Pub Date : 2017-09-01DOI: 10.1109/IDAP.2017.8090213
Serkan Turkeli, B. S. A. Gazioglu, Kenan Kaan Kurt, Hüseyin Tanzer Atay, Yakup Gorur
Finding similar diagnoses for the same region are vital for patients. In this paper, we aim to find the similarity radiology reports based on bag-of-words (BoW) and Fuzzy C-Means Clustering methods. A double-layer structure is applied. Firstly, extracting features from data BoW method is applied and then Fuzzy C-Means algorithm is performed to cluster the blocks into the similar cluster and the non-similar cluster. 457 radiology reports were examined which were collected from a research and education hospital in Istanbul. Data were tested according to the 23 regions and 137 diagnosis. By the opinion of the radiologist a vocabulary consists of these regions and diagnosis were created. Experimental results on data sets have shown that for the standard documents BoW and Fuzzy C-Means Clustering can be used to find similarity.
{"title":"Mining similar radiology reports using BoW and Fuzzy C-means clustering","authors":"Serkan Turkeli, B. S. A. Gazioglu, Kenan Kaan Kurt, Hüseyin Tanzer Atay, Yakup Gorur","doi":"10.1109/IDAP.2017.8090213","DOIUrl":"https://doi.org/10.1109/IDAP.2017.8090213","url":null,"abstract":"Finding similar diagnoses for the same region are vital for patients. In this paper, we aim to find the similarity radiology reports based on bag-of-words (BoW) and Fuzzy C-Means Clustering methods. A double-layer structure is applied. Firstly, extracting features from data BoW method is applied and then Fuzzy C-Means algorithm is performed to cluster the blocks into the similar cluster and the non-similar cluster. 457 radiology reports were examined which were collected from a research and education hospital in Istanbul. Data were tested according to the 23 regions and 137 diagnosis. By the opinion of the radiologist a vocabulary consists of these regions and diagnosis were created. Experimental results on data sets have shown that for the standard documents BoW and Fuzzy C-Means Clustering can be used to find similarity.","PeriodicalId":111721,"journal":{"name":"2017 International Artificial Intelligence and Data Processing Symposium (IDAP)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129284542","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 : 2017-09-01DOI: 10.1109/IDAP.2017.8090252
Doygun Demirol, Gurkan Tuna, Resul Das
Although the Internet offers numerous advantages, it raises many information security risks, especially against young people and children, who are today amongst the largest user groups of mobile and online technologies all around the world. Therefore, to empower and protect Internet users, it is necessary to develop proper strategies and tools to encapsulate their needs, and identify and prevent all types of the information security risks that may arise during the use of the Internet. In this study, a tracking system to ensure the safe use of the Internet is proposed. Considering the distribution of its potential users, the system has an easy-to-use graphical user interface. By recognizing dangerous web sites and IP addresses, the proposed system blocks access to those sites and this way provides reliable Internet access to its users. During the Internet access, it analyzes accessed IP addresses and port numbers in terms of access type and time and informs the user of the corresponding port numbers which must be closed for safe Internet access. Moreover, by continuously checking the host redirection file of the computer it runs on, it identifies redirections from web addresses to specific IP addresses and this way provides protection against phishing attacks which are becoming one of the most common Internet threats. Although the proposed system is a simple application since it is an open source, freeware application designed for children, it can be improved to consider more sophisticated attack types.
