Işık Üniversitesi Kurumsal Akademik Bellek
Işık Üniversitesi tarafından doğrudan ve dolaylı olarak yayınlanan; kitap, makale, tez, bildiri, rapor, araştırma verisi gibi tüm akademik kaynakları uluslararası standartlarda dijital ortamda depolar, Üniversitenin akademik performansını izlemeye aracılık eder, kaynakları uzun süreli saklar ve telif haklarına uygun olarak Açık Erişime sunar.
Güncel Gönderiler
Modeling the effects of soil improvement on train induced random ground-borne vibrations
(Isik University, 2025-05-01) Bayındır, Cihan; Kesten, Ali Sercan; Etminan, Ehsan
Ground-borne vibrations by railway trains are generated at the rail-wheel interface due to the passage of wheels and due to irregularities of wheels and tracks. These vibrations need to be predicted and controlled during the design and service of the railway for the safety and serviceability of the railway to avoid possible vibrationinduced problems such as settlement and differential settlement due to their compaction effect, liquefaction, and discomfort of people. While such railway vibrations are modeled by different techniques, only a few studies do exist to analyze them in the case of soilimproved conditions. In this study, we propose a mathematical framework to study the effects of soil improvement on the ground-borne vibrations induced by railway trains. We use an experimentally calibrated model that utilizes the evolutionary random process approach to model the time-varying transfer functions between the axles of the train and the fixed observation point. The railway is modeled as a Winkler foundation with rail pads and corresponding transfer functions are used. The target area of this study is the Emin¨on¨u-Alibeyk¨oy Tramway Line in ˙Istanbul, which is under construction. Due to poor soil conditions at the specific stations along the proposed tramway route, soil improvement by the application of geo-synthetics is performed at the site and taken into account in our model. The improvement in soil conditions is modeled as increased vertical soil stiffness in the Winkler foundation of the evolutionary random process model. To model the various tramway loading conditions, both the 5-axle and 6-axle tramway configurations with non-uniform axle spacing are considered. We show that by increasing the vertical soil stiffness ksb, the vibration velocity and acceleration levels can be reduced significantly. By implementing the model proposed, we present the reduction of the vibration velocity and acceleration levels as the functions of soil improvement parameters and discuss our findings and the applicability of the model.
Enhancing real estate listings through image classification and enhancement: a comparative study
(Multidisciplinary Digital Publishing Institute (MDPI), 2025-05-22) Küp, Eyüp Tolunay; Sözdinler, Melih; Işık, Ali Hakan; Doksanbir, Yalçın; Akpınar, Gökhan
We extended real estate property listings on the online prop-tech platform. On the platform, the images were classified into the specified classes according to quality criteria. The necessary interventions were made by measuring the platform’s appropriateness level and increasing the advertisements’ visual appeal. A dataset of 3000 labeled images was utilized to compare different image classification models, including convolutional neural networks (CNNs), VGG16, residual networks (ResNets), and the LLaVA large language model (LLM). Each model’s performance and benchmark results were measured to identify the most effective method. In addition, the classification pipeline was expanded using image enhancement with contrastive unsupervised representation learning (CURL). This method assessed the impact of improved image quality on classification accuracy and the overall attractiveness of property listings. For each classification model, the performance was evaluated in binary conditions, with and without the application of CURL. The results showed that applying image enhancement with CURL enhances image quality and improves classification performance, particularly in models such as CNN and ResNet. The study results enable a better visual representation of real estate properties, resulting in higher-quality and engaging user listings. They also underscore the importance of combining advanced image processing techniques with classification models to optimize image presentation and categorization in the real estate industry. The extended platform offers information on the role of machine learning models and image enhancement methods in technology for the real estate industry. Also, an alternative solution that can be integrated into intelligent listing systems is proposed in this study to improve user experience and information accuracy. The platform proves that artificial intelligence and machine learning can be integrated for cloud-distributed services, paving the way for future innovations in the real estate sector and intelligent marketplace platforms.
