Symposium of Future Telecommunication and Technologies (SOFTT)

Design Bus Occupant Detection and Counting System Based on IoT

2019-01-25T11:09:33+07:00

Abstractâ€”Automation transportation can be a solution to a public bus transportation in Indonesia, because seeing the social phenomenon that bus management suspects dishonesty from the operation of the bus. Suspicion in the form of payment income that is not in accordance with the number of passengers of the bus. From this case, bus management need a system to check the correct number of passengers, but still consider the passenger comfort. To solve this case a concept of automated system that count and detect the bus occupant based on Internet of Thigs (IoT) was created. This system would facilitate, ensure information accuracy and provide the accuracy of data for the bus manager so that it can minimized cheating on bus operational. This can also increase the profits of the bus companies and will raise automation in the field of public transportation in Indonesia. Keywordsâ€”Occupant Detection, System Autiomation, IoT

Recent Updates on Zigzag Decodable Codes-based Raptor Coding for Active Safety Transportation (ZEBRA-CODES)

2019-01-25T16:37:38+07:00

Abstractâ€”This paper reports the recent updates on Zigzag decodable codes-based Raptor coding for active safety transportation (ZEBRACODES) for the ï¬rst year of total three years. This project is targeting: (1) a decoding schemes ZEBRA-CODES, (2) optimal degree distribution for Raptor coding for transportation, and (3) New modulation scheme suitable for multiple access providing different level of priorities. Index Termsâ€”Coded random access, Zigzag decodable codes, stopping sets, active safety transportations.

Multiple Access Scheme with Priority for Active Safety Transportations

2019-01-25T16:38:12+07:00

Abstractâ€”We propose a new multiple access technique serving multiple groups of useres with different priorities. The proposed system is consisting of: (a) top emergency, (b) emergency, and (c) normal groups. We give priority based on different accessible time-slot to each group. We expect that the results are useful for further development in future wireless massive communication systems. Index Termsâ€”Internet-of-things, Multiple Access, Coded Random Access, Active Safety Transportations

Dual-band C-band and X-band Array Antenna Microstrip for SAR on UAV

2019-01-25T11:12:32+07:00

Abstractâ€”Imaging RADAR has become important nowadays, collecting more data in anytime and any whether condition has to be challenged. One of RADAR is SAR (Synthetic Aperture Radar) which process echo from reflection wave that has been transmitted. Unmanned Aerial Vehicle (UAV) is the one of airborne which can support SAR to transmit electromagnetic waves from the sky. In this research, microstrip antenna array for dual-band on C-band (5,8 GHz) and X-band (9,65 GHz) has designed. Using both of frequency at the same time, SAR will be obtained more data and more specific. The result of simulation has given the best result from the antenna microstrip array 1x8 that achieved the highest gain in both of frequency which is 9,11 dBi on 5,8 GHz and 5,37 dBi on 9,65 GHz.

Index Termsâ€”Array Antenna, Dual-band Antenna, Microstrip Antenna, SAR

Recent Updates on Prevention and Recovery Networks for Indonesia Natural Disasters based on the Internet-of-Things (Patriot-Net)

2019-01-25T16:40:50+07:00

Abstractâ€”This paper reports the recent updates of Prevention and Recovery Networks for Indonesia Natural Disasters based on the Internetof-Things (PATRIOT-Net) for the ï¬rst year of total three years. The PATRIOT-Net project is targeting on 7 outcomes: (i) optimal high dense internet of things (IoT) networks, (ii) network coding algorithm for heterogeneous sensors and applications, and (iii) optimal routing algorithm, (iv) new IoT devices, (v) mobile cognitive radio base station (MCRBS), (vi) apps, (vii) monitoring room. Index Termsâ€”Internet-of-things, network coding, mobile cognitive radio base station, 2G, 3G, 4G, 5G.

Web and Mobile Application for Disaster Prevention in Padang City

2019-01-25T16:43:17+07:00

Abstractâ€”We propose web and mobile application to prevent disaster (warning system) in Padang City. Internet-of-Things (IoT) sensor upload sensing data to the server. Sensing data from the server is sent to mobile application to show information about four disasters situation such as earthquake, tsunami, landslide, and ï¬‚ood. Application could warn user whenever disaster occur indicated by blinking mark and sound. We expect both of web and mobile application could be useful for disaster prevention in other cities. Index Termsâ€”Disaster, prevention, Internet-of-Things, Application.

