• Journal of Intelligence and Information Systems
• /
• v.14 no.2
• /
• pp.193-206
• /
• 2008

Medical diagnosis can be considered a classification task which classifies disease types from patient's condition data represented by a set of pre-defined attributes. This study proposes a hybrid genetic algorithm based classification method to develop classifiers for multidimensional pattern classification problems related with medical decision making. The classification problem can be solved by identifying separation boundaries which distinguish the various classes in the data pattern. The proposed method fits a finite number of regional agents to the data pattern by combining genetic algorithms and local adaptive operations. The local adaptive operations of an agent include expansion, avoidance and relocation, one of which is performed according to the agent's fitness value. The classifier system has been tested with well-known medical data sets from the UCI machine learning database, showing superior performance to other methods such as the nearest neighbor, decision tree, and neural networks.

• Journal of KIISE:Information Networking
• /
• v.35 no.5
• /
• pp.395-401
• /
• 2008

In order to determine path(s) satisfied with bandwidth-guaranteed in the Internet, routers should have information on network topology and link state. The information is stored in Link State Database (LSDB) located in each router and managed. If link states information is changed, routers inform their neighbor of link state information changed by sending Link State Update (LSU) messages. However, there is trade-off between reflection of actual link state information on LSDB and cost of sending LSU messages. To find a bandwidth-guaranteed path effectively, it is important to decide whether LSU messages are sent or not for the change of link sate. In this paper, we propose a threshold-based LSU algorithm using statistic to effectively decide for sending LSU messages and evaluates its performance by intensive simulations. Simulation results show that the performance of proposed scheme is superior to the existing LSU schemes.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.42 no.5 s.335
• /
• pp.15-22
• /
• 2005

This paper proposes an efficient structure which is able to perform self-detection and self-repair for faults in a digital system by imitating the structure of living beings. The self-repair system is composed of artificial cells, which have homogeneous structures in the two-dimension, and spare cells. An artificial cell is composed of a logic block based on multiplexers, and a genome block, which controls the logic block. The cell is designed using DCVSL (differential cascode voltage switch logic) structure to self-detect faults. If a fault occurs in an artificial cell, it is self-detected by the DCVSL. Then the artificial cells which belong to the column are disabled and reconfigured using both neighbour cells and spare cells to be repaired. A self-repairable 2-bit up/down counter has been fabricated using Hynix $0.35{\mu}m$ technology with $1.14{\times}0.99mm^2$ core area and verified through the circuit simulation and chip test.

• The Journal of Korean Association of Computer Education
• /
• v.5 no.3
• /
• pp.37-44
• /
• 2002

Automatic text classification is a task of assigning predefined categories to free text documents. Its importance is increased to organize and manage a huge amount of text data. There have been some researches on automatic text classification based on machine learning techniques. While most of them was focused on proposal of a new machine learning methods and cross evaluation between other systems, a through evaluation or optimization of a method has been rarely been done. In this paper, we propose an improving method of kNN-based Korean text classification system using heuristic informations about decision function, the number of nearest neighbor, and feature selection method. Experimental results showed that the system with similarity-weighted decision function, global method in considering neighbors, and DF/ICF feature selection was more accurate than simple kNN-based classifier. Also, we found out that the performance of the local method with well chosen k value was as high as that of the global method with much computational costs.

• Journal of KIISE:Software and Applications
• /
• v.30 no.5_6
• /
• pp.540-549
• /
• 2003

In collaborative recommendation two models are generally used: the user model and the article model. A user model learns correlation between users preferences and recommends an article based on other users preferences for the article. Similarly, an article model learns correlation between preferences for articles and recommends an article based on the target user's preference for other articles. In this paper, we investigates various combination methods of the user model and the article model for better recommendation performance. They include simple sequential and parallel methods, perceptron, multi-layer perceptron, fuzzy rules, and BKS. We adopt the multi-layer perceptron for training each of the user and article models. The multi-layer perceptron has several advantages over other methods such as the nearest neighbor method and the association rule method. It can learn weights between correlated items and it can handle easily both of symbolic and numeric data. The combined models outperform any of the basic models and our experiments show that the multi-layer perceptron is the most efficient combination method among them.

