Gazi University Journal of Science Part A: Engineering and Innovation最新文献

Diagnosis of Pneumonia from Chest X-ray Images with Vision Transformer Approach 利用视觉变换器方法从胸部 X 光图像诊断肺炎

Gazi University Journal of Science Part A: Engineering and Innovation Pub Date : 2024-06-14 DOI: 10.54287/gujsa.1464311

Emrah Aslan

{"title":"Diagnosis of Pneumonia from Chest X-ray Images with Vision Transformer Approach","authors":"Emrah Aslan","doi":"10.54287/gujsa.1464311","DOIUrl":"https://doi.org/10.54287/gujsa.1464311","url":null,"abstract":"People can get pneumonia, a dangerous infectious disease, at any time in their lives. Severe cases of pneumonia can be fatal. A doctor would usually examine chest x-rays to diagnose pneumonia. In this work, a pneumonia diagnosis system was developed using publicly available chest x-ray images. Vision Transformer (ViT) and other deep learning models were used to extract features from these images. Vision Transformer (ViT) is an attention-based model used for image processing and understanding as an alternative to the convolutional neural networks traditionally used for this purpose. ViT consists of a series of attention layers, where each attention layer models the relationships between input pixels to represent an image. These relationships are determined by a set of attention heads and then fed into a classifier. ViT performs effectively in a variety of visual tasks, especially when trained on large datasets. The study shows that the ViT model's classification procedure has a high success rate of 95.67%. These results highlight how deep learning models can be used to quickly and accurately diagnose dangerous diseases such as pneumonia in its early stages. The study also shows that the ViT model outperforms current approaches in the biomedical field.","PeriodicalId":510341,"journal":{"name":"Gazi University Journal of Science Part A: Engineering and Innovation","volume":"11 4","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141339053","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}

引用次数: 0

IoT-Based Energy Consumption Prediction Using Transformers 基于物联网的变压器能耗预测

Gazi University Journal of Science Part A: Engineering and Innovation Pub Date : 2024-06-05 DOI: 10.54287/gujsa.1438011

Abdul Amir Alıoghlı, Feyza Yıldırım Okay

{"title":"IoT-Based Energy Consumption Prediction Using Transformers","authors":"Abdul Amir Alıoghlı, Feyza Yıldırım Okay","doi":"10.54287/gujsa.1438011","DOIUrl":"https://doi.org/10.54287/gujsa.1438011","url":null,"abstract":"With the advancement of various IoT-based systems, the amount of data is steadily increasing. The increase of data on a daily basis is essential for decision-makers to assess current situations and formulate future policies. Among the various types of data, time-series data presents a challenging relationship between current and future dependencies. Time-series prediction aims to forecast future values of target variables by leveraging insights gained from past data points. Recent advancements in deep learning-based algorithms have surpassed traditional machine learning-based algorithms for time-series in IoT systems. In this study, we employ Enc&Dec Transformer, the latest advancements in neural networks for time-series prediction problems. The obtained results were compared with Encoder-only and Decoder-only Transformer blocks as well as well-known recurrent based algorithms, including 1D-CNN, RNN, LSTM, and GRU. To validate our approach, we utilize three different univariate time-series datasets collected on an hourly basis, focusing on energy consumption within IoT systems. Our results demonstrate that our proposed Transformer model outperforms its counterparts, achieving a minimum Mean Squared Error (MSE) of 0.020 on small, 0.008 on medium, and 0.006 on large-sized datasets.","PeriodicalId":510341,"journal":{"name":"Gazi University Journal of Science Part A: Engineering and Innovation","volume":"50 25","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141384000","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}

引用次数: 0

Perturbation Augmentation for Adversarial Training with Diverse Attacks 针对多样化攻击的对抗训练的惰性增强

Gazi University Journal of Science Part A: Engineering and Innovation Pub Date : 2024-06-04 DOI: 10.54287/gujsa.1458880