Enhancing mechanical performance of FDM-printed ABS parts through annealing optimization
(John Wiley and Sons Inc, 2025-06-18) Kösemen, Elifnur; Bakkal, Mustafa; Kuzu, Ali Taner
This study examines the impact of annealing on the mechanical properties of acrylonitrile butadiene styrene (ABS) parts produced using fused deposition modeling (FDM). The research investigates how different annealing temperatures (90°C, 105°C, and 120°C), production orientations (upright, on edge, and flat), and infill patterns influence hardness, tensile strength, and impact resistance. Experiments were conducted using a Stratasys F370 printer, and samples were tested following ISO standards for mechanical performance. Results indicated that annealing at 90°C and 105°C generally improved hardness, tensile strength, and impact resistance, particularly for upright and on-edge orientations. However, annealing at 120°C led to a decrease in these properties, likely due to microstructural changes observed through scanning electron microscopy (SEM) and differential scanning calorimetry (DSC) analysis. The study highlights the importance of optimizing production parameters and annealing conditions to achieve desired mechanical properties in FDM-printed ABS parts. These findings may inform post-processing strategies for enhancing the reliability and performance of additive manufactured components, particularly for applications in industries utilizing ABS materials for customized and prototype parts.
Adaptive locally connected recurrent unit (ALCRU)
(Springer Science and Business Media Deutschland GmbH, 2025-07-03) Özçelik, Şuayb Talha; Tek, Faik Boray
Research has shown that adaptive locally connected neurons outperform their fully connected (dense) counterparts, motivating this study on the development of the Adaptive Locally Connected Recurrent Unit (ALCRU). ALCRU modifies the Simple Recurrent Neuron Model (SimpleRNN) by incorporating spatial coordinate spaces for input and hidden state vectors, facilitating the learning of parametric local receptive fields. These modifications add four trainable parameters per neuron, resulting in a minor increase in computational complexity. ALCRU is implemented using standard frameworks and trained with back-propagation-based optimizers. We evaluate the performance of ALCRU using diverse benchmark datasets, including IMDb for sentiment analysis, AdditionRNN for sequence modelling, and the Weather dataset for time-series forecasting. Results show that ALCRU achieves accuracy and loss metrics comparable to GRU and LSTM while consistently outperforming SimpleRNN. In particular, experiments with longer sequence lengths on AdditionRNN and increased input dimensions on IMDb highlight ALCRU’s superior scalability and efficiency in processing complex data sequences. In terms of computational efficiency, ALCRU demonstrates a considerable speed advantage over gated models like LSTM and GRU, though it is slower than SimpleRNN. These findings suggest that adaptive local connectivity enhances both the accuracy and efficiency of recurrent neural networks, offering a promising alternative to standard architectures.
ANN activation function estimators for homomorphic encrypted inference
(Institute of Electrical and Electronics Engineers Inc., 2025-06-13) Harb, Mhd Raja Abou; Çeliktaş, Barış
Homomorphic Encryption (HE) enables secure computations on encrypted data, facilitating machine learning inference in sensitive environments such as healthcare and finance. However, efficiently handling non-linear activation functions, specifically Sigmoid and Tanh, remains a significant computational challenge for encrypted inference using Artificial Neural Networks (ANNs). This study introduces a lightweight, ANN-based estimator designed to accurately approximate activation functions under homomorphic encryption. Unlike traditional polynomial and piecewise linear approximations, the proposed ANN estimators achieve superior accuracy with lower computational overhead associated with bootstrapping or high-degree polynomial techniques. These estimators are trained on plaintext data and seamlessly integrated into encrypted inference pipelines, significantly outperforming conventional methods. Experimental evaluations demonstrate notable improvements, with ANN estimators enhancing accuracy by approximately 2% for Sigmoid and up to 73% for Tanh functions, improving F1-scores by approximately 2% for Sigmoid and up to 88% for Tanh, and markedly reducing Mean Square Error (MSE) by up to 96% compared to polynomial approximations. The ANN estimator achieves an accuracy of 97.70% and an AUC of 0.9997 when integrated into a CNN architecture on the MNIST dataset, and an accuracy of 85.25% with an AUC of 0.9459 on the UCI Heart Disease dataset during ciphertext inference. These results underscore the estimator’s practical effectiveness and computational feasibility, making it suitable for secure and efficient ANN inference in encrypted environments.