Embedded System Design Requirement For The Natural Disaster Early Warning System

2019-01-25T16:43:43+07:00

Abstractâ€”Indonesia is the country with the most volcanoes, meeting many very active continental faults. In addition, the condition of the forest which began to be much deforested, much river siltation and environmental destruction. By looking at such conditions it is appropriate for Indonesia to have an automatic early warning system and can be monitored remotely. One such approach is the application of an integrated Embedded System with a reliable communication system. In the embedded system that is being made, as a processor using the ATMega 328 microcontroller. ATM 328 is an output microcontroller from Atmel which has a RISC architecture where each data execution process is faster than the CISC time operating system and reactive computing implemented in a program structure (using C programming language) in a Main Loop with added functions there are several functions for processing existing raw data sensors, then setting the sensor data format to be ready to be sent to the center control through communication modules. To find out the performance of the system that is made in accordance with the application of natural disaster early warning a trial test or measurement of the main parameter is needed.

Index Termsâ€”Embedded System, design, requirement

Sensor System Design For Tsunami Early Warning System

2019-01-25T16:44:10+07:00

Abstractâ€”Tsunami Early Warning System needs sensor system to detect symptoms of tsunami such as the earthquake, eruption of volcanoes in the sea, as well as the sudden reflux of sea water. This project focus on the sensor system to sensing of tsunami symptoms parameters and transfer the data to IoT Platform. The parameters are earthquake and sea wave condition such as waveform slope, sea water level.

Index Termsâ€” tsunami, sensor system

Header Detection of 5G Mobile Base Station for Wireless Disaster Recovery Networks

2019-01-25T16:44:46+07:00

Abstractâ€”We consider a 5G mobile base station supporting natural disaster recovery networks, where multiple waveform detection is required. In this paper, we propose a practical signal detection using header of the transmitted signal called header detection. The proposed header detection is used in recovery network on disaster area where needs high accuracy signal detection with simple algorithm. We perform a header detection using the cross correlation and capture effect algorithm. This paper considers header detection using Hadamard codes with size of 16Ã—16 which is usually use as header codes on data packet because their length sizes are extendable and guarantee low computational complexity. We found that the capture effect algorithmcanimprovetheaccuracyofheaderdetectionevaluated by mean square error (MSE). Index Termsâ€”Header detection; Heterogeneous waveforms; Disaster recovery networks;

Emotion Classification Based On Eeg Signal Using Support Vector Machine And Independent Component Analysis

2019-01-25T11:13:25+07:00

Abstractâ€”In making a decision, choose the results of the decision. For example when happy, the show will be fine, on the contrary if sad it will provide bad convenience. Emotions include physiological, namely electroencephalographic (EEG) signals from the brain. EEG recording Appears when electrical problems occur in the brain[1]. EEG signals come from DEAP research: The database for Emotion Analysis uses physiological signals from arousal and valence levels and is processed by Independent Component Analysis (ICA). With ICA, existing data will be processed and obtained new data in the form of a matrix. The results of the matrix will be conveyed by Support Vector Machine (SVM) to produce comfortable conditions when happy, relaxed, nervous, and sad. Thus, the results obtained by the data know what percentage of the index is useful when happy, relax, nervous, and sad.

Index Terms--EEG, ICA, SVM

CRLB For DRSS-Based Factor Graph Of Wireless Geolocation

2019-01-25T16:53:20+07:00

Abstractâ€”This paper derives a Crammer Rao lower bound (CRLB) for wireless geolocation technique based on differential received signal strength (DRSS) measured parameter and factor graph framework. Then, the CRLB is derived by modifying the Fisher information matrix (FIM). In particular, the derivation inside the FIM focuses on Jacobian matrix having relationship between the target coordinate and the measured DRSS. Finally, the simulation results show that the derived CRLB is excellence with the lowest root mean square error (RMSE) curve. Index Termsâ€”CRLB, DRSS, FIM, Factor Graph, Geolocation

