Pub Date : 2018-07-01DOI: 10.1109/JCSSE.2018.8457378
Sirinthra Chantharaj, Kissada Pornratthanapong, Pitchayut Chitsinpchayakun, Teerapong Panboonyuen, P. Vateekul, S. Lawawirojwong, Panu Srestasathiern, Kulsawasd Jitkajornwanich
Semantic Segmentation is a fundamental task in computer vision and remote sensing imagery. Many applications, such as urban planning, change detection, and environmental monitoring, require the accurate segmentation; hence, most segmentation tasks are performed by humans. Currently, with the growth of Deep Convolutional Neural Network (DCNN), there are many works aiming to find the best network architecture fitting for this task. However, all of the studies are based on very-high resolution satellite images, and surprisingly; none of them are implemented on medium resolution satellite images. Moreover, no research has applied geoinformatics knowledge. Therefore, we purpose to compare the semantic segmentation models, which are FCN, SegNet, and GSN using medium resolution images from Landsat-8 satellite. In addition, we propose a modified SegNet model that can be used with remote sensing derived indices. The results show that the model that achieves the highest accuracy RGB bands of medium resolution aerial imagery is SegNet. The overall accuracy of the model increases when includes Near Infrared (NIR) and Short-Wave Infrared (SWIR) band. The results showed that our proposed method (our modified SegNet model, named RGB-IR-IDX-MSN method) outperforms all of the baselines in terms of mean F1 scores.
{"title":"Semantic Segmentation On Medium-Resolution Satellite Images Using Deep Convolutional Networks With Remote Sensing Derived Indices","authors":"Sirinthra Chantharaj, Kissada Pornratthanapong, Pitchayut Chitsinpchayakun, Teerapong Panboonyuen, P. Vateekul, S. Lawawirojwong, Panu Srestasathiern, Kulsawasd Jitkajornwanich","doi":"10.1109/JCSSE.2018.8457378","DOIUrl":"https://doi.org/10.1109/JCSSE.2018.8457378","url":null,"abstract":"Semantic Segmentation is a fundamental task in computer vision and remote sensing imagery. Many applications, such as urban planning, change detection, and environmental monitoring, require the accurate segmentation; hence, most segmentation tasks are performed by humans. Currently, with the growth of Deep Convolutional Neural Network (DCNN), there are many works aiming to find the best network architecture fitting for this task. However, all of the studies are based on very-high resolution satellite images, and surprisingly; none of them are implemented on medium resolution satellite images. Moreover, no research has applied geoinformatics knowledge. Therefore, we purpose to compare the semantic segmentation models, which are FCN, SegNet, and GSN using medium resolution images from Landsat-8 satellite. In addition, we propose a modified SegNet model that can be used with remote sensing derived indices. The results show that the model that achieves the highest accuracy RGB bands of medium resolution aerial imagery is SegNet. The overall accuracy of the model increases when includes Near Infrared (NIR) and Short-Wave Infrared (SWIR) band. The results showed that our proposed method (our modified SegNet model, named RGB-IR-IDX-MSN method) outperforms all of the baselines in terms of mean F1 scores.","PeriodicalId":338973,"journal":{"name":"2018 15th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129099463","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 : 2018-07-01DOI: 10.1109/JCSSE.2018.8457355
Chanachai Puttaruksa, Pinyo Taeprasartsit
In order to achieve a more accurate deep learning model, we need large amount of data. For imaging application, color data augmentation is usually required. Color jittering is a common current practice for such augmentation where color values in image are slightly adjusted. Unfortunately, color values between two cameras may be significantly different. This makes the current practice ineffective. This work proposes to map color values among cameras by using deep learning to learn color-mapping parameters. In this way, we can augment color data by converting an image from one camera to another image whose colors seemingly are taken from another camera. This allows a machine to learn a model that can deal with input images from multiple cameras without actually using training data from multiple cameras. These parameters can also be employed to calibrate colors in order that all cameras produce the same color tone. The proposed neural network architecture which employs fully connected layers and batch normalization outperforms an existing method and can be systematically performed for any camera pairs to extend its applications in other scenarios.