Duygu Serbes, Inci M. Baytas

{"title":"Perturbation Augmentation for Adversarial Training with Diverse Attacks","authors":"Duygu Serbes, Inci M. Baytas","doi":"10.54287/gujsa.1458880","DOIUrl":"https://doi.org/10.54287/gujsa.1458880","url":null,"abstract":"Adversarial Training (AT) aims to alleviate the vulnerability of deep neural networks to adversarial perturbations. However, the AT techniques struggle to maintain the performance on natural samples while improving the deep model’s robustness. The absence of perturbation diversity in generated during the adversarial training degrades the generalizability of the robust models, causing overfitting to particular perturbations and a decrease in natural performance. This study proposes an adversarial training framework that augments adversarial directions from a single-step attack to address the trade-off between robustness and generalization. Inspired by feature scattering adversarial training, the proposed framework computes a principal adversarial direction with a single-step attack that finds a perturbation disrupting the inter-sample relationships in the mini-batch during adversarial training. The principal direction obtained at each iteration is augmented by sampling new adversarial directions within a region spanning 45 degrees from the principal adversarial direction. The proposed adversarial training approach does not require extra backpropagation steps in adversarial direction augmentation. Therefore, generalization of the robust model is improved without posing an additional burden on the feature scattering adversarial training. Experiments on CIFAR-10, CIFAR-100, SVHN, Tiny-ImageNet, and The German Traffic Sign Recognition Benchmark consistently improve the accuracy on adversarial with an almost pristine natural performance.","PeriodicalId":510341,"journal":{"name":"Gazi University Journal of Science Part A: Engineering and Innovation","volume":"8 11","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141267381","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}

引用次数: 0

A No-Code Automated Machine Learning Platform for the Energy Sector 能源行业无代码自动化机器学习平台

Gazi University Journal of Science Part A: Engineering and Innovation Pub Date : 2024-06-04 DOI: 10.54287/gujsa.1473782

Ezgi Avci

{"title":"A No-Code Automated Machine Learning Platform for the Energy Sector","authors":"Ezgi Avci","doi":"10.54287/gujsa.1473782","DOIUrl":"https://doi.org/10.54287/gujsa.1473782","url":null,"abstract":"The energy sector is increasingly leveraging machine learning (ML) to enhance efficiency, sustainability, and decision-making processes. However, the complexity of ML model development and deployment poses significant challenges, particularly in adapting to diverse data environments and operational requirements. This paper presents the development of a No-Code Automated ML (Auto-ML) platform specifically tailored for the energy industry. We introduce a framework that automates critical steps in the ML pipeline, including data preprocessing, feature engineering, model selection, and hyperparameter tuning, while integrating domain-specific knowledge to handle unique industry challenges such as fluctuating energy demands and regulatory compliance. The platform employs a modular architecture that allows for customization and scalability, facilitating its adaptation across various energy sub-sectors, including renewable energy, oil and gas, and power distribution. The results underline the platform's potential to democratize advanced analytical capabilities within the energy industry, making sophisticated data-driven insights accessible even to non-expert users. This paper contributes to the field by detailing the architectural considerations, implementation challenges, and practical impacts of Auto-ML in a sector critical to global infrastructure and economic stability.","PeriodicalId":510341,"journal":{"name":"Gazi University Journal of Science Part A: Engineering and Innovation","volume":"3 3","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141266019","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}

引用次数: 0

Zeolite A Thin Film Growth on Silica Optical Fibers 在二氧化硅光纤上生长沸石 A 薄膜

Gazi University Journal of Science Part A: Engineering and Innovation Pub Date : 2024-02-15 DOI: 10.54287/gujsa.1409513

Sezin GALİOĞLU ÖZALTUĞ

{"title":"Zeolite A Thin Film Growth on Silica Optical Fibers","authors":"Sezin GALİOĞLU ÖZALTUĞ","doi":"10.54287/gujsa.1409513","DOIUrl":"https://doi.org/10.54287/gujsa.1409513","url":null,"abstract":"Thin films of zeolite A (Linde type A, LTA) were fabricated on silica optical fibers using dip coating method, followed by a secondary growth process. Zeolite A seed crystals, with crystal sizes ranging from 100 to 250 nm, were synthesized through a hydrothermal method and coated onto the silica optical fiber using the dip coating technique. The one and two-time dip coating resulted in island-like growth, while three-time dip coating process led to the intergrowth of seed crystals, resulting in nearly continuous films on the optical fiber. The seed crystals, initially deposited as a monolayer, were grown by secondary growth method, leading to film thicknesses of 255.3 nm ± 10.1 nm (average ± standard deviation). Upon extending the secondary growth reaction time from 2 hours to 4 hours, the film thickness increased from 255.3 nm ± 10.1 nm to 649.6 nm ± 28.1 nm. Epitaxial growth was identified as the mechanism through which the seed crystals grow. The nanoporous zeolite A films formed on the optical fiber exhibit potential applications as gas sensors. These sensors function by detecting changes in optical reflection caused by the differential adsorption of gases.","PeriodicalId":510341,"journal":{"name":"Gazi University Journal of Science Part A: Engineering and Innovation","volume":"14 6","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-02-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139774038","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}