Deep Learning For Short Packet Transmissions Over Flat Fading Channels

2019-01-25T16:53:45+07:00

Abstractâ€”This paper investigates the performances of a deepneural-networks scheme for decoding polar coded short packets that are transmitted over frequency-ï¬‚at fading. Computer simulation results conï¬rm that the proposed technique achieves the coding gain with learning epoch larger than 212. Index Termsâ€”polar codes, packet combining, deep learning, short packet, lossy forwarding

Design Of An Ultra Wideband Antenna For Noncontact Respiratory Monitoring

2019-01-25T11:14:10+07:00

Abstractâ€”Non contact respiratory monitoring is the development of wireless communication systems in telemedicine. A key component of non-contact respiratory monitoring is an antenna. The antenna must be lightweight so that it is compatible with other supporting system components. The paper presents the design of an optimized radiating element that satisfies challenging requirements: ultra wide-band (UWB), directive radiation pattern, small dimensions (in order to reduce its invasiveness) and low cost for realization.

Index Termsâ€”Non contact respiratory monitoring;UWB; antenna

Minangkabau And Sunda Tribes Detection Based On Lip Print Pattern Using Discrete Cosine Transform (DCT) And Learning Vector Quantization (LVQ)

2019-01-25T11:14:44+07:00

Abstractâ€”Forensic science is one of the medical sciences to be able to express the identity of an individual from gender to race and ethnicity. In general, forensic science can be interpreted as an application or use of certain knowledge in the interests of law enforcement and justice. In an individual there is a unique pattern that is different from other individuals, even though the specifics will be different for each individual. In general, fingerprints are used but have limitations, namely the lack of resistance to the fingers so that another alternative is lip prints. Lips are an alternative way to get data from individuals. On the lips there are unique patterns that each individual is different, namely the description of sulci on the mucosa of the upper lip and lower lip, as well as fingerprints. Lipstick can be used so that it can help forensic science in solving existing cases. This Paper to identify the Minangkabau and Sundanese tribes so as to minimize the scope of individual searches in the application of forensic science. The method used is the registration of lip print images By using feature extraction method of Discrete Cosine Transform (DCT) and and for classification using Learning Vector Quantization (LVQ).

EXIT Analysis For 5G NR QC-LDPC Codes Based On Base Graphs 1 And 2

2019-01-25T16:45:12+07:00

Abstractâ€”Quasi-Cyclic (QC) Low Density Parity Check (LDPC) codes constructed based on Base Graphs (BG) 1 and 2 have been standardized as channel coding schemes for the ï¬fth telecommunication generation (5G) New Radio (NR). This paper analyses the behaviour of QC-LDPC codes of 5G NR based on BG-1 and BG-2 to evaluate BG selection for practical applications. In this paper, we consider a single carrier transmission under additive white Gaussian noise (AWGN) channels. We use Extrinsic Information Transfer (EXIT) chart to evaluate and predict the performances as well as the behaviour of the QCLDPC codes based on BG-1 and BG-2. Index Termsâ€”QC-LDPC codes, EXIT Analysis, Channel coding, 5G NR.

Outage Performances Of 5G Channel Model Considering Temperature Effects At 28 Ghz

2019-01-25T16:45:50+07:00

Abstractâ€”The 5th telecommunication generation (5G) is expected to be deployed worldwide in 2020 including Indonesia for future better services using 1-100 GHz band, which is sensitive to the environments. For optimal 5G deployment in Indonesia, this paper studies 5G channel model based on software simulation by considering the inï¬‚uence of temperature, where Telkom University, Bandung, Indonesia, is chosen as a representative location for the 5G channel model. We consider a frequency of 28 GHz, which is one of the golden frequencies for pratical 5G applications. From the model, we obtain Power Delay Proï¬le (PDP) representing the 5G Telkom University channel model. Based on the PDP, this paper calculates the outage performances to predict the 5G performance of Indonesia regardless any 5G technologies used, since the outage performance is based on the theoretical Shannon channel capacity limit for a probability of error asymptotically small (close to zero). Index Termsâ€”5G Channel Model, Temperature, Power Delay Proï¬le, Outage Probability, Capacity

