Articles

Access the latest knowledge in applied science, electrical engineering, computer science and information technology, education, and health.

Advanced Search
Filter Icon

Filters article

Years

FAQ Arrow
Single Year
Range Year
0
0

Source Title

FAQ Arrow

Authors

FAQ Arrow

27,404 Article Results

Oil Palm Yield Forecasting Based on Weather Variables Using Artificial Neural Network

10.11591/ijeecs.v3.i3.pp626-633
Nadia Dwi Kartika , I Wayan Astika , Edi Santosa
Forecasting of oil palm yield has become a main factor in the management of oil palm industries for proper planning and decision making in order to avoid monthly high cost in harvesting. Predicting future value of oil palm yield with minimum error becomes an important issue recently. A lot of factors determine the productivity of oil palm and weather variables play an important role that affect plant growth and development that may reduce yield significantly. This research used secondary data of yield and weather variables available in company administration. It proposed feed forward neural network with back propagation learning algorithm to build a monthly yield forecasting model. The optimization procedure of ANN architecture obtained the best using 60 neurons in input layer, five hidden layers and one neuron in the output layer. Training data were from January 2005 to June 2008 while testing data were from July 2008 to December 2009. ANN architecture using five hidden layers gave the best accuracy with MAE 0.5346 and MSE 0.4707 while the lowest accuracy occurred by using two hidden layers with MAE 1.5843and MSE 4.087.
Volume: 3
Issue: 3
Page: 626-633
Publish at: 2016-09-01

A Comparison of Retweet Prediction Approaches: The Superiority of Random Forest Learning Method

10.12928/telkomnika.v14i3.3150
Hendra; Maranatha Christian University Bunyamin , Tomas; Czech Technical University Tunys
We consider the following retweet prediction task: given a tweet, predict whether it will be retweeted. In the past, a wide range of learning methods and features has been proposed for this task. We provide a systematic comparison of the performance of these learning methods and features in terms of prediction accuracy and feature importance. Specifically, from each previously published approach we take the best performing features and group these into two sets: user features and tweet features. In addition, we contrast five learning methods, both linear and non-linear. On top of that, we examine the added value of a previously proposed time-sensitive modeling approach. To the authors’ knowledge this is the first attempt to collect best performing features and contrast linear and non-linear learning methods. We perform our comparisons on a single dataset and find that user features such as the number of times a user is listed, number of followers, and average number of tweets published per day most strongly contribute to prediction accuracy across selected learning methods. We also find that a random forest-based learning, which has not been employed in previous studies, achieves the highest performance among the learning methods we consider. We also find that on top of properly tuned learning methods the benefits of time-sensitive modeling are very limited.
Volume: 14
Issue: 3
Page: 1052-1058
Publish at: 2016-09-01

Flow Fair Sampling Based on Multistage Bloom Filters

10.12928/telkomnika.v14i3.3648
Liu; Yancheng Institute of Technology Yuanzhen , Huang; Yancheng Institute of Technology Shurong , Liu; Yancheng Institute of Technology Jianzhao
Network traffic distribution is heavy-tailed. Most of network flows are short and carry very few packets, and the number of large flows is small. Traditional random sampling tends to sample more large flows than short ones. However, many applications depend on per-flow traffic other than just large flows. A flow fair sampling based on multistage Bloom filters is proposed. The total measurement interval is divided into n child time intervals. In each child time interval, employ multistage Bloom filters to query the incoming packet’s flow whether exists in flow information table or not, if exists, sample the packet with static sampling rate which is inversely proportional to the estimation flow traffic up to the previous time interval. If it is a new flow’s first packet, create its flow information and insert it into the multistage Bloom filters. The results show that the proposed algorithm is accurate especially for short flows and easy to extend.
Volume: 14
Issue: 3
Page: 1142-1149
Publish at: 2016-09-01

Classification of Motorcyclists not Wear Helmet on Digital Image with Backpropagation Neural Network