{"title":"A simple logging system for safe internet use","authors":"Doygun Demirol, Gurkan Tuna, Resul Das","doi":"10.1109/IDAP.2017.8090252","DOIUrl":"https://doi.org/10.1109/IDAP.2017.8090252","url":null,"abstract":"Although the Internet offers numerous advantages, it raises many information security risks, especially against young people and children, who are today amongst the largest user groups of mobile and online technologies all around the world. Therefore, to empower and protect Internet users, it is necessary to develop proper strategies and tools to encapsulate their needs, and identify and prevent all types of the information security risks that may arise during the use of the Internet. In this study, a tracking system to ensure the safe use of the Internet is proposed. Considering the distribution of its potential users, the system has an easy-to-use graphical user interface. By recognizing dangerous web sites and IP addresses, the proposed system blocks access to those sites and this way provides reliable Internet access to its users. During the Internet access, it analyzes accessed IP addresses and port numbers in terms of access type and time and informs the user of the corresponding port numbers which must be closed for safe Internet access. Moreover, by continuously checking the host redirection file of the computer it runs on, it identifies redirections from web addresses to specific IP addresses and this way provides protection against phishing attacks which are becoming one of the most common Internet threats. Although the proposed system is a simple application since it is an open source, freeware application designed for children, it can be improved to consider more sophisticated attack types.","PeriodicalId":111721,"journal":{"name":"2017 International Artificial Intelligence and Data Processing Symposium (IDAP)","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128841871","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 : 2017-09-01DOI: 10.1109/IDAP.2017.8090322
Buket Toptaş, D. Hanbay
Image processing based systems take important place in systems used to detect fires in open spaces. Vision-based systems can detect fires in open spaces from distances and detect the fire at an early stage. In this study, a fire/flame detection method based on the color analysis of the fire image is presented. The proposed method consists of three steps. First, the image in RGB space is converted to the HSI color space. Then a color filter is applied to determine the fire/flame candidate zone. In the second stage, fake fire zones within the candidate zone identified as fire are eliminated. In this phase, image difference and gauss mixture model is used to recover the fake fire areas. In the third step, the result of the two methods is subjected to AND processing. The AND operation ensures to detect the exact flame zone. As a result, the proposed algorithm has been tested using fire video images. The highest calculated accuracy is 96%.
{"title":"Flame detection using HSI color space","authors":"Buket Toptaş, D. Hanbay","doi":"10.1109/IDAP.2017.8090322","DOIUrl":"https://doi.org/10.1109/IDAP.2017.8090322","url":null,"abstract":"Image processing based systems take important place in systems used to detect fires in open spaces. Vision-based systems can detect fires in open spaces from distances and detect the fire at an early stage. In this study, a fire/flame detection method based on the color analysis of the fire image is presented. The proposed method consists of three steps. First, the image in RGB space is converted to the HSI color space. Then a color filter is applied to determine the fire/flame candidate zone. In the second stage, fake fire zones within the candidate zone identified as fire are eliminated. In this phase, image difference and gauss mixture model is used to recover the fake fire areas. In the third step, the result of the two methods is subjected to AND processing. The AND operation ensures to detect the exact flame zone. As a result, the proposed algorithm has been tested using fire video images. The highest calculated accuracy is 96%.","PeriodicalId":111721,"journal":{"name":"2017 International Artificial Intelligence and Data Processing Symposium (IDAP)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126889092","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 : 2017-09-01DOI: 10.1109/IDAP.2017.8090316
Ebubekir Buber, Ozgur Koray Sahingoz
Son yillardaki farkli alanlarda gelişen teknolojiler ile birlikte makine öğrenmesi konsepti hayatimizin hemen hemen her sayisallaşan alanina girmiştir. Özellikle yapay sinir ağlari konusundaki gelişmeler, paralel hesaplama ortamlarindaki atilimlar ve derin öğrenme gibi yeni alanlarla birlikte makine öğrenmesi yaklaşimi farkli uygulama alanlarinda kullanilmaya başlanmiştir. Ancak bu süreçte kullanilan parametreler öğrenme yaklaşiminin performansini ciddi boyutlarda etkilemektedir. Bu çalişmamizda özel bir uygulama platformu için kullanilan değişik parametrelerin sistem performansina etkisi değerlendirilmiş ve en uygun parametre değerlerinin (Eğitim Adimi Sayisi (Epoch), Katman ve Nöron Sayisi (Neuron Size), Öğrenme Katsayisi (Learning Rate) ve Mini-Batch Boyutu (Mini-Batch Size)) nasil seçildiği gösterilmiştir.