{"title":"Color Data Augmentation through Learning Color-Mapping Parameters between Cameras","authors":"Chanachai Puttaruksa, Pinyo Taeprasartsit","doi":"10.1109/JCSSE.2018.8457355","DOIUrl":"https://doi.org/10.1109/JCSSE.2018.8457355","url":null,"abstract":"In order to achieve a more accurate deep learning model, we need large amount of data. For imaging application, color data augmentation is usually required. Color jittering is a common current practice for such augmentation where color values in image are slightly adjusted. Unfortunately, color values between two cameras may be significantly different. This makes the current practice ineffective. This work proposes to map color values among cameras by using deep learning to learn color-mapping parameters. In this way, we can augment color data by converting an image from one camera to another image whose colors seemingly are taken from another camera. This allows a machine to learn a model that can deal with input images from multiple cameras without actually using training data from multiple cameras. These parameters can also be employed to calibrate colors in order that all cameras produce the same color tone. The proposed neural network architecture which employs fully connected layers and batch normalization outperforms an existing method and can be systematically performed for any camera pairs to extend its applications in other scenarios.","PeriodicalId":338973,"journal":{"name":"2018 15th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"135 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128830042","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 : 2018-07-01DOI: 10.1109/JCSSE.2018.8457363
Warawut Kesornsukhon, P. Visutsak, S. Ratanasanya
Chromatic Aberration (CA) is an active research topic in the digital era since everybody communicates through digital photos more than they ever do in the past. The digital photos or digital images are not only used to record the precious memories of people but they are also used as a mean to share those precious time with other people. The digital images thus have significant impact to our everyday life as well as the social. CA can distort the memories represented by the digital images since it is a blurring of colors especially Red and Blue colors. CA is a result from using low quality lens, which are part of digital cameras. The low quality lens disperses the light out of the incident point on the other lens and this phenomenon cause the aberration of colors. There were several attempts to detect the CA using pixel-based algorithms and filtering techniques. Some attempts spent too much effort to detect CA but got poor results. However, none of the previous attempts investigated the detection of CA using image segmentation. Therefore, this paper applies image segmentation method to detect CA and compares its performances to the existing methods of detecting CA. The unique characteristics of CA are applied to the selected image segmentation method in order to make it be able to identify the CA segments in the digital images. The preliminary experiments showed that the proposed exploitation can bring out the ability to detect CA with impressive results. The accuracy of the proposed method is up to 95.25% on the average with low false positive rate of 0.90% on the average. Moreover, the proposed method is 42.73% faster than the previous method on the average.
{"title":"Chromatic Aberration Detection Based on Image Segmentation","authors":"Warawut Kesornsukhon, P. Visutsak, S. Ratanasanya","doi":"10.1109/JCSSE.2018.8457363","DOIUrl":"https://doi.org/10.1109/JCSSE.2018.8457363","url":null,"abstract":"Chromatic Aberration (CA) is an active research topic in the digital era since everybody communicates through digital photos more than they ever do in the past. The digital photos or digital images are not only used to record the precious memories of people but they are also used as a mean to share those precious time with other people. The digital images thus have significant impact to our everyday life as well as the social. CA can distort the memories represented by the digital images since it is a blurring of colors especially Red and Blue colors. CA is a result from using low quality lens, which are part of digital cameras. The low quality lens disperses the light out of the incident point on the other lens and this phenomenon cause the aberration of colors. There were several attempts to detect the CA using pixel-based algorithms and filtering techniques. Some attempts spent too much effort to detect CA but got poor results. However, none of the previous attempts investigated the detection of CA using image segmentation. Therefore, this paper applies image segmentation method to detect CA and compares its performances to the existing methods of detecting CA. The unique characteristics of CA are applied to the selected image segmentation method in order to make it be able to identify the CA segments in the digital images. The preliminary experiments showed that the proposed exploitation can bring out the ability to detect CA with impressive results. The accuracy of the proposed method is up to 95.25% on the average with low false positive rate of 0.90% on the average. Moreover, the proposed method is 42.73% faster than the previous method on the average.","PeriodicalId":338973,"journal":{"name":"2018 15th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124381856","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 : 2018-07-01DOI: 10.1109/JCSSE.2018.8457358
Songsri Tangsripairoj, Premmanat Natseevatana
This paper describes the design and development of the business intelligence system for radio communication licensing textbf(BISRCL) for the Office of the National Broadcasting and Telecommunications Commission of Thailand (NBTC). Data comes from two main sources, which are the Frequency Management System (FMS) and the Automated Spectrum Management System (ASMS). This data is integrated by passing through the Extraction-Transformation-Load (ETL) process and stored in data marts. Online analytical processing (OLAP) techniques are exploited to analyze the data in many different perspectives. Moreover, the BISRCL system provides an interactive dashboard, which is a web application for operational and managerial staff, both in the NBTC central office and regional offices, to support their needs. Besides, various forms of analysis reports can be generated by the system. After the prototype system was developed, it was tested and evaluated by the NBTC officers. The user satisfaction survey results show that by overall the users were satisfied with the BISRCL system at the ‘Good’ level. By comparing the proposed BISRCL system to the existing FMS and ASMS systems, users can access the information they need from the BISRCL system faster, more accurate and more efficient than the existing systems. The ultimate goal of this research project is to provide meaningful and valuable information to the NBTC officers and executives for more effective strategic and operational insights and decisionmaking in radio communication licensing.