10.12928/telkomnika.v14i3.3486
Sutikno; Diponegoro University Sutikno , Indra; Diponegoro University Waspada , Nurdin; Diponegoro University Bahtiar , Priyo Sidik; Diponegoro University Sasongko
One of the world’s leading causes of death is traffic accidents. Data from World Health Organization (WHO) that there are 1.25 million people in the world die each year. Meanwhile, based on data obtained from Statistics Indonesia, traffic accidents from 2006 to 2013 continue to increase. Of all these accidents, the largest accident occurred at motorcyclists, especially motorcyclists who not wearing standard helmet. In controlling the motorcyclists, police view directly at the highway, so that there are weaknesses which there are still a possibility of motorcyclist offenders who are undetectable especially for motorcyclists who are not wear helmet. This paper explains research on image classification of human head wearing a helmet and not wearing a helmet with backpropagation neural network algorithm. The test results of this analysis is the application can performs classification with 86.67% accuracy rate. This research can be developed into a larger system and integrated that can be used to detect motorcyclists who are not wearing helmet.
Volume: 14
Issue: 3
Page: 1128-1133
Publish at: 2016-09-01

A New Semi-supervised Clustering Algorithm Based on Variational Bayesian and Its Application

10.12928/telkomnika.v14i3.3805
Shoulin Yin , Jie Liu , Lin Teng
Biclustering algorithm is proposed for discovering matrix with biological significance in gene expression data matrix and it is used widely in machine learning which can cluster the row and column of matrix. In order to further improve the performance of biclustering algorithm, this paper proposes a semi-supervised clustering algorithm based on variational Bayesian. Firstly, it introduces supplementary information of row and column for biclustering process and represents corresponding joint distribution probability model. In addition, it estimates the parameter of joint distribution probability model based on variational Bayesian learning method. Finally, it estimates the performance of proposed algorithm through synthesized data and real gene expression data set. Experiments show that normalized mutual information of this paper’s new method is better than relevant biclustering algorithms for biclustering analysis.
Volume: 14
Issue: 3
Page: 1150-1156
Publish at: 2016-09-01

Compressive Sensing Algorithm for Data Compression on Weather Monitoring System

10.12928/telkomnika.v14i3.3021
Rika; Indonesian Institute of Sciences (LIPI) Sustika , Bambang; Indonesian Institute of Sciences (LIPI) Sugiarto
Compressive sensing (CS) is new data acquisition algorithm that can be used for compression. CS theory certifies that signals can be recovered from far fewer samples or measurements than Nyquist rate. On this paper, the compressive sensing technique is applied for data compression on our weather monitoring system. On this weather monitoring system, compression using compressive sensing with fewer samples or measurements means minimizing sensing and overall energy cost. Our focus on this paper lies in the selection of matrix for representation basis under which the weather data are sparsely represented. We evaluated three types of representation basis using data from real measurement. By comparing performance of data recovery, result show that DCT (Discrete Cosine Transform) is the best performance on sparsifying weather data
Volume: 14
Issue: 3
Page: 974-980
Publish at: 2016-09-01

Hybrid Hierarchical Collision Detection Based on Data Reuse

10.12928/telkomnika.v14i3.3590
Jiancai; South China University of Technology Hu , Kejing; South China University of Technology He , Xiaobin; South China University of Technology Lin , Funan; South China University of Technology Lin
To improve the efficiency of collision detection between rigid bodies in complex scenes, this paper proposes a method based on hybrid bounding volume hierarchies for collision detection. In order to improve the simulation performance, the method is based on weighted oriented bounding box and makes dense sampling on the convex hulls of the geometric models. The hierarchical bounding volume tree is composed of many layers. The uppermost layer adopts a cubic bounding box, while lower layers employ weighted oriented bounding box. In the meantime, the data of weighted oriented bounding box is reused for triangle intersection check. We test the method using two scenes. The first scene contains two Buddha models with totally 361,690 triangle facets. The second scene is composed of 200 models with totally 115, 200 triangle facets. The experiments verify the effectiveness of the proposed method.
Volume: 14
Issue: 3
Page: 1077-1082
Publish at: 2016-09-01