Outage Performances Of 5G Channel Model Considering Humidity Effects

2019-01-25T16:46:12+07:00

Abstractâ€”The ï¬fth telecommunication generation (5G) is predicted to be deployed in 2020, where high frequencies of 1â€“100 GHz expected to be utilized are causing high attenuations. We consider communications under a 5G channel model for operating frequency of 3.3 GHz with bandwidth of 60 MHz based on 5G New Radio (NR) standard. In this paper we propose a 5G channel model based on Telkom University realï¬eldenvironmentsunderhumidityeffectsderivedusingcomputer simulations for 5G NR implementation in Indonesia. The channel model is represented by the Power Delay Proï¬le (PDP), which is conï¬rmed by outage performance of Telkom University 5G channel model. Index Termsâ€”Power Delay Proï¬le, Outage Performance

Study On Ofdm Numerology 4 Of 5g New Radio (Nr) Under Indonesia 5g Channel Model

2019-01-25T16:46:35+07:00

Abstractâ€”With data transfer speed expected around 20 Gbps, the ï¬fth generation of cellular communication (5G) New Radio (NR) is estimated to serve services interconnected to all heterogeneous wireless networks. However, several optimal parameters are not yet known, such as Fast Fourier Transform (FFT) size, cyclic preï¬x (CP) length, block length, coding rate, and bandwidth that suitable for Indonesia. This paper proposes study on 5G NR technology performances using Orthogonal Frequency Division Multiplexing (OFDM) Numerology 4 under Indonesia 5GChannelModeltoservevariousapplicationsof5Gtechnology infuture.Thispaperinvestigatesthecharacteristicsofbandwidth distribution and parameters for the application of 5G NR for OFDM Numerology 4. The results show that with parameters of OFDM Numerology 4, the 5G channel model has 13 paths which the outage performances are derived for coding rates R = 1/2 and R = 1. Index Termsâ€”Cellular Communication, 5G New Radio (NR), OFDM Numerology, Indonesia 5G Channel Model

Recent Updates of Preparing Optimal 5G Indonesia Networks (5G-POINT

2019-01-25T16:47:11+07:00

Abstractâ€”This paper reports the recent updates of Preparing Optimal 5G Indonesia Networks (5G-POINT) for the ï¬rst year of total two years. This project is targeting: (i) the Indonesia ï¬fth generation (5G) of telecommunication channel model, (ii) Theoretical 5G outage performances of Indonesia, and (iii) Framework of 5G channel measurement for any locations in Indonesia, and (iv) Dissemination in terms of books, whitepapers and tutorials. Index Termsâ€”5G Channel Model, channel measurement, power delay proï¬le, outage probability.

A Study of Familiarity Effects Classification in Human EEG Signal Using Hjorth Descriptor

2019-01-25T11:18:56+07:00

Abstractâ€”There are a lot of researches related to human EEG signal that has been done before, but there are only a few of research related to familiarity effects in human EEG signal. Hence, this paper will classify human EEG signal while feeling familiar. This paper is using secondary data taken from DEAP: A Database for Emotion Analysis using Physiological Signals. The feature extraction method used is Hjorth Descriptor. The feature classification chosen is Coarse K-Nearest Neighbor, because the accuracy is 5.55% higher than the average of all KNN methods.

Keywordâ€”EEG, Familiarity, Hjorth Descriptor, Coarse K-NN, K-Nearest Neighbor.

Tonic Clonic Seizure Classification Based on EEG Signal Using Artificial Neural Network Method

2019-01-25T11:23:18+07:00

Abstractâ€”An instrument to record the activity of brainwave in specific time called Electroencephalography (EEG). EEG signal can be used to analyze the epilepsy disease. One of a signal that appears when a seizure happens called Tonic Clonic Seizure (TCSZ) signal. Brainwave of seizure patient has a low frequency with a tighter pattern than brainwave of normal people. The purpose of this research is to classify between the tonic clonic seizure signal and normal EEG signal using Artificial Neural Network (ANN) Backpropagation method. At first, the features of signals will be extracted by using Mel Frequency Cepstral Coefficients (MFCC). The output of MFCC will be the input for ANN Backpropagation classifier. The result of this research has reached 100% of accuracy value with 13-MFCCs features and 25-MFCCs features.

