Pub Date : 2018-10-01DOI: 10.1109/TENCON.2018.8650138
Otgontsetseg Sukhbaatar, K. Ogata, T. Usagawa
Learning management systems generate a large amount of data, where knowledge discovery is possible using data mining techniques. We proposed simple prediction scheme using decision tree analysis for purpose of classification to identify dropout prone students in the middle of the semester based on previous year’s course characteristics for that course. The data included 717 students’ online activities in compulsory, sophomore level course with blended learning styles, 79% of the actual dropout students were predicted correctly.
{"title":"Mining Educational Data to Predict Academic Dropouts: a Case Study in Blended Learning Course","authors":"Otgontsetseg Sukhbaatar, K. Ogata, T. Usagawa","doi":"10.1109/TENCON.2018.8650138","DOIUrl":"https://doi.org/10.1109/TENCON.2018.8650138","url":null,"abstract":"Learning management systems generate a large amount of data, where knowledge discovery is possible using data mining techniques. We proposed simple prediction scheme using decision tree analysis for purpose of classification to identify dropout prone students in the middle of the semester based on previous year’s course characteristics for that course. The data included 717 students’ online activities in compulsory, sophomore level course with blended learning styles, 79% of the actual dropout students were predicted correctly.","PeriodicalId":132900,"journal":{"name":"TENCON 2018 - 2018 IEEE Region 10 Conference","volume":"95 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124755660","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/TENCON.2018.8650263
S. Phaiboon, P. Phokharatkul
This paper presents a semi deterministic path loss model at frequency of 28 GHz for fifth generation (5G). We modified Xia model which was used for frequency of 0.9 GHz to 2.0 GHz with a correction factor in order to use for higher frequency. Two important parameters, average building height and distance of mobile from last roof top were considered while measured data in the dense urban environment around New York University's (NYU) Manhattan campus were used for modeling and verification. Comparisons between the modified and conventional models provide all NLOS propagation including transverse, lateral, and staircase.
{"title":"Extended Xia Semi Deterministic Model for a Frequency of 28 GHz","authors":"S. Phaiboon, P. Phokharatkul","doi":"10.1109/TENCON.2018.8650263","DOIUrl":"https://doi.org/10.1109/TENCON.2018.8650263","url":null,"abstract":"This paper presents a semi deterministic path loss model at frequency of 28 GHz for fifth generation (5G). We modified Xia model which was used for frequency of 0.9 GHz to 2.0 GHz with a correction factor in order to use for higher frequency. Two important parameters, average building height and distance of mobile from last roof top were considered while measured data in the dense urban environment around New York University's (NYU) Manhattan campus were used for modeling and verification. Comparisons between the modified and conventional models provide all NLOS propagation including transverse, lateral, and staircase.","PeriodicalId":132900,"journal":{"name":"TENCON 2018 - 2018 IEEE Region 10 Conference","volume":"69 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129507946","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/TENCON.2018.8650284
Taewoo Kim, Eunju Yang, S. Bae, Chan-Hyun Youn
As the deep learning (DL) has widely been used for application domains such as image classifications, natural language processing, and speech recognition, various software frameworks have been developed. They provide users with efficient programming interfaces for developing the DL applications. The optimization techniques within these frameworks generally are different from each other, which leads to different processing times for even the same applications. However, it is difficult that end users consider performance differences in processing time due to incompatible programming interface among the DL frameworks. These differences might cause redundant efforts and costs for end users to develop and maintain the applications. In this paper, we introduce an integrated deep learning engine (IDLE), a novel interface working on the top of the existing DL frameworks, which provides a convenient, flexible and scalable programming interface developing the DL applications for end users regardless of DL frameworks. Besides, we also propose a novel adaptive task scheduling scheme for training DL applications in a cluster with different GPUs. We implement our platform on the heterogeneous GPU cluster, and the results show that the proposed scheduling algorithm improves cost efficiency processing various DL applications.
