Bayu Kanigoro

Weather forecasting using deep learning techniques

Weather forecasting has gained attention many researchers from various research communities due to its effect to the global human life. The emerging deep learning techniques in the last decade coupled with the wide availability of massive weather observation data and the advent of information and computer technology have motivated many researches to explore hidden hierarchical pattern in the large volume of weather dataset for weather forecasting. This study investigates deep learning techniques for weather forecasting. In particular, this study will compare prediction performance of Recurrence Neural Network (RNN), Conditional Restricted Boltzmann Machine (CRBM), and Convolutional Network (CN) models. Those models are tested using weather dataset provided by BMKG (Indonesian Agency for Meteorology, Climatology, and Geophysics) which are collected from a number of weather stations in Aceh area from 1973 to 2009 and El-Nino Southern Oscilation (ENSO) data set provided by International Institution such as National Weather Service Center for Environmental Prediction Climate (NOAA). Forecasting accuracy of each model is evaluated using Frobenius norm. The result of this study expected to contribute to weather forecasting for wide application domains including flight navigation to agriculture and tourism.

Recurrent gradient descent adaptive learning rate and momentum neural network for rainfall forecasting

This paper presents a recurrent optimized heuristic Artificial Neural Network (ANN) uses Recurrent Elman Gradient Descent Adaptive Learning Rate and Momentum method with various parameter values. The values is approached by using El-Nino Southern Oscillation (ENSO) variable, which are Southern Oscillation Index (SOI), Wind, Outgoing Long Wave Radiation (OLR), and Sea Surface Temperature (SST) to forecast regional monthly rainfall in Bongan Bali. The data sets used in this paper are separated by two groups. First group of data has training data and testing data which is 75% and 25% respectively. The latter group of data consists of 50% training data and 50% testing data. First data group produces the maximum R2 leap 74.6% and the latter produces the maximum R2 leap 49.8%.

Adaptive Neuro Fuzzy Inference System for Diagnosing Dengue Hemorrhagic Fever

Death of effect disease Dengue Hemorrhagic Fever (DHF) in Jakarta is high. To diagnosis patient of DHF, also render help immediately, and prevent death. Expert system can be used for the diagnosis of DHF. This research is conducted by the model system of Neuro-Fuzzy. One of them is ANFIS for conducting the diagnosis and medical procedure of DHF. Sample of research consist patient data suffering from DHF and dengue fever (DF), 32 cases of DHF and 32 cases of DF. Examination of model is conducted by some phase with data before and after verification. The examination of the model consists of amount and type of membership function. The conclusion of this research is ANFIS can be used selectively for conduct diagnosis of disease of DHF and can be used better if fulfilled minimum condition according to regulation which have been known such as haemorrhage manifestation and finding of fever. Model use the type of membership function of Gaussian yield accuracy 86,67 %, type of trapezoid yield accuracy 40 %, type of bells yield accuracy 40 % and type of triangular yield accuracy 40 %.

STEGANOGRAPHY APPLICATION PROGRAM USING THE ID3V2 IN THE MP3 AUDIO FILE ON MOBILE PHONE

The MPEG Layer-III File or MP3 can contain the ID3v2 tag which able to store information not only the song and singer’s name but also song lyric and alblum cover figure. The steganography technique on audio generally only make use of the main body of the audio file, but has not ever use the part of the ID3v2 to hide an message. This study an steganography technique which make use of this ID3v2 tag space where message is encrypted using McEliece cryptosystem method which applying the public key in the form of matrix then inserted into the MP3 file applying Before All Frames (BAF) method, whereas the public key is inserted into album cover figure which is stored in the ID3v2 tag.

WEB BASED CONFERENCE SYSTEM FOR INTELLIGENCE TELEPRESENCE ROBOT: A FRAMEWORK

Once an individual has access to the Internet, there is a wide variety of different methods of communication and information exchange over the network, one of them is telepresence robot. This study presents a web framework for web conference system of intelligent telepresence robot. The robot is controlled using web conference system from Google App Engine, so the manager/supervisor at office/industry can direct the robot to the intended person to start a discussion/inspection. We build a web application for controlling the multiclient telepresence robot. Experimental result shows the ability of robot to be controlled remotely and smoothly using our framework.

Parental control application on Android platform

Purpose The purpose of this study is to build the android-based applications for parent to supervise, control and restrict the children in using smartphone and tablet. Design/methodology/approach The method used is the method of analysis, i.e. collect the necessary data. The data is collected from journals, articles, books, questionnaires, and the similar application analysis. Findings Evaluation of existing applications made to perform an evaluation of the user that the GPS tracking feature of the Parental Control Application facilitate searching children location by pointing the spot where they use Android device , the application history feature of The Parental Control facilitate finding out what application that accessed by children through their Android-based application and the time limit feature of the Parental Control Application facilitate restriction on time that children spend on using Android-based so that they will not be gadget addict. Originality/value The authors report of the development ...

Ordinary Kriging and Spatial Autocorrelation Identification to Predict Peak Ground Acceleration in Banda Aceh City, Indonesia

Peak ground acceleration (PGA) is a measure of earthquake acceleration in the ground. The prediction information about PGA is important to minimize the effect of earthquake. The method for prediction is Ordinary Kriging. It is geostatistic method used to predict data in certain locations which have autocorrelation. The sample data used in this research are PGA in Meuraxa, Banda Aceh 2006. The steps of research methodology consist of autocorrelations identify by Moran’s I and LISA, build semivariograms, and prediction by Ordinary Kriging. The results is Ordinary Kriging can be applied to predict PGA. It was shown by evaluate of mean and MSE value. According to mean value of three prediction, all models (Gaussian, Spherical, and Exponential) have mean 0,3534; 0,3584; and 0,3555 which approaches the actual PGA mean 0.34. According to MSE value, it can be seen that all models have small MSE or relatively closed to zero.

Earthquake Ground Motion Attenuation Modeling Using Levenberg-Marquardt and Brute-Force Method

In this paper, we discuss the results of research on the optimization modeling of ground motion attenuation in the subduction zone of the model Youngs et al. [1] using two methods: the Levenberg-Marquard and Bruce-Force method. This modeling is particularly important in the case of seismicity. Given that it takes a good model for predicting the strength of earthquakes in order to reduce the risk of the impact of natural disasters. Two major contributions of this study are ground motion attenuation model specific to the subduction zone that has been optimized with the Levenberg-Marquard method and Bruce-Force uses a model Youngs et al. [1] and a proof that the Levenberg-Marquard method for optimization model is better than Bruce-Force method. The Levenberg-Marquardt method has been proven to provide more accurate results on the modeling of ground motion attenuation which is indicated by a very small deviation between the values of PGA predictable results with the PGA actual values.

Ball Distance Estimation and Tracking System of Humanoid Soccer Robot

Modern Humanoid Soccer Robots in uncontrolled environments need to be based on vision and versatile. This paper propose a method for object measurement and ball tracking method using Kalman Filter for Humanoid Soccer, because the ability to accurately track a ball is one of the important features for processing high-definition image. A color-based object detection is used for detecting a ball while PID controller is used for controlling pan tilt camera system. We also modify the robots controller CM-510 in order able to communicate efficiently using main controller. The proposed method is able to determine and estimate the position of a ball and kick the ball correctly with the success percentage greater than 90%. We evaluate and present the performance of the system.