-216
• /
• 2004

Fuzzy neural networks have been successfully applied to analyze/generate predictive rules for medical or diagnostic data. However, most approaches proposed so far have not considered the weights for the membership functions much. This paper presents a neural network with weighted fuzzy membership functions. In our approach, the membership functions can capture the concentrated and essential information that affects the classification of the input patterns. To verify the performance of the proposed model, well-known Iris data set is performed. According to the results, the weighted membership functions enhance the prediction accuracy. The architecture of the proposed neural network with weighted fuzzy membership functions and the details of experimental results for the data set is discussed in this paper.

• Journal of Korea Society of Digital Industry and Information Management
• /
• v.9 no.4
• /
• pp.1-10
• /
• 2013

Recently many documents have been written by HTML due to the expansion of the Internet and WWW. HTML allows scripting languages to interact between an user and an browser. Sometimes there might have problems with some functions of the script language. Some functions have the same name although the different contents and vice versa. It presents tremendous difficulties to maintain the functions. And, a file includes many functions shared among documents. If a user tries to see a HTML document, the file will be sent to the web browser in the user system. The network resource is wasted due to moving all data in the file through the network although some functions in the file is not used. So, we should design and implement the code library of a script language with a database. Each function has the name and arguments. A system of the code library manages functions with them. The system has the functions consistency and reduces the network resource consumption because there are only necessary functions in documents. Beside it is easy for the system to modify and reuse the functions.

• Journal of Broadcast Engineering
• /
• v.17 no.1
• /
• pp.37-48
• /
• 2012

These days many DLNA home media network services start in home network area. DLNA networking softwares included in DLNA media devices contains all of server, renderer and controller functions. But a home media device in home media network environment not uses all of DLNA functions but uses only limited functions. Using all of integrated DLNA functions wastes networking and processing resources especially in home media network, because a home media device uses very limited functions. For example, a smart-TV has a main role of playback and a network attached storage has a main role of media content storage. In this paper, we analyze integrated DLNA networking softwares and design DLNA software models representing each home media device so as to utilize device resources efficiently. And we implement DLNA softwares for smart-TV and smart-phone and test resource efficiency.