{"title":"IDLE: Integrated Deep Learning Engine with Adaptive Task Scheduling on Heterogeneous GPUs","authors":"Taewoo Kim, Eunju Yang, S. Bae, Chan-Hyun Youn","doi":"10.1109/TENCON.2018.8650284","DOIUrl":"https://doi.org/10.1109/TENCON.2018.8650284","url":null,"abstract":"As the deep learning (DL) has widely been used for application domains such as image classifications, natural language processing, and speech recognition, various software frameworks have been developed. They provide users with efficient programming interfaces for developing the DL applications. The optimization techniques within these frameworks generally are different from each other, which leads to different processing times for even the same applications. However, it is difficult that end users consider performance differences in processing time due to incompatible programming interface among the DL frameworks. These differences might cause redundant efforts and costs for end users to develop and maintain the applications. In this paper, we introduce an integrated deep learning engine (IDLE), a novel interface working on the top of the existing DL frameworks, which provides a convenient, flexible and scalable programming interface developing the DL applications for end users regardless of DL frameworks. Besides, we also propose a novel adaptive task scheduling scheme for training DL applications in a cluster with different GPUs. We implement our platform on the heterogeneous GPU cluster, and the results show that the proposed scheduling algorithm improves cost efficiency processing various DL applications.","PeriodicalId":132900,"journal":{"name":"TENCON 2018 - 2018 IEEE Region 10 Conference","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129545298","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/TENCON.2018.8650114
Leandro David Cuayo, Jorelle Kerbee Culla, Jeneson Gualvez, Siward Enoch Padua, R. J. Gallano
This paper presents a methodology to build a wireless SCADA Remote Terminal Unit (RTU) through a microcontroller. Using a microcontroller as a SCADA RTU eliminates the complexity of building current SCADA systems. It minimizes the space it covers and removes long communication lines as it communicates through Wi-Fi. The proposed methodology uses three microcontrollers to support the data acquisition, the server, and the supervisory control branch of the SCADA system. The study proves that the microcontrollers are able to handle real-time monitoring with the voltage, current readings and the states of the circuit breakers, as well as supervisory control by turning the circuit breakers on and off remotely.
{"title":"Development of a Wireless Microcontroller-based SCADA RTU","authors":"Leandro David Cuayo, Jorelle Kerbee Culla, Jeneson Gualvez, Siward Enoch Padua, R. J. Gallano","doi":"10.1109/TENCON.2018.8650114","DOIUrl":"https://doi.org/10.1109/TENCON.2018.8650114","url":null,"abstract":"This paper presents a methodology to build a wireless SCADA Remote Terminal Unit (RTU) through a microcontroller. Using a microcontroller as a SCADA RTU eliminates the complexity of building current SCADA systems. It minimizes the space it covers and removes long communication lines as it communicates through Wi-Fi. The proposed methodology uses three microcontrollers to support the data acquisition, the server, and the supervisory control branch of the SCADA system. The study proves that the microcontrollers are able to handle real-time monitoring with the voltage, current readings and the states of the circuit breakers, as well as supervisory control by turning the circuit breakers on and off remotely.","PeriodicalId":132900,"journal":{"name":"TENCON 2018 - 2018 IEEE Region 10 Conference","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129570033","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/TENCON.2018.8650137
Melannie B. Mendoza, Renehl John C. Alejandro, Franz Nichol P. Cadao, Maecie Claire G. Rayo
This paper describes the process, methods and experimental verification and validation of a flood forecasting system. The main objective of this design project is to develop a predictive flood monitoring system using the Hidden Markov Model that will provide an alternate route once the level of the flood reaches a certain range that is not passable to vehicles. Moreover, it will have a local database that will receive the information that will be sent by the pressure sensor and the rain gauge. The present value of the rain gauge and the historical data of flood and rain will be the basis of the prediction of the road as to whether it will be passable, not passable to light vehicles and not passable to heavy vehicles. It will also have a mobile application for public notifications of the flood level, rain intensity, road condition, and percentage of passability of the road. This system is deemed as a faster method of alerting relevant authorities and the public.