{"title":"A Business Intelligence System for Radio Communication Licensing: A Case Study of The National Broadcasting and Telecommunications Commission of Thailand","authors":"Songsri Tangsripairoj, Premmanat Natseevatana","doi":"10.1109/JCSSE.2018.8457358","DOIUrl":"https://doi.org/10.1109/JCSSE.2018.8457358","url":null,"abstract":"This paper describes the design and development of the business intelligence system for radio communication licensing textbf(BISRCL) for the Office of the National Broadcasting and Telecommunications Commission of Thailand (NBTC). Data comes from two main sources, which are the Frequency Management System (FMS) and the Automated Spectrum Management System (ASMS). This data is integrated by passing through the Extraction-Transformation-Load (ETL) process and stored in data marts. Online analytical processing (OLAP) techniques are exploited to analyze the data in many different perspectives. Moreover, the BISRCL system provides an interactive dashboard, which is a web application for operational and managerial staff, both in the NBTC central office and regional offices, to support their needs. Besides, various forms of analysis reports can be generated by the system. After the prototype system was developed, it was tested and evaluated by the NBTC officers. The user satisfaction survey results show that by overall the users were satisfied with the BISRCL system at the ‘Good’ level. By comparing the proposed BISRCL system to the existing FMS and ASMS systems, users can access the information they need from the BISRCL system faster, more accurate and more efficient than the existing systems. The ultimate goal of this research project is to provide meaningful and valuable information to the NBTC officers and executives for more effective strategic and operational insights and decisionmaking in radio communication licensing.","PeriodicalId":338973,"journal":{"name":"2018 15th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128059280","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 : 2018-07-01DOI: 10.1109/JCSSE.2018.8457375
W. Kurdthongmee, K. Suwannarat, Praepaka Panyuen, Naruedom Sae-Ma
Sawmills in Thailand demand an automatic approach to correctly detect rubberwood piths. This is a starting point to maximize the yield of slabs per lumber. Knowing the pith location at both cross-section sides of the lumber makes it possible to rotate the lumber in such a way that both piths are parallel to the saws. Then, a knot, the likely to defect part which runs along the length of the lumber, can be removed. In this paper, we propose an algorithm to accelerate the process of approximating the pith location of rubberwoods. The algorithm employs histogram of oriented gradients (HOG) and a set of relevant histogram bin indices to significantly reduce the number of line segments to be later used in a complex group of lines intersection part of the algorithm. This is in contrast to previously proposed algorithms that employ all edge points to create a huge amount of line segments which consume extremely high processing time. The results confirm that 3,315 times performance is reached at 0.52 reduction of detection error in average compared to the state of the art implementation on a set of 35 cross-section rubberwood images taken by a normal camera.