{"title":"Makine Öğrenmesi sistemi ile görüntü İşleme ve en uygun parametrelerin ayarlanmasi","authors":"Ebubekir Buber, Ozgur Koray Sahingoz","doi":"10.1109/IDAP.2017.8090316","DOIUrl":"https://doi.org/10.1109/IDAP.2017.8090316","url":null,"abstract":"Son yillardaki farkli alanlarda gelişen teknolojiler ile birlikte makine öğrenmesi konsepti hayatimizin hemen hemen her sayisallaşan alanina girmiştir. Özellikle yapay sinir ağlari konusundaki gelişmeler, paralel hesaplama ortamlarindaki atilimlar ve derin öğrenme gibi yeni alanlarla birlikte makine öğrenmesi yaklaşimi farkli uygulama alanlarinda kullanilmaya başlanmiştir. Ancak bu süreçte kullanilan parametreler öğrenme yaklaşiminin performansini ciddi boyutlarda etkilemektedir. Bu çalişmamizda özel bir uygulama platformu için kullanilan değişik parametrelerin sistem performansina etkisi değerlendirilmiş ve en uygun parametre değerlerinin (Eğitim Adimi Sayisi (Epoch), Katman ve Nöron Sayisi (Neuron Size), Öğrenme Katsayisi (Learning Rate) ve Mini-Batch Boyutu (Mini-Batch Size)) nasil seçildiği gösterilmiştir.","PeriodicalId":111721,"journal":{"name":"2017 International Artificial Intelligence and Data Processing Symposium (IDAP)","volume":"124 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121425763","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 : 2017-09-01DOI: 10.1109/IDAP.2017.8090174
Ümit Budak, A. Şengür, Yaman Akbulut
Exudates are among the first signs of diabetic retinopathy and one of the main causes of vision loss in diabetic patients. In this study, an approach based on clustering and morphological image processing has been proposed for detection of retinal exudates. Contrast-limited adaptive histogram equalization technique is used to make the location of the exudate areas more specific. In addition, the k-means clustering algorithm determines the locations of candidate regions. According to experimental results, it was observed that a majority of the pixels of the exudate regions were detected.
{"title":"Localization of macular edema region from color retinal images for detection of diabetic retinopathy","authors":"Ümit Budak, A. Şengür, Yaman Akbulut","doi":"10.1109/IDAP.2017.8090174","DOIUrl":"https://doi.org/10.1109/IDAP.2017.8090174","url":null,"abstract":"Exudates are among the first signs of diabetic retinopathy and one of the main causes of vision loss in diabetic patients. In this study, an approach based on clustering and morphological image processing has been proposed for detection of retinal exudates. Contrast-limited adaptive histogram equalization technique is used to make the location of the exudate areas more specific. In addition, the k-means clustering algorithm determines the locations of candidate regions. According to experimental results, it was observed that a majority of the pixels of the exudate regions were detected.","PeriodicalId":111721,"journal":{"name":"2017 International Artificial Intelligence and Data Processing Symposium (IDAP)","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114183167","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 : 2017-09-01DOI: 10.1109/IDAP.2017.8090319
M. Rashid, A. Muhtaroğlu
Data processing performed by adder circuits need to achieve low delay and low power at the same time while maintaining low cost, due to the steep growth in mobile computation devices. Recently proposed 1-bit full adder design that hybridizes transmission gates (TG) and standard CMOS offers significant PDP improvement. Two full adder implementations are presented in this paper which further optimizes the previously presented circuits: First (CKT1) deploys GDI-cell based XNOR module to decrease PDP, while the second circuit (CKT2) reduces the worst case delay with equivalent PDP. Simulation results indicate the proposed CKT1 has 4.8% and 2.5% reduced PDP for realistic cascade and FO4 loads respectively, with 16% improved cost compared to literature. CKT2 maintains comparable PDP with 11.3% and 2% improved delay for realistic cascade and FO4 loads respectively.
{"title":"Power delay product optimized hybrid full adder circuits","authors":"M. Rashid, A. Muhtaroğlu","doi":"10.1109/IDAP.2017.8090319","DOIUrl":"https://doi.org/10.1109/IDAP.2017.8090319","url":null,"abstract":"Data processing performed by adder circuits need to achieve low delay and low power at the same time while maintaining low cost, due to the steep growth in mobile computation devices. Recently proposed 1-bit full adder design that hybridizes transmission gates (TG) and standard CMOS offers significant PDP improvement. Two full adder implementations are presented in this paper which further optimizes the previously presented circuits: First (CKT1) deploys GDI-cell based XNOR module to decrease PDP, while the second circuit (CKT2) reduces the worst case delay with equivalent PDP. Simulation results indicate the proposed CKT1 has 4.8% and 2.5% reduced PDP for realistic cascade and FO4 loads respectively, with 16% improved cost compared to literature. CKT2 maintains comparable PDP with 11.3% and 2% improved delay for realistic cascade and FO4 loads respectively.","PeriodicalId":111721,"journal":{"name":"2017 International Artificial Intelligence and Data Processing Symposium (IDAP)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114259888","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 : 2017-09-01DOI: 10.1109/IDAP.2017.8090264
T. Sağ, H. Kahramanli
In this paper, a novel approach for classification rule mining is presented. The remarkable relationship between the rule extraction procedure and the concept of multiobjective optimization is emphasized. The range values of features composing the rules are handled as decision variables in the modelled multiobjective optimization problem. The proposed method is applied to three well-known datasets in literature. These are Iris, Haberman's Survival Data and Pima Indians Diabetes Datasets obtained from machine learning repository of University of California at Irvine (UCI). The classification rules are extracted with 100% accuracy for all datasets. These experimental results are the best outcomes found in literature so far.