{"title":"Development of Manila Road Predictive Flood Monitoring System Using Hidden Markov Model","authors":"Melannie B. Mendoza, Renehl John C. Alejandro, Franz Nichol P. Cadao, Maecie Claire G. Rayo","doi":"10.1109/TENCON.2018.8650137","DOIUrl":"https://doi.org/10.1109/TENCON.2018.8650137","url":null,"abstract":"This paper describes the process, methods and experimental verification and validation of a flood forecasting system. The main objective of this design project is to develop a predictive flood monitoring system using the Hidden Markov Model that will provide an alternate route once the level of the flood reaches a certain range that is not passable to vehicles. Moreover, it will have a local database that will receive the information that will be sent by the pressure sensor and the rain gauge. The present value of the rain gauge and the historical data of flood and rain will be the basis of the prediction of the road as to whether it will be passable, not passable to light vehicles and not passable to heavy vehicles. It will also have a mobile application for public notifications of the flood level, rain intensity, road condition, and percentage of passability of the road. This system is deemed as a faster method of alerting relevant authorities and the public.","PeriodicalId":132900,"journal":{"name":"TENCON 2018 - 2018 IEEE Region 10 Conference","volume":"97 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127101119","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/TENCON.2018.8650060
Dhritiman Das, Siddharth Deshmukh
In this paper we propose a non-uniform quantization technique for reporting the test statistics computed by local cognitive radio (CR) sensors to the fusion center. Local CR sensors are assumed to compute the test statistics based on conventional energy detection technique and fusion center on accumulating the respective test statistics makes final decision about presence or absence of primary user (PU). In order to communicate the test statistics to the fusion center over a band limited channel, the locally computed test statistics are quantized using d bit non-uniform quantizer. The number of bits can be selected on the basis of system specification or amount of backhaul communication that can be supported by the system. The proposed non-uniform quantization is performed on the basis of likelihood function, which is defined as the probability of null and alternate hypothesis when test statistic is known. Next at fusion center the reported local quantized test statistics are combined using optimal weights to get global test statistic. Finally the global test statistic is compared with a threshold to decide for presence or absence of PU. Our simulation results illustrate that the performance of proposed non-uniform quantization is better than conventional uniform quantization.
{"title":"Non-Uniform Quantization based Reporting in Cooperative Cognitive Radio","authors":"Dhritiman Das, Siddharth Deshmukh","doi":"10.1109/TENCON.2018.8650060","DOIUrl":"https://doi.org/10.1109/TENCON.2018.8650060","url":null,"abstract":"In this paper we propose a non-uniform quantization technique for reporting the test statistics computed by local cognitive radio (CR) sensors to the fusion center. Local CR sensors are assumed to compute the test statistics based on conventional energy detection technique and fusion center on accumulating the respective test statistics makes final decision about presence or absence of primary user (PU). In order to communicate the test statistics to the fusion center over a band limited channel, the locally computed test statistics are quantized using d bit non-uniform quantizer. The number of bits can be selected on the basis of system specification or amount of backhaul communication that can be supported by the system. The proposed non-uniform quantization is performed on the basis of likelihood function, which is defined as the probability of null and alternate hypothesis when test statistic is known. Next at fusion center the reported local quantized test statistics are combined using optimal weights to get global test statistic. Finally the global test statistic is compared with a threshold to decide for presence or absence of PU. Our simulation results illustrate that the performance of proposed non-uniform quantization is better than conventional uniform quantization.","PeriodicalId":132900,"journal":{"name":"TENCON 2018 - 2018 IEEE Region 10 Conference","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127536357","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/TENCON.2018.8650503
A. Sulaiman, F. Abdullah, M. Z. Jamaludin, A. Ismail, M. Mahdi
We investigate an intensity influence towards the flatness of multiwavelength fiber laser (MWFL) based on intensity dependent transmission (IDT) mechanism. The intensity is varied by changing semiconductor optical amplifier (SOA) current and throughput port ratio. Owing to the IDT mechanism, the multiwavelength flatness is degraded with the increment of SOA current. The change of throughput port ratio of optical splitter from 10% to 90% has also affected a worse multiwavelength flatness. The flattest multiwavelength spectrum is achieved at SOA current and throughput port of 150 mA and 10%, respectively, with the lasing lines are counted up to 300 channels within 3 dB uniformity.