Study on Community’s Land Allocation in Long Pahangai District

10.11591/ijeecs.v3.i3.pp564-571
Dito Cahya Renaldi , I Nengah Surati Jaya , Omo Rusdiana
Land use allocation for community has been a crucial process for supporting the spatial allocation either at the regency or provincial level. This study was emphasized on the analysis of land allocation at the district level. The study applied a linear programming approach to optimize the land use in Long Pahangai District then linked with the spatial information. The optimization considered several factors, i.e., land productivity, the degree of erosion and the preference of the community living in the study area. To support the optimization, the availability of land use was determined by considering the land capability using the query tools in the Geographic Information System. The level of land capability applied five constraints, namely, slope, drainage, soil texture, effective depth and erosion. The study found that the optimal allocation of land use in the study area are primary forest of 6,635.11 ha (25.19%), secondary forest of 19,025.7 ha (71.9%), mixed plantation area of 289.61 ha (1.1%), settlement area of 8.3 ha (0.03%) and rice field of 487.35 ha (1.844%). This optimal allocation might increase the community income per capita by approximately 80% from 9,602,000.- to 17,275,171.-/capita/ha/year.  
Volume: 3
Issue: 3
Page: 564-571
Publish at: 2016-09-01

GPU CUDA accelerated Image Inpainting using Fourth Order PDE equation

10.12928/telkomnika.v14i3.3412
Edwin Prananta , Pranowo; Atma Jaya Yogyakarta University Pranowo , Djoko; Atma Jaya Yogyakarta University Budianto
This paper describes the technique to accelerate inpainting process using fourth order PDE equation using GPU CUDA. Inpainting is the process of filling in missing parts of damaged images based on information gleaned from surrounding areas. It uses the GPU computation advantage to process PDE equation into parallel process. Fourth order PDE will be solved using parallel computation in GPU. This method can speed up the computation time up to 36x using NVDIA GEFORCE GTX 670.
Volume: 14
Issue: 3
Page: 1009-1015
Publish at: 2016-09-01

Prediction Model of Smelting Endpoint of Fuming Furnace Based on Grey Neural Network

10.12928/telkomnika.v14i3.3713
Song; School of Mechanical Engineering, Anyang Institute of Technology,Henan,China Qiang , WU; School of Mechanical Engineering, Anyang Institute of Technology,Henan,China Yaochun
Since grey theory and neural network could improve prediction precision, the technology of combination prediction was proposed in this study. Then the algorithm was simulated by Matlab using practical data of a fuming furnace. The results reveal that the smelting endpoint of fuming furnace could be accurately predicted with this model by referring to small sample and information. Therefore, GNN model is effective with the advantages of high precision, fewer samples required and simple calculation.
Volume: 14
Issue: 3
Page: 941-947
Publish at: 2016-09-01

Recognition of Fission Signals Based on Wavelet Analysis and Neural Network

10.12928/telkomnika.v14i3.3544
Li; South West University of Science & Technology Li , Liu Keqi , Hu; Academy of Engineering Physics Gen
Because of the particularity of the uranium components, the nondestructive measuring technique is needed to detect the radioactivity of the component in certain container and identify their property to recognize all kinds of uranium components. This paper establishes a set of samples with the same shape, different weight and abundance of uranium by simulation. Secondly the cross-correlation function of time-relation signal between the source detector and the detector could be calculated. Lastly the result of cross-correlation functions is through micro-wavelet analysis to obtain feature vector which is related to the quality and abundance property of target uranium components. This vector is used to train neural network and help to identify the quality and abundance of unknown uranium components.
Volume: 14
Issue: 3
Page: 1016-1023
Publish at: 2016-09-01

Particle Swarm Optimization Performance: Comparison of Dynamic Economic Dispatch with Dantzig-Wolfe Decomposition

10.12928/telkomnika.v14i3.4054
Mohd Ruddin; Universiti Teknikal Malaysia Melaka Ab Ghani , Saif Tahseen; Ministry of Electricity Hussein , Zanariah; Universiti Teknikal Malaysia Melaka Jano , Tole; Universitas Ahmad Dahlan Sutikno
Economic Dispatch (ED) problem, in practice, is a nonlinear, non-convex type,which has developed gradually into a serious task management goal in the planning phase of the power system. The prime purpose of Dynamic Economic Dispatch (DED) is to minimize generators’ total cost of the power system. DED is to engage the committed generating units at a minimum cost to meet the load demand while fulfilling various constraints. Utilizing heuristic, population-based, and advanced optimization technique, Particle Swarm Optimization (PSO), represents a challenging problem with large dimension in providing a superior solution for DED optimization problem. The feasibility of the PSO method has been demonstrated technically, and economically for two different systems, and it is compared with the Dantzig-Wolfe technique regarding the solution quality and simplicity of implementation. While Dantzig-Wolfe method has its intrinsic drawbacks and positive features, PSO algorithm is the finest and the most appropriate solution. Conventional techniques have been unsuccessful to present compatible solutions to such problems due to their susceptibility to first estimates and possible entrapment into local optima which may complicate computations.
Volume: 14
Issue: 3
Page: 1042-1051
Publish at: 2016-09-01

