Browsing by Author "Oztoprak, Kasim"
Now showing 1 - 2 of 2
- Results Per Page
- Sort Options
Conference Object Citation Count: 4Protocol and connectivity based overlay level capacity calculation of P2P networks(Ieee Computer Soc, 2006) Kılıç, Hürevren; Kilic, Hurevren; Computer EngineeringIn this paper, we proposed a metric for P2P networks based on Shannon's L-channel capacity calculation idea. The metric calculates the maximum rate of information (in bits per second) that can be transmitted over P2P network (a.k.a. combinatorial capacity) caused by protocol and overlay-level connectivity. We suggest P2P systems to be modeled as a discrete noiseless channel on which the protocol together with dynamically changing overlay-level instant connectivity topology defines a Shannon Language. In experimental works, we applied the metric first to the Gnutella 0.6 protocol for which message traffic explosion is a known problem and then to its time-based clustering version. The obtained results are compared with other two known metrics' namely, number of query hits and unit query-hit response time, results and potential correlations among them are discussed.Article Citation Count: 0Two-Stage Feature Generator for Handwritten Digit Classification(Mdpi, 2023) Tora, Hakan; Tora, Hakan; Oztoprak, Kasim; Butun, Ismail; Airframe and Powerplant MaintenanceIn this paper, a novel feature generator framework is proposed for handwritten digit classification. The proposed framework includes a two-stage cascaded feature generator. The first stage is based on principal component analysis (PCA), which generates projected data on principal components as features. The second one is constructed by a partially trained neural network (PTNN), which uses projected data as inputs and generates hidden layer outputs as features. The features obtained from the PCA and PTNN-based feature generator are tested on the MNIST and USPS datasets designed for handwritten digit sets. Minimum distance classifier (MDC) and support vector machine (SVM) methods are exploited as classifiers for the obtained features in association with this framework. The performance evaluation results show that the proposed framework outperforms the state-of-the-art techniques and achieves accuracies of 99.9815% and 99.9863% on the MNIST and USPS datasets, respectively. The results also show that the proposed framework achieves almost perfect accuracies, even with significantly small training data sizes.