引用次数: 0

Zeolite A Thin Film Growth on Silica Optical Fibers 在二氧化硅光纤上生长沸石 A 薄膜

Gazi University Journal of Science Part A: Engineering and Innovation Pub Date : 2024-02-15 DOI: 10.54287/gujsa.1409513

Sezin GALİOĞLU ÖZALTUĞ

{"title":"Zeolite A Thin Film Growth on Silica Optical Fibers","authors":"Sezin GALİOĞLU ÖZALTUĞ","doi":"10.54287/gujsa.1409513","DOIUrl":"https://doi.org/10.54287/gujsa.1409513","url":null,"abstract":"Thin films of zeolite A (Linde type A, LTA) were fabricated on silica optical fibers using dip coating method, followed by a secondary growth process. Zeolite A seed crystals, with crystal sizes ranging from 100 to 250 nm, were synthesized through a hydrothermal method and coated onto the silica optical fiber using the dip coating technique. The one and two-time dip coating resulted in island-like growth, while three-time dip coating process led to the intergrowth of seed crystals, resulting in nearly continuous films on the optical fiber. The seed crystals, initially deposited as a monolayer, were grown by secondary growth method, leading to film thicknesses of 255.3 nm ± 10.1 nm (average ± standard deviation). Upon extending the secondary growth reaction time from 2 hours to 4 hours, the film thickness increased from 255.3 nm ± 10.1 nm to 649.6 nm ± 28.1 nm. Epitaxial growth was identified as the mechanism through which the seed crystals grow. The nanoporous zeolite A films formed on the optical fiber exhibit potential applications as gas sensors. These sensors function by detecting changes in optical reflection caused by the differential adsorption of gases.","PeriodicalId":510341,"journal":{"name":"Gazi University Journal of Science Part A: Engineering and Innovation","volume":"230 ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-02-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139833776","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}

引用次数: 0

A COMPARATIVE STUDY FOR PRIVACY-AWARE RECOMMENDATION SYSTEMS 隐私感知推荐系统的比较研究

Gazi University Journal of Science Part A: Engineering and Innovation Pub Date : 2024-02-09 DOI: 10.54287/gujsa.1393692

Yavuz Canbay, Anıl Utku

{"title":"A COMPARATIVE STUDY FOR PRIVACY-AWARE RECOMMENDATION SYSTEMS","authors":"Yavuz Canbay, Anıl Utku","doi":"10.54287/gujsa.1393692","DOIUrl":"https://doi.org/10.54287/gujsa.1393692","url":null,"abstract":"Recommendation systems are sophisticated processes for information filtering designed to offer users tailored recommendations based on their preferences and interests. Users need help to choose between options as the amount of information on the web grows. As a result, it is critical to deliver personalized recommendations to consumers to promote user loyalty and satisfaction. Because recommender systems use sensitive user information such as ratings, comments, likes, and dislikes, this information can be leaked if no privacy measures are taken. As a result, we presented a comparison of privacy-aware recommendation systems in this paper. Two experiments are carried out. In the first experiment, we examined collaborative filtering algorithms on perturbed ratings and then compared hybrid, collaborative, and content-based algorithms on perturbed ratings in the second experiment. According to the results, the Singular Value Decomposition++ (SVDpp) algorithm presented the lowest Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) values for epsilon 100, with 0.8889 and 0.6822, respectively. Furthermore, for epsilon 100, the hybrid filtering technique had the lowest RMSE and MAE rates of 0.90664 and 0.69813, respectively.","PeriodicalId":510341,"journal":{"name":"Gazi University Journal of Science Part A: Engineering and Innovation","volume":"358 5","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-02-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139848335","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}

引用次数: 0

A COMPARATIVE STUDY FOR PRIVACY-AWARE RECOMMENDATION SYSTEMS 隐私感知推荐系统的比较研究

Gazi University Journal of Science Part A: Engineering and Innovation Pub Date : 2024-02-09 DOI: 10.54287/gujsa.1393692

Yavuz Canbay, Anıl Utku

{"title":"A COMPARATIVE STUDY FOR PRIVACY-AWARE RECOMMENDATION SYSTEMS","authors":"Yavuz Canbay, Anıl Utku","doi":"10.54287/gujsa.1393692","DOIUrl":"https://doi.org/10.54287/gujsa.1393692","url":null,"abstract":"Recommendation systems are sophisticated processes for information filtering designed to offer users tailored recommendations based on their preferences and interests. Users need help to choose between options as the amount of information on the web grows. As a result, it is critical to deliver personalized recommendations to consumers to promote user loyalty and satisfaction. Because recommender systems use sensitive user information such as ratings, comments, likes, and dislikes, this information can be leaked if no privacy measures are taken. As a result, we presented a comparison of privacy-aware recommendation systems in this paper. Two experiments are carried out. In the first experiment, we examined collaborative filtering algorithms on perturbed ratings and then compared hybrid, collaborative, and content-based algorithms on perturbed ratings in the second experiment. According to the results, the Singular Value Decomposition++ (SVDpp) algorithm presented the lowest Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) values for epsilon 100, with 0.8889 and 0.6822, respectively. Furthermore, for epsilon 100, the hybrid filtering technique had the lowest RMSE and MAE rates of 0.90664 and 0.69813, respectively.","PeriodicalId":510341,"journal":{"name":"Gazi University Journal of Science Part A: Engineering and Innovation","volume":" 32","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-02-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139788360","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}