{"title":"A Fast Algorithm to Approximate the Pith Location of Rubberwood Timber from a Normal Camera Image","authors":"W. Kurdthongmee, K. Suwannarat, Praepaka Panyuen, Naruedom Sae-Ma","doi":"10.1109/JCSSE.2018.8457375","DOIUrl":"https://doi.org/10.1109/JCSSE.2018.8457375","url":null,"abstract":"Sawmills in Thailand demand an automatic approach to correctly detect rubberwood piths. This is a starting point to maximize the yield of slabs per lumber. Knowing the pith location at both cross-section sides of the lumber makes it possible to rotate the lumber in such a way that both piths are parallel to the saws. Then, a knot, the likely to defect part which runs along the length of the lumber, can be removed. In this paper, we propose an algorithm to accelerate the process of approximating the pith location of rubberwoods. The algorithm employs histogram of oriented gradients (HOG) and a set of relevant histogram bin indices to significantly reduce the number of line segments to be later used in a complex group of lines intersection part of the algorithm. This is in contrast to previously proposed algorithms that employ all edge points to create a huge amount of line segments which consume extremely high processing time. The results confirm that 3,315 times performance is reached at 0.52 reduction of detection error in average compared to the state of the art implementation on a set of 35 cross-section rubberwood images taken by a normal camera.","PeriodicalId":338973,"journal":{"name":"2018 15th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"56 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131798568","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}
This study presents a density-based incremental clustering algorithm which incorporates the concept of fuzzy set in clustering. Unlike other existing fuzzy clustering algorithms which are c-mean clustering where the number of clusters must be pre-defined, the proposed algorithm incorporates the concept of fuzzy set into density-based clustering where the number of clusters is not restricted. Moreover, the proposed algorithm uses incremental clustering usually employed in stream data clustering, leading to linear computation time, rather than quadratic computation time as in other density-based clustering. The proposed algorithm outperforms other existing density-based clustering algorithms in terms of both clustering results and computation time. As a result, the proposed algorithm can much efficiently process large data sets than other density-based clustering algorithms.
{"title":"A Fuzzy Density-based Incremental Clustering Algorithm","authors":"Sirisup Laohakiat, Photchanan Ratanajaipan, Leenhapat Navaravong, Rachanee Ungrangsi, Krissada Maleewong","doi":"10.1109/JCSSE.2018.8457385","DOIUrl":"https://doi.org/10.1109/JCSSE.2018.8457385","url":null,"abstract":"This study presents a density-based incremental clustering algorithm which incorporates the concept of fuzzy set in clustering. Unlike other existing fuzzy clustering algorithms which are c-mean clustering where the number of clusters must be pre-defined, the proposed algorithm incorporates the concept of fuzzy set into density-based clustering where the number of clusters is not restricted. Moreover, the proposed algorithm uses incremental clustering usually employed in stream data clustering, leading to linear computation time, rather than quadratic computation time as in other density-based clustering. The proposed algorithm outperforms other existing density-based clustering algorithms in terms of both clustering results and computation time. As a result, the proposed algorithm can much efficiently process large data sets than other density-based clustering algorithms.","PeriodicalId":338973,"journal":{"name":"2018 15th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"255 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132661748","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 : 2018-07-01DOI: 10.1109/JCSSE.2018.8457388
Noppawat Samdangdech, S. Phiphobmongkol
The visual estimation of log volume and size distribution of eucalyptus logs on a truck is a challenging task. In Thailand, inspectors at paper mills typically perform this task. The information is used to determine whether the logs pass the criteria for the mill and to find the appropriate price. This method is far from accurate and not efficient. This paper presents a new approach to automatically detects eucalyptus logend cut area from rear-end images of eucalyptus timber trucks. The method used machine learning and image processing techniques. It consists of three parts: timber truck detection, log segmentation, and log counting. The proposed system was tested with 300 images of timber truck dataset and achieved an average accuracy of 94.45% in log segmentation and 2.71% of false negative.
{"title":"Log-End Cut-Area Detection in Images Taken from Rear End of Eucalyptus Timber Trucks","authors":"Noppawat Samdangdech, S. Phiphobmongkol","doi":"10.1109/JCSSE.2018.8457388","DOIUrl":"https://doi.org/10.1109/JCSSE.2018.8457388","url":null,"abstract":"The visual estimation of log volume and size distribution of eucalyptus logs on a truck is a challenging task. In Thailand, inspectors at paper mills typically perform this task. The information is used to determine whether the logs pass the criteria for the mill and to find the appropriate price. This method is far from accurate and not efficient. This paper presents a new approach to automatically detects eucalyptus logend cut area from rear-end images of eucalyptus timber trucks. The method used machine learning and image processing techniques. It consists of three parts: timber truck detection, log segmentation, and log counting. The proposed system was tested with 300 images of timber truck dataset and achieved an average accuracy of 94.45% in log segmentation and 2.71% of false negative.","PeriodicalId":338973,"journal":{"name":"2018 15th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"71 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115981276","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 : 2018-07-01DOI: 10.1109/JCSSE.2018.8457354
Pakapol Krongbaramee, Yuthapong Somchit
A Software Defined Networking (SDN) has been deployed in the current network system. Together with Network Virtualization (NVF), it makes the network become more flexible. The firewall can be implemented in SDN. However, with the limitation of earlier version of OpenFlow protocol used in SDN, the stateful firewall could not be implemented with the SDN standard. The development of OpenFlow enables some features that can be used for implementing the stateful firewall. In this work, we implement the stateful firewall in the SDN switch on the data plane. The Open vSwitch is used. We also evaluate the performance of the SDN stateful firewall. The results show that our SDN stateful firewall can work correctly with small overhead increased in SDN switches.
