• International Journal of Computer Science & Network Security
• /
• 제22권1호
• /
• pp.167-174
• /
• 2022

In metaheuristic algorithms such as Genetic Algorithm (GA), initial population has a significant impact as it affects the time such algorithm takes to obtain an optimal solution to the given problem. In addition, it may influence the quality of the solution obtained. In the machine learning field, feature selection is an important process to attaining a good performance model; Genetic algorithm has been utilized for this purpose by scientists. However, the characteristics of Genetic algorithm, namely random initial population generation from a vector of feature elements, may influence solution and execution time. In this paper, the use of a statistical algorithm has been introduced (Chi2) for feature relevant checks where p-values of conditional independence were considered. Features with low p-values were discarded and subject relevant subset of features to Genetic Algorithm. This is to gain a level of certainty of the fitness of features randomly selected. An ensembled-based learning model for Hepatitis has been developed for Hepatitis C stage classification. 1385 samples were used using Egyptian-dataset obtained from UCI repository. The comparative evaluation confirms decreased in execution time and an increase in model performance accuracy from 56% to 63%.

• Steel and Composite Structures
• /
• 제31권5호
• /
• pp.427-435
• /
• 2019

The estimation of moment and rotation in steel rack connections could be significantly helpful parameters for designers and constructors in the initial designing and construction phases. Accordingly, Extreme Learning Machine (ELM) has been optimized to estimate the moment and rotation in steel rack connection based on variable input characteristics as beam depth, column thickness, connector depth, moment and loading. The prediction and estimating of ELM has been juxtaposed with genetic programming (GP) and artificial neural networks (ANNs) methods. Test outcomes have indicated a surpass in accuracy predicting and the capability of generalization in ELM approach than GP or ANN. Therefore, the application of ELM has been basically promised as an alternative way to estimate the moment and rotation of steel rack connection. Further particulars are presented in details in results and discussion.

• 한국정보통신학회논문지
• /
• 제25권4호
• /
• pp.515-522
• /
• 2021

본 논문에서는 관형 철탑의 용접부 결함을 상시적으로 감시하기 위하여 초음파 탐상 신호에 대한 기계학습 알고리즘의 적용 방법을 제시하고 분석하였다. 기계학습 방법으로는 유전자 알고리즘에 의한 특징 선택과 서포트 벡터머신을 이용한 탐상 신호 분류 방법을 사용하였다. 특징 선택에서는 30개의 후보 특징들 가운데 피크, 히스토그램 하한 경계, 정규 음로그우도가 선택되었으며, 이들은 결함의 깊이에 따른 신호의 차이를 명확하게 나타내었다. 또한, 선택된 특징들을 서포트 벡터 머신에 적용한 결과 정상 부위와 결함 부위를 완벽하게 분류할 수 있는 것으로 나타났다. 따라서 본 연구의 결과는 향후 초음파 신호 기반 결함 성장 조기 감지시스템의 개발과 이를 통한 에너지 송전 관련 산업에 유용하게 사용될 수 있을 것으로 기대된다.

• 인터넷정보학회논문지
• /
• 제13권4호
• /
• pp.45-52
• /
• 2012

감염된 미생물에서 유래한 단백질 펩타이드가 HLA에 결합하여 숙주의 세포표면에 제시되면, T 세포가 이를 인식하여 면역반응을 유발함으로써 감염원을 제거하게 된다. HLA와 펩타이드간의 결합이 안정적일수록 T 세포반응이 강하게 일어나 효율적으로 감염원을 제거할 수 있다고 알려져 있다. 따라서 특정 HLA에 안정적으로 결합할 수 있는 펩타이드(HLA binder)를 찾아낼 수 있다면 감염질환이나 암의 예방을 위한 펩타이드 백신의 개발에 활용될 수 있다. 그런데 HLA는 매우 다형하기 때문에 하나의 집단 내에서도 어느 정도의 빈도를 가지는 대립유전자의 수가 매우 많다. 따라서 이들 모든 대립유전자들에 대해 가능한 펩타이드조합을 제작한 후 직접 실험을 통해 안정적으로 결합하는 펩타이드를 찾아내는 것은 매우 비효율적이다. 이를 극복하기 위하여 특정 HLA에 안정적으로 결합하는 펩타이드를 예측하는 정보전산적인 방법이 최근 개발되어 왔다. 이들 방법을 통해 제시된 펩타이드에 대해서만 직접 생물학적 실험을 시행함으로써 연구자는 검증해야 할 후보 펩타이드의 수를 현격히 감소시킬 수 있게 된다. 본 논문에서는 HLA 결합 펩타이드 예측을 위해 기계학습을 이용한 방법을 소개할 뿐만 아니라, 지금까지 HLA 결합 펩타이드 예측에 시도된 적이 없는 '지식기반 유전자 알고리즘(knowledge-based genetic algorithm)'이라는 새로운 모델을 제시하고자 한다. 이것은 유전자알고리즘(GA)에 기반한 것이었지만 전문가 지식을 접목함으로써 GA보다 더 향상된 성능으로 한국인에 흔한 HLA에 결합하는 펩타이드를 예측하였다. 뿐만 아니라 이것은 결합하는 펩타이드의 규칙을 한국인에 흔한 HLA 대립유전자에 대하여 추출해 줄 수 있는 새로운 방법이었다.