引用次数: 0

Kinetics of Nonenzymatic Browning Reactions in Pumpkin Puree During Storage 南瓜泥在储存过程中的非酶褐变反应动力学

Gazi University Journal of Science Part A: Engineering and Innovation Pub Date : 2024-02-06 DOI: 10.54287/gujsa.1400745

Feryal Karadeni̇z, Betül Işik, Osman Aslanali̇, Soner Kaya, Fatma Mi̇di̇lli̇

{"title":"Kinetics of Nonenzymatic Browning Reactions in Pumpkin Puree During Storage","authors":"Feryal Karadeni̇z, Betül Işik, Osman Aslanali̇, Soner Kaya, Fatma Mi̇di̇lli̇","doi":"10.54287/gujsa.1400745","DOIUrl":"https://doi.org/10.54287/gujsa.1400745","url":null,"abstract":"Pumpkins (Cucurbita moschata) are a great source of essential nutrients counting vitamins, minerals, carotenoids, and dietary fiber. Due to their healthy composition, it draws consumer attention. Pumpkin puree is used for preparation of baby foods, soups, flour, jam, jellies, and desserts. To assess the main quality parameters of the foods such as color and hydroxymethylfurfural (HMF) formation, CIE-L*a*b* color changes and HMF formation which is also an indicator of Maillard browning reactions were evaluated storing the pumpkin puree at 27°, 37° and 47°C for 17 weeks. Kinetic parameters for HMF formation and color changes were calculated. The findings showed that HMF quantity linearly increased with the temperature and storage duration following the reaction model of zero order. The values of b* coordinate, Lightness, Chroma, and hue lessened linearly as a* coordinate values increased linearly fitting zero order reaction kinetic. By means of Arrhenius equation, the temperature dependency of the rate constant of color variation was demonstrated and the values of the activation energy (kj mol-1) were calculated as 76.15, 30.60, 46.08, 28.44, 27.61 and 38.32 for HMF formation, Lightness, a* coordinate, b* coordinate, Chroma, and hue, respectively.","PeriodicalId":510341,"journal":{"name":"Gazi University Journal of Science Part A: Engineering and Innovation","volume":"115 18","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139802055","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}

引用次数: 0

General Purpose and Focused Invention, Market Value, and Productivity 通用和重点发明、市场价值和生产力

Gazi University Journal of Science Part A: Engineering and Innovation Pub Date : 2024-02-06 DOI: 10.54287/gujsa.1357879

Burak Dindaroğlu

{"title":"General Purpose and Focused Invention, Market Value, and Productivity","authors":"Burak Dindaroğlu","doi":"10.54287/gujsa.1357879","DOIUrl":"https://doi.org/10.54287/gujsa.1357879","url":null,"abstract":"I study returns to general purpose and focused invention at the firm level for a panel of manufacturing firms in the U.S. for the period 1976-1995, by studying their relationships with market value and Total Factor Productivity. I construct stocks of patents that lie at the two relevant tails of the distribution of the generality index (Trajtenberg et al., 1997; Henderson et al, 1998) to measure general purpose and focused invention at the firm level. In line with expectations, there is a market value premium to focused invention, and a productivity premium to general purpose invention. Estimates for the value of focus indicate that moving a single patent from the upper tail of the generality distribution to the lower tail would increase market value by .24×q million 1992 dollars on average, where q is Tobin’s q. The firm with the average general patent stock would gain 6.7×q million in market value if all its patents at the highest quartile of the generality distribution were moved to the lowest. In terms of the value of general purposeness, moving all its focused patents to the general category increases Total Factor Productivity by 2.3% to 2.8%, and five-year productivity growth by 3.9% to 5.2%, for the average firm. A potential implication is that corporate basic research is associated with significant long-term benefits in terms of productivity growth.","PeriodicalId":510341,"journal":{"name":"Gazi University Journal of Science Part A: Engineering and Innovation","volume":"387 4-6","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139860215","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}

引用次数: 0