{"title":"SOA-based Multiwavelength Fiber Laser Assisted by Intensity Dependent Transmission Mechanism","authors":"A. Sulaiman, F. Abdullah, M. Z. Jamaludin, A. Ismail, M. Mahdi","doi":"10.1109/TENCON.2018.8650503","DOIUrl":"https://doi.org/10.1109/TENCON.2018.8650503","url":null,"abstract":"We investigate an intensity influence towards the flatness of multiwavelength fiber laser (MWFL) based on intensity dependent transmission (IDT) mechanism. The intensity is varied by changing semiconductor optical amplifier (SOA) current and throughput port ratio. Owing to the IDT mechanism, the multiwavelength flatness is degraded with the increment of SOA current. The change of throughput port ratio of optical splitter from 10% to 90% has also affected a worse multiwavelength flatness. The flattest multiwavelength spectrum is achieved at SOA current and throughput port of 150 mA and 10%, respectively, with the lasing lines are counted up to 300 channels within 3 dB uniformity.","PeriodicalId":132900,"journal":{"name":"TENCON 2018 - 2018 IEEE Region 10 Conference","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129928448","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/TENCON.2018.8650218
Joy Nathalie M. Avelino, Carmi Anne Loren Mora, J. P. Balaquit
The role of big data and machine intelligence in the field of information security is gaining importance as malicious attackers use evasion techniques (polymorphism, encryption, obfuscation) to bypass signature-based detection. As most threats propagate through the network, it is important to have proactive techniques to discover an infection before it damages a computer.This paper will examine how header-based information as well as other characteristics in the HTTP network traffic can be used to train a machine learning model to capture malicious behavior.Network streams tagged as malicious are preprocessed and clustered. It has been found that features in the raw byte stream augmented with handcrafted features are useful in learning the characteristics of network threats.In specific clusters formed, it is possible to identify certain threats targeting a specific server, or if there are characteristics that can be observed in the injected code for exploit detection.Clustering malicious network traffic leads to a better understanding of protection against these types of threats, identification of connected malware campaigns, and insight on future trends.
{"title":"Ahead of the Curve: A Deeper Understanding of Network Threats Through Machine Learning","authors":"Joy Nathalie M. Avelino, Carmi Anne Loren Mora, J. P. Balaquit","doi":"10.1109/TENCON.2018.8650218","DOIUrl":"https://doi.org/10.1109/TENCON.2018.8650218","url":null,"abstract":"The role of big data and machine intelligence in the field of information security is gaining importance as malicious attackers use evasion techniques (polymorphism, encryption, obfuscation) to bypass signature-based detection. As most threats propagate through the network, it is important to have proactive techniques to discover an infection before it damages a computer.This paper will examine how header-based information as well as other characteristics in the HTTP network traffic can be used to train a machine learning model to capture malicious behavior.Network streams tagged as malicious are preprocessed and clustered. It has been found that features in the raw byte stream augmented with handcrafted features are useful in learning the characteristics of network threats.In specific clusters formed, it is possible to identify certain threats targeting a specific server, or if there are characteristics that can be observed in the injected code for exploit detection.Clustering malicious network traffic leads to a better understanding of protection against these types of threats, identification of connected malware campaigns, and insight on future trends.","PeriodicalId":132900,"journal":{"name":"TENCON 2018 - 2018 IEEE Region 10 Conference","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130202986","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/TENCON.2018.8650321
Denise Lee, Mei Yu Soh, T. Teo, K. Yeo
The adoption of IoT enabled devices have led to higher risks of EMI issues, especially in their power management. Particularly, rectifiers for IoT devices are especially concerning. To mitigate this, one approach is to use wide-bandgap semiconductor devices for power management, which reduces the devices’ susceptibility and emission of conducted EMI. One such example are GaN semiconductors. However, for low voltage, low frequency systems, the literature concerning GaN performance with regards to conducted EMI have been sparse. This is particularly important for industries such as the medical field which might require low voltage power circuits with components more adept to handling EMI. In this paper, an evaluation will be carried out to compare transistors’ performance in the design of a low voltage, low frequency rectifier of 12 Vac at 50 Hz. The results of the simulation using IGBT, GaN FETs and MOSFETs are then shown and discussed.