• 한국지능시스템학회:학술대회논문집
• /
• 한국지능시스템학회 2007년도 추계학술대회 학술발표 논문집
• /
• pp.283-286
• /
• 2007

This paper proposes a genetic algorithm (GA) approach to instance selection in case-based reasoning (CBR) for the prediction of Korea Stock Price Index (KOSPI). CBR has been widely used in various areas because of its convenience and strength in complex problem solving. Nonetheless, compared to other machine learning techniques, CBR has been criticized because of its low prediction accuracy. Generally, in order to obtain successful results from CBR, effective retrieval of useful prior cases for the given problem is essential. However, designing a good matching and retrieval mechanism for CBR systems is still a controversial research issue. In this paper, the GA optimizes simultaneously feature weights and a selection task for relevant instances for achieving good matching and retrieval in a CBR system. This study applies the proposed model to stock market analysis. Experimental results show that the GA approach is a promising method for instance selection in CBR.

• 한국농공학회논문집
• /
• 제64권5호
• /
• pp.27-39
• /
• 2022

The smart farm is recognized as a solution for future farmers having positive effects on the sustainability of the poultry industry. Intelligent microclimate control can be a key technology for broiler production which is extremely vulnerable to abnormal indoor air temperatures. Furthermore, better control of indoor microclimate can be achieved by accurate prediction of indoor air temperature. This study developed predictive models for internal air temperature in a mechanically-ventilated broiler house based on the data measured during three rearing periods, which were different in seasonal climate and ventilation operation. Three machine learning models and a mechanistic model based on thermal energy balance were used for the prediction. The results indicated that the all models gave good predictions for 1-minute future air temperature showing the coefficient of determination greater than 0.99 and the root-mean-square-error smaller than 0.306℃. However, for 1-hour future air temperature, only the mechanistic model showed good accuracy with the coefficient of determination of 0.934 and the root-mean-square-error of 0.841℃. Since the mechanistic model was based on the mathematical descriptions of the heat transfer processes that occurred in the broiler house, it showed better prediction performances compared to the black-box machine learning models. Therefore, it was proven to be useful for intelligent microclimate control which would be developed in future studies.

• 전자공학회논문지 IE
• /
• 제44권3호
• /
• pp.35-41
• /
• 2007

본 논문에서는 보상신호를 수반하는 인공지능 기반의 가상 로봇 학습 행위 모델을 제안하고 이 모델을 3가지 환경에 적용시킨 후에 보상 방법에 따른 가상 로봇의 학습 속도를 비교 검토하였다. 결과로서 환경이 다소 복잡하면 즉, 로봇 집단의 크기, 먹이 수, 장애물 수가 다소 많은 경우 학습 세대가 충분하다면 강화 보상 방법이 강화와 억제를 혼합한 보상 방법 보다 우월함을 알 수 있었다. 하지만 복잡하지 않은 환경에서는 혼합 보상 방법이 우수했다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1997년도 하계학술대회 논문집 D
• /
• pp.781-783
• /
• 1997

Genetic Algorithm is a kind of an evolution programming based on natural evolution principle. It applied to probabilistic searching, machine learning and optimization, and many good results were reported. Generator maintenance scheduling is an optimization Problem with constraints. This paper applied a genetic algorithm to generator maintenance scheduling problem and tested on sample systems. The results are compared with heuristic method and branch-and-bound method.

• ETRI Journal
• /
• 제44권2호
• /
• pp.327-338
• /
• 2022

Classification models for human action recognition require robust features and large training sets for good generalization. However, data augmentation methods are employed for imbalanced training sets to achieve higher accuracy. These samples generated using data augmentation only reflect existing samples within the training set, their feature representations are less diverse and hence, contribute to less precise classification. This paper presents new data augmentation and action representation approaches to grow training sets. The proposed approach is based on two fundamental concepts: virtual video generation for augmentation and representation of the action videos through robust features. Virtual videos are generated from the motion history templates of action videos, which are convolved using a convolutional neural network, to generate deep features. Furthermore, by observing an objective function of the genetic algorithm, the spatiotemporal features of different samples are combined, to generate the representations of the virtual videos and then classified through an extreme learning machine classifier on MuHAVi-Uncut, iXMAS, and IAVID-1 datasets.

• International Journal of Internet, Broadcasting and Communication
• /
• 제16권2호
• /
• pp.119-126
• /
• 2024

The volume of genomic data is constantly increasing in various modern industries and research fields. This growth presents new challenges and opportunities in terms of the quantity and diversity of genetic data. In this paper, we propose a distributed cloud system for integrating and managing large-scale gene databases. By introducing a distributed data storage and processing system based on the Hadoop Distributed File System (HDFS), various formats and sizes of genomic data can be efficiently integrated. Furthermore, by leveraging Spark on YARN, efficient management of distributed cloud computing tasks and optimal resource allocation are achieved. This establishes a foundation for the rapid processing and analysis of large-scale genomic data. Additionally, by utilizing BigQuery ML, machine learning models are developed to support genetic search and prediction, enabling researchers to more effectively utilize data. It is expected that this will contribute to driving innovative advancements in genetic research and applications.