• 한국컴퓨터정보학회논문지
• /
• 제21권5호
• /
• pp.79-89
• /
• 2016

In this paper, we proposed the object classification method using genetic and dynamic random forest consisting of optimal combination of unit tree. The random forest can ensure good generalization performance in combination of large amount of trees by assigning the randomization to the training samples and feature selection, etc. allocated to the decision tree as an ensemble classification model which combines with the unit decision tree based on the bagging. However, the random forest is composed of unit trees randomly, so it can show the excellent classification performance only when the sufficient amounts of trees are combined. There is no quantitative measurement method for the number of trees, and there is no choice but to repeat random tree structure continuously. The proposed algorithm is composed of random forest with a combination of optimal tree while maintaining the generalization performance of random forest. To achieve this, the problem of improving the classification performance was assigned to the optimization problem which found the optimal tree combination. For this end, the genetic algorithm methodology was applied. As a result of experiment, we had found out that the proposed algorithm could improve about 3~5% of classification performance in specific cases like common database and self infrared database compare with the existing random forest. In addition, we had shown that the optimal tree combination was decided at 55~60% level from the maximum trees.

• 정보처리학회논문지A
• /
• 제9A권3호
• /
• pp.295-300
• /
• 2002

본 논문에서는 3차원 입출력 알고리즘을 랜덤 인터리버와 s-랜덤 인터리버에 적용하였으며, 이를 터보코드 인터리버에 적용하여 성능을 분석하였다. 인터리버의 성능은 인접 데이터간 최소 거리에 의해 결정되어지므로, 인접 데이터간의 최소거리를 증가시키는 방법으로 인터리버의 성능을 향상 시켰다. 3차원 알고리즘을 적용한 인터리버는 3차원 저장공간을 이용해 입력 데이터를 저장하고 랜덤하게 추출하는 방식이다. 이러한 방식은 기존의 랜덤 인터리버와 s-랜덤 인터리버에 비해 인접 데이터간 최소거리와 평균거리를 증가시킨다. 컴퓨터 시뮬레이션을 이용하여 3차원 알고리듬을 적용한 터보코드의 성능을 분석하였으며, 전송 환경을 가우시안 채널로 설정하였다.

• 한국정보통신학회논문지
• /
• 제12권2호
• /
• pp.360-368
• /
• 2008

본 논문에서는 다중 서비스 다중 사용자 OFDMA 시스템의 임의 접근채널에서 차별화된 서 비스 품질을 제공하기 위하여 backoff기반 임의 접근 알고리즘을 제안하고, 임의 접근 채널의 주 자원인 부채널의 수와 PN-코드의 수가 주어진 경우, 제안된 알고리즘의 성능을 각 서비스 클래스 접속 확률의 함수로서, 각 서비스 클래스의 접속 성공 확률, 처리율, 블로킹 확률, 접속 지연관점에서 분석한다. 수치적 분석을 통하여 제안된 backoff 기반 임의 접근 알고리즘이 서로 다른 서비스 클래스의 임의접근 시도들에게 차등한 서비스 품질을 제공할 수 있음을 보였다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2006년도 심포지엄 논문집 정보 및 제어부문
• /
• pp.321-323
• /
• 2006

The IEEE 802.15.4 uses a CSMA/CA algorithm on access of media. The CSMA/CA algorithm does Random Backoff before the data is transmitted to avoid collisions. The random backoff is a kind of unavoidable delays and introduces the side effect of energy consumptions. To cope with those problems we propose a new media access algorithm, the Priority Based Ordered Media Access (PBOMA) algorithm, which uses different IFSs. The PBOMA algorithm uses Sampling Rate and Beacon Interval to get a different access probability(or IFS). The access probability is higher, the IFS is shorter. Note that The transfer of urgent data uses tone signal to transmit it immediately. The proposed algorithm is expected to reduce the energy consumptions and the delay.