Keywordsâ€”TCSZ, ANN Backpropagation, MFCC, Epilepsy

Design of a MIMO Antenna for Broadband 5G Using Stepped Cut at Four Corners Method

2019-01-25T11:24:07+07:00

Abstractâ€”A MIMO antenna using SCFC Method for the Broadband 5G communication is proposed in this paper. The proposed antenna consists of two antenna, it operating at 14000- 16000 MHz. The antenna designed in this letter are different from any design of 5G antennas, the antenna of this paper is used based on operating frequency and bandwidth requirement , it can be applied for the IOT technology needed. According to the simulation results, a total bandwidth of the antenna is 2.348 MHz, and the VSWR is 1.26 with Gain 1.23 dB over the band- frequency of 1400016000 MHz, it will met the needs of IOT for future 5G applications. Index Termsâ€”Stepped Cut at Four Corners, 5G operation, MIMO antenna

Classification NREM in EEG Signal for Detection Depth of Sleep Using HJORTH Descripto

2019-01-25T14:09:03+07:00

Abstractâ€” Sleep is one of rest which is marked by decrease in physical activity, awareness, and decrease in respond for external stimulation. Sleep is important in returning body energy, maintain body resilience even cognitive function in the body. There is two kind of sleep that is Rapid Eye Movement (REM) and Non Rapid Eye Movement (NREM) that must be experienced by someone for reach depth of sleep. If one of that not achieved then it can disturbance in the body. Detection of NREM and REM can be done with analyze EEG signal. In this research has been successfully classify NREM and REM wave based on EEG signal for delta wave. A time series method analysis that is HJORTH Descriptor used to get signal feature of that. HJORTH activity, mobility, and complexity showing many value for each category. From simulation has been reach 70,9% accuracy using Fine Gaussian Support Vector Machine.
Keywordsâ€”NREM, REM, HJORTH Descriptor, SVM

A Study of Arousal Classification Based on EEG Signal with Support Vector Machine Method

2019-01-25T14:09:33+07:00

Abstract---The development of Brain Computer Interface technology nowadays has spread out in a case of classifying emotions based on brain signal (EEG) in human, which in this work using a set of secondary data from DEAP. One of the emotion parameters being focused on here is arousal with the range from low (uninterested) to high (excited). This study is applying Principal Component Analysis as the feature extraction. Not only that, feature extraction also being done statistically. As for feature classification is using Support Vector Machine with the maximum accuracy that only able to reach 59,4% which still needs improvements in the system for future works. Keywords---EEG, PCA, SVM, Emotion Classification

Array Mikrostrip Antenna With Triangular Patch Using Dgs Technique For Mimo 2x2 At Frequency Of 15 Ghz

2019-01-25T14:10:22+07:00

Abstractâ€”This paper proposes MIMO antenna for 5G communications. The antenna proposed in this paper is mimo 2x2 antenna using triangular patch. The antenna works at a frequency of 15 GHz. This paper use the DGS method with rhombus shape. The material used is Duroid Roger 5880 with â†‹r = 2,2 and a material of thickness 1,575 mm. Based on the simulation results, this antenna have bandwidth greater than 1 GHz over than frequency 14,5-15,5 GHz, and gain of 9,8 dB with radiation patterns is uni directional.

Keywordsâ€”MIMO antenna, triangular patch, DGS method.

The Utilization GPS Radio Occultation Data to Improve Numerical Weather Prediction Skill through Assimilation Data Procedure Using WRF3DVAR Technique over Jakarta Region

2019-01-25T14:11:05+07:00

Abstractâ€”The model of Weather Research and ForecastingAdvanced Research WRF (WRF-ARW) is one of Numerical Weather Prediction (NWP) model which often used to study and to predict weather phenomenon in the atmosphere. Initial condition and boundary condition are two essential elements which needed by WRF model in order to produce forecast. Initial condition is a part of WRF that needs to be corrected in order to make the prediction more accurate. Various methods have been developed to improve the initial condition one of them through data assimilation. There are several methods of assimilation data process which combines NWP products with information from different types of observation, one of them is Three Dimensional Variational (3D-VAR). The purpose of this research is to analyze and compare the accuracy of Weather Research Forecasting (WRF) prediction before and after assimilation the Global Positioning System Radio Occultation (GPS RO) Refractivity data, where the GPS data will be assimilated into the WRF-ARW model by 3D-VAR technique to simulate rain event in Jakarta area on 14 until 16 February 2018. Verification technique to quantify the accuracy of the assimilation model was conducted towards 24 hours accumulated rainfall. The result of this research shows that by applying the data assimilation procedure of the GPS RO Refractivity which goes into WRF-ARW model can increase the accuracy predictions level of heavy rainfall phenomenon which is occurred at that time where able to predict the occurrence for the first category correctly through the percentage of average POD reaching 66% with a prediction error rate of average rainfall (POFD) of 26.1%. Furthermore, for the light rain category, on average only around 59.2% of events can be predicted correctly and with an average percentage of 12.5% prediction errors. Keywordsâ€”Weather Prediction, WRF-ARW, GPS RO Refractivity
I