{"title":"Classification rule mining approach based on multiobjective optimization","authors":"T. Sağ, H. Kahramanli","doi":"10.1109/IDAP.2017.8090264","DOIUrl":"https://doi.org/10.1109/IDAP.2017.8090264","url":null,"abstract":"In this paper, a novel approach for classification rule mining is presented. The remarkable relationship between the rule extraction procedure and the concept of multiobjective optimization is emphasized. The range values of features composing the rules are handled as decision variables in the modelled multiobjective optimization problem. The proposed method is applied to three well-known datasets in literature. These are Iris, Haberman's Survival Data and Pima Indians Diabetes Datasets obtained from machine learning repository of University of California at Irvine (UCI). The classification rules are extracted with 100% accuracy for all datasets. These experimental results are the best outcomes found in literature so far.","PeriodicalId":111721,"journal":{"name":"2017 International Artificial Intelligence and Data Processing Symposium (IDAP)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125121988","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 : 2017-09-01DOI: 10.1109/IDAP.2017.8090183
Z. Cebeci, A. T. Kavlak, Figen Yildiz
Unsupervised fuzzy clustering is an important tool for finding the meaningful patterns in data sets. In fuzzy clustering analyses, the performances of clustering algorithms are mostly compared using several internal fuzzy validity indices. However, since the well-known fuzzy indices have originally been proposed for working with membership degrees produced by the traditional Fuzzy c-means Clustering (FCM) algorithm, these indices cannot be used for possibilistic algorithms that produce typicality matrices instead of fuzzy membership matrices. Even more, the variants of FCM and PCM such as Possibilistic Fuzzy C-means (PFCM) and Fuzzy Possibilistic C-means (FPCM) simultaneously result with probabilistic and possibilistic membership degrees. Thus, some kind of validity indices are needed for working with both of these results. For this purpose, a few extended and generalized validity indices has been proposed in recent years. In this paper, the performances of these indices were examined for validating the clustering results from Unsupervised Possibilistic Fuzzy Clustering (UPFC), FCM and PCM algorithms. The findings showed that generalized versions of the fuzzy validity indices based on normalization of typicality degrees can be successfully used to validate the results from PCM, UPFC and the variants of FCM and PCM.
{"title":"Validation of fuzzy and possibilistic clustering results","authors":"Z. Cebeci, A. T. Kavlak, Figen Yildiz","doi":"10.1109/IDAP.2017.8090183","DOIUrl":"https://doi.org/10.1109/IDAP.2017.8090183","url":null,"abstract":"Unsupervised fuzzy clustering is an important tool for finding the meaningful patterns in data sets. In fuzzy clustering analyses, the performances of clustering algorithms are mostly compared using several internal fuzzy validity indices. However, since the well-known fuzzy indices have originally been proposed for working with membership degrees produced by the traditional Fuzzy c-means Clustering (FCM) algorithm, these indices cannot be used for possibilistic algorithms that produce typicality matrices instead of fuzzy membership matrices. Even more, the variants of FCM and PCM such as Possibilistic Fuzzy C-means (PFCM) and Fuzzy Possibilistic C-means (FPCM) simultaneously result with probabilistic and possibilistic membership degrees. Thus, some kind of validity indices are needed for working with both of these results. For this purpose, a few extended and generalized validity indices has been proposed in recent years. In this paper, the performances of these indices were examined for validating the clustering results from Unsupervised Possibilistic Fuzzy Clustering (UPFC), FCM and PCM algorithms. The findings showed that generalized versions of the fuzzy validity indices based on normalization of typicality degrees can be successfully used to validate the results from PCM, UPFC and the variants of FCM and PCM.","PeriodicalId":111721,"journal":{"name":"2017 International Artificial Intelligence and Data Processing Symposium (IDAP)","volume":"164 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122468076","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 : 2017-09-01DOI: 10.1109/IDAP.2017.8090232
E. Kose, Gülçin Mühürcü, Aydin Muhurcu, Buse Sevim
In this work, it is aimed to control the output of Buck Converter which steps down a DC voltage level. Discrete Time PI Algorithm is chosen as control algorithm. The controller parameters of Kp and Ki is calculated by means of the optimization process to increase the efficiency of power transmission of Buck Converter instead of classical methods like Pole Placement Method. It is chosen Shuffled Frog Leaping Algorithm (SFLA) which is an iterative algorithm as optimization algorithm. The results obtained from simultaneous executions of control process is discussed by simulating in Matlab-Simulink.