A Sentiment Knowledge Discovery Model in Twitter’s TV Content Using Stochastic Gradient Descent Algorithm

10.12928/telkomnika.v14i3.2671
Lira; Bogor Institute of Agriculture Ruhwinaningsih , Taufik; Bogor Institute of Agriculture Djatna
The use of social media that the explosive can be a rich source for data mining. Meanwhile, the development of television programs become increased and varied so motivate people to make comments on it’s via social media. Social network contains abundant information which is unstructured, heterogeneous, high dimensional and incremental in nature. Abundant data can be a rich source of information but it is difficult to identify manually. The contributions of this research are to perform preprocessing to address unstructured data, a lot of noise and heterogeneous; find patterns of information and knowledge of social media user activities in the form of positive and negative sentiment on twitter TV content. Some methodologies and techniques are used to perform preprocessing. They are eliminates punctuation and symbols, eliminates number, replace numbers into letters, translation of Alay words, eliminate stop word and Stemming Porter Algorithm. Methodology of this study was used Stochastic Gradient Descent (SGD).The text that has been through preprocessing produces a more structured text, reducing noise and reducing the diversity of text. So, preprocessing affect to the correctly classified istances and processing time. The experiment results reveal that the use of SGD for discovery of the positive and negative sentiment tends to be faster for large data or stream data. Correctly classified instance with a maximum of 88%.
Volume: 14
Issue: 3
Page: 1067-1076
Publish at: 2016-09-01

Quasi-Newton Method for Absolute Value Equation Based on Upper Uniform Smoothing Approximation

10.12928/telkomnika.v14i3.3785
Longquan; Shaanxi University of Technology Yong , Shouheng; Shaanxi University of Technology Tuo
In this paper, an upper uniform smooth approximation function of absolute value function is proposed, and some properties of uniform smooth approximation function are studied. Then, absolute value equation (AVE), Ax - |x| = b, where A is a square matrix whose singular values exceed one, is transformed into smooth optimization problem by using the upper uniform smooth approximation function, and solved by quasi-Newton method. Numerical results in solving given AVE problems demonstrated that our algorithm is valid and superior to lower uniform smooth approximation function.
Volume: 14
Issue: 3
Page: 1134-1141
Publish at: 2016-09-01

Medical Image Contrast Enhancement via Wavelet Homomorphic Filtering Transform

10.12928/telkomnika.v14i3.3118
Xinmin; Hunan University of Commerce Zhou , Ying; Tongji University Zheng , Lina; Hunan University of Commerce Tan , Junchan; Hunan University of Commerce Zhao
A novel enhancement algorithm for magnetic resonance (MR) images based on spatial homomorphic filtering transform is proposed in this paper. By this method, the source image is decomposed into different sub-images by dyadic wavelet transform. Homomorphic filtering functions are applied in performing filtering of corresponding sub-band images to attenuate the low frequencies as well as amplify the high frequencies, and a linear adjustment is carried out on the low frequency of the highest level. Later, inverse dyadic wavelet transform is applied to reconstruct the object image. Experiment results on MR images illustrate that the proposed method can eliminate non-uniformity luminance distribution effectively, some subtle tissues can be improved effectually, and some weak sections have not been smoothed by the novel method. 
Volume: 14
Issue: 3
Page: 1203-1212
Publish at: 2016-09-01
1 2 3
...
1826 1827
Show 1462 of 1827

Discover Our Library

Embark on a journey through our expansive collection of articles and let curiosity lead your path to innovation.

Explore Now
Library 3D Ilustration