Classiï¬cation Tree Performance Analysis on ICA-based Functional Near-infrared Spectroscopy Signals

2019-01-25T16:31:40+07:00

Abstractâ€”This paper proposes a method to classify between the clean and the contaminated signal by motion artifact (MA) signal on functional near-infrared spectroscopy (fNIRS) signals, by extracting the signal features based on independent component analysis(ICA)andstatisticalmodelsusingclassiï¬cationtreeasthe classiï¬er. The extracted features such as kurtosis, skewness, mean, variance, standard deviation, interquartile range, and weight vector are used in a classiï¬cation tree as the prediction model for class classiï¬cation. The result of this paper is to acknowledge the performance of classiï¬cation tree to classify the fNIRS signals, which results in 88.9% accuracy, 81% sensitivity, 100% speciï¬city, and 0.83 value of area under convergence (AUC). Index Termsâ€”fNIRS, ICA, feature extraction, classiï¬cation tree

Performance Analysis on Fine-tuned Region-based CNN for Object Recognition

2019-01-25T16:32:22+07:00

Abstractâ€”In this time, machine learning technology has developed rapidly, especially on deep learning architecture where many models have been created and already has a good result. However, to create a good system it took a long process such as designing a model architecture, creating a vast number of dataset and test the model many times to obtain the best performance. It is practically hard to create such a system. Therefore ï¬ne-tuning is a fascinating thing to discuss, by merely taking a model that has been known having a good result and conï¬gure the model to suit our needs, ï¬ne-tuning becomes more popular method than to create a model from scratch. In this experiment, we tried to ï¬ne-tune the R-CNN model where the pre-trained model used the Residual network architecture. Our best is at 90%. Index Termsâ€”Deep learning, Neural network, Fine-tuning

Analysis of Transfer Learning on Faster R-CNN for Vehicle Detection

2019-01-25T16:33:15+07:00

Abstractâ€”Computer vision is one of the favorite research topics recently, especially for object detection task. Faster Regionbased Convolutional Neural Network (R-CNN) is a state-ofthe-art object detection algorithm. This method has an excellent performance inï¬‚uenced by several parameters such as the number of convolution layers, epoch, padding scheme, network initialization, etc. In this paper, we perform an analysis of the impact of transfer learning using pre-trained AlexNet on Faster R-CNN for vehicle detection. Transfer learning method enables us to use a small amount of training data and training time to achieve good performance. Based on the experimental results, the performance of transfer learning has signiï¬cant improvement by 15.9% compared to the full-training model with mAP of 73.1% at 10th epoch. Keywordsâ€“Convolutional neural network; Transfer learning; Object detection; Vehicle detection

The Effect of Proactive Flow on Network Downtime in Software Defined Network Using Opendaylight Controller

2019-01-25T16:33:44+07:00

Abstractâ€”Computer vision is one of the favorite research topics recently, This paper will analyze the effect of proactive flow on network downtime when there is a link failure. The network is designed using a data center network model and OpenDaylight as SDN controller. Proactive flow has global view of the network before the first packet arrives. Refers to that, the changes of the network when sending packets will have an impact on the continuity of packets transmission. The result shows that downtime due to the link failure during packets transmission is 3.5 seconds.

Keywordsâ€” Software Defined Network, OpenFlow, OpenDaylight, Proactive Flow

Performance Analysis of Metric Threshold in SURF for Object Tracking

2019-01-25T16:35:01+07:00

Abstractâ€”Object tracking has a lot of progress every year and gives something new, so the trackers and method itself is getting better. Many researchers are engaged in one of the ï¬elds of computer vision to provide good beneï¬ts for human life in the ï¬eld of Internet of Things. Feature extraction is needed in object tracking processes. One of the scale and rotationinvariant local feature extraction methods, namely Speeded-Up Robust Feature (SURF). In implementing it on object tracking, SURF will extract features from two frames and match them. Then, the Random Sample Consensus (RANSAC) as a rejection correspondence using an inlier on the features obtained and then performs the estimation transformation which is applied to the input match the reference image. In this paper, we analyze one of the parameters of SURF, namely Metric Threshold as a parameter that determines the strongest feature threshold. From the evaluation results, it was found that the default parameters gave non-optimal results.