{"title":"SFLA based PI parameter optimization for optimal controlling of a Buck converter's voltage","authors":"E. Kose, Gülçin Mühürcü, Aydin Muhurcu, Buse Sevim","doi":"10.1109/IDAP.2017.8090232","DOIUrl":"https://doi.org/10.1109/IDAP.2017.8090232","url":null,"abstract":"In this work, it is aimed to control the output of Buck Converter which steps down a DC voltage level. Discrete Time PI Algorithm is chosen as control algorithm. The controller parameters of Kp and Ki is calculated by means of the optimization process to increase the efficiency of power transmission of Buck Converter instead of classical methods like Pole Placement Method. It is chosen Shuffled Frog Leaping Algorithm (SFLA) which is an iterative algorithm as optimization algorithm. The results obtained from simultaneous executions of control process is discussed by simulating in Matlab-Simulink.","PeriodicalId":111721,"journal":{"name":"2017 International Artificial Intelligence and Data Processing Symposium (IDAP)","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122708542","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 : 2017-09-01DOI: 10.1109/IDAP.2017.8090194
Emine Cengil, A. Cinar, Zafer Güler
Deep learning obtains successful results in solving many machine learning problems. In this study, image classification process is performed by using Convolutional Neural Network (CNN) which is the most used architecture of deep learning. Image classification is used in a lot of basic field like medicine, education and security. Conditions that correct classification has vital importance may be especially in medicine field. Therefore, improved methods are needed in this issue. Although several algorithms for image classification have been developed over the years, they have not been used with the discovery of Convolutional Neural Networks. Convolutional Neural Networks provide better results than existing methods in the literature due to advantages such as processing by extracting hidden features, allowing parallel processing thanks to parallel structure, and real time operation. Furthermore, we use Convolutional Neural Networks in the proposed method. In this study, the image classification process is performed by using like a LeNet network model. The caffe library, which is often used for deep learning, is utilized. Our method is trained and tested with images of cats and dogs taken from the kaggle dataset. 10.000 tagged data is used for training and 5.000 unlabeled data is used for testing. Owing to Convolutional Neural Networks allow parallel processing, GPU technology has been used. In our method is used GPU technology and classification is evaluated with acceptable accuracy rate and speed performance.
{"title":"A GPU-based convolutional neural network approach for image classification","authors":"Emine Cengil, A. Cinar, Zafer Güler","doi":"10.1109/IDAP.2017.8090194","DOIUrl":"https://doi.org/10.1109/IDAP.2017.8090194","url":null,"abstract":"Deep learning obtains successful results in solving many machine learning problems. In this study, image classification process is performed by using Convolutional Neural Network (CNN) which is the most used architecture of deep learning. Image classification is used in a lot of basic field like medicine, education and security. Conditions that correct classification has vital importance may be especially in medicine field. Therefore, improved methods are needed in this issue. Although several algorithms for image classification have been developed over the years, they have not been used with the discovery of Convolutional Neural Networks. Convolutional Neural Networks provide better results than existing methods in the literature due to advantages such as processing by extracting hidden features, allowing parallel processing thanks to parallel structure, and real time operation. Furthermore, we use Convolutional Neural Networks in the proposed method. In this study, the image classification process is performed by using like a LeNet network model. The caffe library, which is often used for deep learning, is utilized. Our method is trained and tested with images of cats and dogs taken from the kaggle dataset. 10.000 tagged data is used for training and 5.000 unlabeled data is used for testing. Owing to Convolutional Neural Networks allow parallel processing, GPU technology has been used. In our method is used GPU technology and classification is evaluated with acceptable accuracy rate and speed performance.","PeriodicalId":111721,"journal":{"name":"2017 International Artificial Intelligence and Data Processing Symposium (IDAP)","volume":"130 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122901391","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：