当前网络系统中已经部署了SDN (Software Defined Networking)。它与网络虚拟化(NVF)一起使网络变得更加灵活。防火墙可以在SDN网络中实现。但是，由于SDN中使用的OpenFlow协议的早期版本的限制，无法使用SDN标准实现状态防火墙。OpenFlow的开发提供了一些可用于实现有状态防火墙的特性。在这项工作中，我们在数据平面的SDN交换机上实现了有状态防火墙。使用Open vSwitch。我们还评估了SDN状态防火墙的性能。结果表明，我们设计的SDN状态防火墙可以在SDN交换机增加少量开销的情况下正常工作。
{"title":"Implementation of SDN Stateful Firewall on Data Plane using Open vSwitch","authors":"Pakapol Krongbaramee, Yuthapong Somchit","doi":"10.1109/JCSSE.2018.8457354","DOIUrl":"https://doi.org/10.1109/JCSSE.2018.8457354","url":null,"abstract":"A Software Defined Networking (SDN) has been deployed in the current network system. Together with Network Virtualization (NVF), it makes the network become more flexible. The firewall can be implemented in SDN. However, with the limitation of earlier version of OpenFlow protocol used in SDN, the stateful firewall could not be implemented with the SDN standard. The development of OpenFlow enables some features that can be used for implementing the stateful firewall. In this work, we implement the stateful firewall in the SDN switch on the data plane. The Open vSwitch is used. We also evaluate the performance of the SDN stateful firewall. The results show that our SDN stateful firewall can work correctly with small overhead increased in SDN switches.","PeriodicalId":338973,"journal":{"name":"2018 15th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"79 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132136456","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 : 2018-07-01DOI: 10.1109/JCSSE.2018.8457395
Charinya Wangwatcharakul, S. Wongthanavasu
The recommender system is an efficient tool for online application, which exploits historical user rating on item to make recommendations on items to users. This paper aims to enhance dynamic recommender systems under volatile user preference drifts. It proposed an algorithm to solve sparse data by using Gaussian mixture model to fill in data matrix for sparsity reduction and improve more completely ratings prediction. Subsequently, it utilizes item clustering and linear regression technique to predict the future interests of users in category based and additionally uses the nearest neighbor method to prevent over-fitting. The experimental results show that the proposed approach provides the better performance on rating prediction when compared with the state-of-the-art dynamic recommendation algorithms.
{"title":"Improving Dynamic Recommender System Based on Item Clustering for Preference Drifts","authors":"Charinya Wangwatcharakul, S. Wongthanavasu","doi":"10.1109/JCSSE.2018.8457395","DOIUrl":"https://doi.org/10.1109/JCSSE.2018.8457395","url":null,"abstract":"The recommender system is an efficient tool for online application, which exploits historical user rating on item to make recommendations on items to users. This paper aims to enhance dynamic recommender systems under volatile user preference drifts. It proposed an algorithm to solve sparse data by using Gaussian mixture model to fill in data matrix for sparsity reduction and improve more completely ratings prediction. Subsequently, it utilizes item clustering and linear regression technique to predict the future interests of users in category based and additionally uses the nearest neighbor method to prevent over-fitting. The experimental results show that the proposed approach provides the better performance on rating prediction when compared with the state-of-the-art dynamic recommendation algorithms.","PeriodicalId":338973,"journal":{"name":"2018 15th International Joint Conference on Computer Science and Software Engineering (JCSSE)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130858409","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：