• 제어로봇시스템학회논문지
• /
• 제20권8호
• /
• pp.854-867
• /
• 2014

This paper deals with the performance comparison of a PSO algorithm inspired in the process of simulating the behavior pattern of the organisms. The PSO algorithm finds the optimal solution (fitness value) of the objective function based on a stochastic process. Generally, the stochastic process, a random function, is used with the expression related to the velocity included in the PSO algorithm. In this case, the random function of the normal distribution (Gaussian) or uniform distribution are mainly used as the random function in a PSO algorithm. However, in this paper, because the probability distribution which is various with 2 shape parameters can be expressed, the performance comparison of a PSO algorithm using the beta probability distribution function, that is a random function which has a high degree of freedom, is introduced. For performance comparison, 3 functions (Rastrigin, Rosenbrock, Schwefel) were selected among the benchmark Set. And the convergence property was compared and analyzed using PSO-FIW to find the optimal solution.

• Journal of Information Processing Systems
• /
• 제20권2호
• /
• pp.215-225
• /
• 2024

With the rapid development of Internet of Things (IoT) and big data technology, a large amount of data will be generated during the operation of related industries. How to classify the generated data accurately has become the core of research on data mining and processing in IoT industry chain. This study constructs a classification model of IoT industry chain based on improved random forest algorithm and text analysis, aiming to achieve efficient and accurate classification of IoT industry chain big data by improving traditional algorithms. The accuracy, precision, recall, and AUC value size of the traditional Random Forest algorithm and the algorithm used in the paper are compared on different datasets. The experimental results show that the algorithm model used in this paper has better performance on different datasets, and the accuracy and recall performance on four datasets are better than the traditional algorithm, and the accuracy performance on two datasets, P-I Diabetes and Loan Default, is better than the random forest model, and its final data classification results are better. Through the construction of this model, we can accurately classify the massive data generated in the IoT industry chain, thus providing more research value for the data mining and processing technology of the IoT industry chain.

• 한국산학기술학회논문지
• /
• 제13권3호
• /
• pp.1147-1152
• /
• 2012

본 논문은 의학 분야에서 사용되는 하드웨어 발생기 디지털 난수 출력열의 특성 처리 분석을 주요 이슈로 한다. 하드웨어 이진 난수를 기반으로 하는 난수발생기의 출력열은 지연, 지터, 온도 등의 요소로부터 영향을 받는다. 본 논문에서는 하드웨어 출력 난수열에 영향을 주는 주요 요소에 대해 살펴보고, 출력열과 암호알고리즘, 부호알고리즘이 결합된 출력열의 난수성을 분석하였다. 분석된 결과는 난수성 주요 검증 항목에 의해 평가되었다.

• Journal of applied mathematics & informatics
• /
• 제5권2호
• /
• pp.357-372
• /
• 1998

In this paper we introduce and study a new class of random completely generalized strongly nonlinear quasi -comple- mentarity problems with non-compact valued random fuzzy map-pings and construct some new iterative algorithms for this kind of random fuzzy quasi-complementarity problems. We also prove the existence of random solutions for this class of random fuzzy quasi-complementarity problems and the convergence of random iterative sequences generated by the algorithms.

• 한국전자거래학회지
• /
• 제6권2호
• /
• pp.167-178
• /
• 2001

Key is generally formed using the Random Number. How to make the Random Number is to cast coin or dice as classical method, to form the Real Random Number with Hardware and to make the Pseudo Random Number by means of utilizing mathematical algorithm. This thesis presented NRNG(New Random Number Generator) which put self-development Hardware to use as Key Generation Method and inspected to compare the Real Random Number with the Pseudo Random Number and special properties which PRNG(Pseudo-Random Number Generator) creates.

• 한국인공지능학회지
• /
• 제10권1호
• /
• pp.21-25
• /
• 2022

Korea has relatively less crime than other countries. However, the crime rate is steadily increasing. Many people think the crime rate is decreasing, but the crime arrest rate has increased. The goal is to check the relationship between CCTV and the crime rate as a way to lower the crime rate, and to identify the correlation between areas without CCTV and areas without CCTV. If you see a crime that can happen at any time, I think you should use a random forest algorithm. We also plan to use machine learning random forest algorithms to reduce the risk of overfitting, reduce the required training time, and verify high-level accuracy. The goal is to identify the relationship between CCTV and crime occurrence by creating a crime prevention algorithm using machine learning random forest techniques. Assuming that no crime occurs without CCTV, it compares the crime rate between the areas where the most crimes occur and the areas where there are no crimes, and predicts areas where there are many crimes. The impact of CCTV on crime prevention and arrest can be interpreted as a comprehensive effect in part, and the purpose isto identify areas and frequency of frequent crimes by comparing the time and time without CCTV.