{"title":"Evaluation of Low Voltage Rectifier Design Using IGBT, MOSFET, and GaN FETs","authors":"Denise Lee, Mei Yu Soh, T. Teo, K. Yeo","doi":"10.1109/TENCON.2018.8650321","DOIUrl":"https://doi.org/10.1109/TENCON.2018.8650321","url":null,"abstract":"The adoption of IoT enabled devices have led to higher risks of EMI issues, especially in their power management. Particularly, rectifiers for IoT devices are especially concerning. To mitigate this, one approach is to use wide-bandgap semiconductor devices for power management, which reduces the devices’ susceptibility and emission of conducted EMI. One such example are GaN semiconductors. However, for low voltage, low frequency systems, the literature concerning GaN performance with regards to conducted EMI have been sparse. This is particularly important for industries such as the medical field which might require low voltage power circuits with components more adept to handling EMI. In this paper, an evaluation will be carried out to compare transistors’ performance in the design of a low voltage, low frequency rectifier of 12 Vac at 50 Hz. The results of the simulation using IGBT, GaN FETs and MOSFETs are then shown and discussed.","PeriodicalId":132900,"journal":{"name":"TENCON 2018 - 2018 IEEE Region 10 Conference","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130613763","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/TENCON.2018.8650350
Y. Seo, Deokjin Seo, Chang-Sung Jeong
With the development of media including newspaper written by robots and many unreliable sources, it’s getting hard to distinguish whether the news is true or not. In this paper, we shall present a novel fake news detection model, FaNDeR(Fake News Detection model using media Reliability) which can efficiently classify the level of truth for the news in the question answering system based on modified CNN deep learning model. Our model reflects the reliability of various medias by training with the input dataset which contains the truthfulness of each media as well as that of the proposition. Our model is designed for higher accuracy with media dataset in terms of data augmentation, batch size control and model modification. We shall show that our model has higher accuracy over statistical approach by reflecting the tendency of truth level for each media through the training of the dataset collected so far.
随着媒体的发展，包括机器人写的报纸和许多不可靠的消息来源，很难区分新闻是真是假。在本文中，我们将提出一种新的假新闻检测模型FaNDeR(fake news detection model using media Reliability)，它可以基于改进的CNN深度学习模型对问答系统中的新闻的真实程度进行有效的分类。我们的模型通过使用包含每种媒体真实性以及命题真实性的输入数据集进行训练来反映各种媒体的可靠性。我们的模型在数据增强、批量大小控制和模型修改方面针对媒体数据集设计了更高的精度。我们将通过迄今收集的数据集的训练来反映每种媒体的真实水平的趋势，从而表明我们的模型比统计方法具有更高的准确性。
{"title":"FaNDeR: Fake News Detection Model Using Media Reliability","authors":"Y. Seo, Deokjin Seo, Chang-Sung Jeong","doi":"10.1109/TENCON.2018.8650350","DOIUrl":"https://doi.org/10.1109/TENCON.2018.8650350","url":null,"abstract":"With the development of media including newspaper written by robots and many unreliable sources, it’s getting hard to distinguish whether the news is true or not. In this paper, we shall present a novel fake news detection model, FaNDeR(Fake News Detection model using media Reliability) which can efficiently classify the level of truth for the news in the question answering system based on modified CNN deep learning model. Our model reflects the reliability of various medias by training with the input dataset which contains the truthfulness of each media as well as that of the proposition. Our model is designed for higher accuracy with media dataset in terms of data augmentation, batch size control and model modification. We shall show that our model has higher accuracy over statistical approach by reflecting the tendency of truth level for each media through the training of the dataset collected so far.","PeriodicalId":132900,"journal":{"name":"TENCON 2018 - 2018 IEEE Region 10 Conference","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130649700","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：