Face Recognition Using the Direct GLCM and K-NN Methods

2019-01-25T16:35:35+07:00

Abstractâ€”This paper introduces a new face recognition method based on the gray-level co-occurrence matrix (GLCM). This method directly uses GLCM by converting a matrix into a vector that can be used as a feature vector for the classiï¬cation process, this method is called direct GLCM. The classiï¬cation process used is K-Nearest Neighbor (K-NN), in which the classiï¬cation process compares the features contained in K-NN namely Euclidean distance, Cityblock, Chebychev, and Minkowski. The results show that using direct GLCM as a feature vector in the recognition process using the K-NN classiï¬cation with the Cityblock feature produces an accuracy of 84.29%, FAR 6.67% and FRR 9.05%. Index Termsâ€”Face recognition, Gray-Level Co-occurrence Matrix, K-Nearest Neighbor, Cityblock

Generalized Non-Specific Seizure Based on EEG Signal Using Artificial Neural Network Method

2019-01-25T16:36:01+07:00

Abstractâ€” Epilepsy is a seizure that occurs in the human brain. To find out, this research is to detect one of the signals in epilepsy patients, Generalized Non-Specific Seizure (GNSZ) recorded by using Electroencephalography (EEG). The dataset is a GNSZ signal taken from Temple University EEG Corpus. In this study, Hjorth Descriptor Method was used a feature extraction to process signals in time domain, where the output of this method is represented using three parameters, activity, mobility, complexity, and Artificial Neural Network (ANN) as a classification. In this study, the results of feature extraction from the GNSZ signal on the recording of EEG signals are compared with the characteristics of the normal signal. The results of this research has got 95,83% accuracy by using activity and complexity parameters. Other results obtained are GNSZ signals recognized as normal or vice versa.
Keywordsâ€”Epilepsy, GNSZ, Hjorth Descriptor, ANN

Impact Analysis of Type-2 Fuzzy Logic for Weighted Multiple Instance Learning Performance on Motion Blur and Low-Resolution Attributes

2019-01-25T16:36:25+07:00

Abstractâ€”Type-2 fuzzy logic is an extension of type-1 fuzzy logic. Type-2 fuzzy logic can model an uncertainty better than type-1 fuzzy logic. Because it can model uncertainty well, type-2 fuzzy logic is very well used for decision making. In weighted multiple instance learning (WMIL), the tracker has not been able to decide whether the tracker has failed or not. In this paper, we analyze the inï¬‚uence of Type-2 fuzzy logic for WMIL performance on the motion blur and low-resolution attributes. Based on the experimental results, WMILâ€™s performance blur increasedby0.0725whileinthelow-resolutionattribute,WMILâ€™s performance decreased by -0.0045 when compared to WMIL without fuzzy logic type-2. The parameters used to analyze performance are success plot and precision plot. Index Termsâ€”Type-2 fuzzy logic, weighted multiple instance learning, motion blur, low-resolution, object tracking.

Focal Non-Specific Seizure Classification Based on EEG Signal Using Artificial Neural Network Method

2019-01-25T16:36:45+07:00

Abstractâ€”In EEG signals Epilepsy has various signals. One of them is Focal Non-Specific Seizure which has features that are different from other signals. This paper will detect the presence of a signal surge in the Focal Non-Specific Seizure. This detection is based on brain activity using Epilepsy EEG which will be compared with normal people's training data. Focal Non-Specific Seizure Signals will go through the preprocessing process. Then the preprocessing results are feature extraction with Independent Component Analysis (ICA) which will then be classified with Artificial Neural Networks (ANN). From this study the greatest accuracy of one of the characteristics is weight vactor of 90%.

Keywordsâ€” Focal Non-Specific Seizure, Independent Component Analysis (ICA), Artificial Neural Network (ANN)