• Journal of the Korean Institute of Gas
• /
• v.5 no.4 s.16
• /
• pp.85-91
• /
• 2001

This paper presents safety analysis of LNG leakage in a prestressed concrete outer tank, which is strongly related on the leak checking effects of the PC structure with and without a residual compression zone based on the BS 7777 codes. The full containment type outer tank which is constructed by a prestressed concrete may be destroyed by leaked cryogenic fluids. The FE calculated results show that the total leak checking time of the PC structure with $10\%$ residual compression zone is about 9 days for $-162^{\circ}C$ liquids. But, three primary pumps in an inner tank may operate to send cryogenic fluids for 6 days, which are stored in an inner tank of $140,000m^3$ capacity This means that the prestressed concrete outer tank may be safe for $-162^{\circ}C$ cryogenic fluids leaked from the demolished inner tank.

• Journal of Korea Multimedia Society
• /
• v.4 no.5
• /
• pp.406-413
• /
• 2001

Ordinary high-speed block matching algorithms have a disadvantage that they need to get MAD (Mean Absolute Distance) as many as the number of search points due to comparing the MAD between the current frame's search block and the reference frame's search block. To solve such disadvantage of high-speed block matching algorithm, the proposed high-speed DS algorithm employs a MMAD calculation method using a specific characteristic that neighboring pixels have almost same values. In this thesis, we can get rid of unnecessary MAD calculation between the search point block by the new calculation method which uses the previously calculated MAD as the current search point and by breaking from the established MAD calculation method which calculates the MAD of a new search point by each search stage. Comparing with the established high-speed block matching algorithm, this new calculation's estimated movement error was shown as similar, and th total calculation amount decreased by $2FN^2Ep$.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.35 no.5B
• /
• pp.721-733
• /
• 2010

In this paper, we propose a scalable content-based routing scheme for the tactical mobile ad-hoc networks where the nodes usually move together forming a group. Our scheme is classified as a content-based routing scheme based on publish and subscribe model. The nodes compose a cluster according to their position and interest. Our scheme allows multi-level cluster nesting. As a publish/subscribe model does, each node announces its interest to the cluster head who aggregates and announces the interests from cluster members to its neighbor cluster head. Actual data messages are delivered to the end node via cluster header in the reverse direction of the interest announcement. Using Qualnet simulator, we evaluated the performance of our scheme in comparison with two well-known content-based routing schemes. Simulation results show that our scheme maintains higher message delivery ratio as number of nodes increases whereas the two schemes show much lower delivery ratio. In addition, as the group mobility gets faster our scheme incurs less or similar control message overhead compared with the two schemes.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.10 no.4
• /
• pp.479-486
• /
• 1999

In this paper the system performance degradation resulting from nonlinear transmitter power amplifier which is essential to increase the efficiency is analyzed in a forward link CDMA system. The power amplifier is modeled by power series model which includes only odd-order terms. The effects of power amplifier's nonlinearity such as intersymbol interference, phase distortion on the RF system performance were visualized by examining the distorted time domain waveforms, signal vector constellation. And through the investigation of the power spectrum density of the transmitted signal, spectral regrowth or sideband regrowth which is result from amplitude distortion can be seen. All these characteristics result in BER performance degradation due to other user interferences and intersymbol interference. The analysis technique described here applies not only to power amplifier but also to any other nonlinear components such as mixers and switches. Also the effects of adjacent channel interference and supurious emission can be analysed between different systems.

• The KIPS Transactions:PartB
• /
• v.8B no.5
• /
• pp.579-586
• /
• 2001

Digital watermarking is to embed imperceptible mark into image, video, audio, and text data to prevent the illegal copy of multimedia data. arbitrary modification, and also illegal sales of the copies without agreement of copyright ownership. In this paper, we study for the embedding and extraction of watermark key using wavelet in the luminance signal in order to implement the system to protect the copyright for image MPEG. First, the original image is analyzed into frequency domain by discrete wavelet transform. The RSA(Rivest, Shamir, Aldeman) public key of the coded target is RUN parameter of VLD(variable length coding). Because the high relationship among the adjacent RUN parameters effect the whole image, it prevents non-authorizer not to possess private key from behaving illegally. The Results show that the proposed method provides better moving picture and the distortion more key of insert than direct coded method on low-frequency domain based DCT.