Bilgisayar Mühendisliği Bölümü Koleksiyonu
Permanent URI for this collectionhttps://hdl.handle.net/20.500.11779/1940
Browse
18 results
Search Results
Article Citation - WoS: 3Citation - Scopus: 3Array Bp-Xor Codes for Hierarchically Distributed Matrix Multiplication(IEEE, 2022) Arslan, Şuayb Şefik; Arslan, Şefik Şuayb; 02.02. Department of Computer Engineering; 02. Faculty of Engineering; 01. MEF UniversityA novel fault-tolerant computation technique based on array Belief Propagation (BP)-decodable XOR (BP-XOR) codes is proposed for distributed matrix-matrix multiplication. The proposed scheme is shown to be configurable and suited for modern hierarchical compute architectures such as Graphical Processing Units (GPUs) equipped with multiple nodes, whereby each has many small independent processing units with increased core-to-core communications. The proposed scheme is shown to outperform a few of the well–known earlier strategies in terms of total end-to-end execution time while in presence of slow nodes, called stragglers. This performance advantage is due to the careful design of array codes which distributes the encoding operation over the cluster (slave) nodes at the expense of increased master-slave communication. An interesting trade-off between end-to-end latency and total communication cost is precisely described. In addition, to be able to address an identified problem of scaling stragglers, an asymptotic version of array BP-XOR codes based on projection geometry is proposed at the expense of some computation overhead. A thorough latency analysis is conducted for all schemes to demonstrate that the proposed scheme achieves order-optimal computation in both the sublinear as well as the linear regimes in the size of the computed product from an end-to-end delay perspective.Article Citation - WoS: 4Citation - Scopus: 6Data Repair-Efficient Fault Tolerance for Cellular Networks Using Ldpc Codes(IEEE, 2022) Haytaoglu, Elif; Arslan, Şefik Şuayb; Arslan, Şuayb Şefik; 02.02. Department of Computer Engineering; 02. Faculty of Engineering; 01. MEF UniversityThe base station-mobile device communication traffic has dramatically increased recently due to mobile data, which in turn heavily overloaded the underlying infrastructure. To decrease Base Station (BS) interaction, intra-cell communication between local devices, known as Device-to-Device, is utilized for distributed data caching. Nevertheless, due to the continuous departure of existing nodes and the arrival of newcomers, the missing cached data may lead to permanent data loss. In this study, we propose and analyze a class of LDPC codes for distributed data caching in cellular networks. Contrary to traditional distributed storage, a novel repair algorithm for LDPC codes is proposed which is designed to exploit the minimal direct BS communication. To assess the versatility of LDPC codes and establish performance comparisons to classic coding techniques, novel theoretical and experimental evaluations are derived. Essentially, the theoretical/numerical results for repair bandwidth cost in presence of BS are presented in a distributed caching setting. Accordingly, when the gap between the cost of downloading a symbol from BS and from other local network nodes is not dramatically high, we demonstrate that LDPC codes can be considered as a viable fault-tolerance alternative in cellular systems with caching capabilities for both low and high code rates.Conference Object Citation - WoS: 3Citation - Scopus: 3Data Repair in Bs-Assisted Distributed Data Caching(IEEE, 2020) Arslan, Şefik Şuayb; Haytaoğlu, Elif; Arslan, Şuayb Şefik; 02.02. Department of Computer Engineering; 02. Faculty of Engineering; 01. MEF UniversityIn this paper, centralized and independent repair approaches based on device-to-device communication for the repair of the lost nodes have been investigated in a cellular network where distributed caching is applied whose fault tolerance is provided by erasure codes. The caching mechanisms based on Reed-Solomon codes and minimum bandwidth regenerating codes are adopted. The proposed approaches are analyzed in a simulation environment in terms of base station utilization load during the repair process. Based on the intuitive assumption that the base station is usually more costly than device-to-device communication, the centralized repair approach demonstrates a better performance than the independent repair approaches on the number of symbols retrieved from the base station. On the other hand, the centralized approach has not achieved a dramatic reduction in the number of symbols downloaded from the other devices.Conference Object Citation - WoS: 1Citation - Scopus: 1Adaptive Boosting of Dnn Ensembles for Brain-Computer Interface Spellers(IEEE, 2021) Çatak, Yiğit; Arslan, Şefik Şuayb; Özkan, Hüseyin; Güney, Osman Berke; Koç, Emirhan; Arslan, Şuayb Şefik; 02.02. Department of Computer Engineering; 02. Faculty of Engineering; 01. MEF UniversitySteady-state visual evoked potentials (SSVEP) are commonly used in brain computer interface (BCI) applications such as spelling systems, due to their advantages over other paradigms. In this study, we develop a method for SSVEP-based BCI speller systems, using a known deep neural network (DNN), which includes transfer and ensemble learning techniques. We test performance of our method on publicly available benchmark and BETA datasets with leave-one-subject-out procedure. Our method consists of two stages. In the first stage, a global DNN is trained using data from all subjects except one subject that is excluded for testing. In the second stage, the global model is fine-tuned to each subject whose data are used in the training. Combining the responses of trained DNNs with different weights for each test subject, rather than an equal weight, provide better performance as brain signals may differ significantly between individuals. To this end, weights of DNNs are learnt with SAMME algorithm with using data belonging to the test subject. Our method significantly outperforms canonical correlation analysis (CCA) and filter bank canonical correlation analysis (FBCCA) methods.Conference Object Citation - WoS: 2Citation - Scopus: 1Average Bandwidth-Cost Vs. Storage Trade-Off for Bs-Assisted Distributed Storage Networks(IEEE, 2021) Tengiz, Ayse Ceyda; Arslan, Şefik Şuayb; Pusane, Ali Emre; Arslan, Şuayb Şefik; Pourmandi, Massoud; 02.02. Department of Computer Engineering; 02. Faculty of Engineering; 01. MEF UniversityIn this study, we consider a hierarchically structured base station (BS)-assisted cellular system equipped with a backend distributed data storage in which nodes randomly arrive and depart the cell. We numerically motivate and characterize the fundamental trade-off between the average repair bandwidth cost versus storage space where BS communication cost (higher than that of local) and link capacity constraints exist while the number of failed nodes can vary dynamically. We establish the capacity region that is most relevant to 5G and beyond networks, which are layered by design. We hope that this study shall motivate novel regeneration code constructions that will be able to achieve the presented limits.Conference Object Fault-Tolerant Strassen-Like Matrix Multiplication(IEEE, 2020) Arslan, Şuayb Şefik; Arslan, Şefik Şuayb; Güney, Osman B.; Oblokulov, Muhtasham; 02.02. Department of Computer Engineering; 02. Faculty of Engineering; 01. MEF UniversityIn this study, we propose a simple method for fault-tolerant Strassen-like matrix multiplications. The proposed method is based on using two distinct Strassen-like algorithms instead of replicating a given one. We have realized that using two different algorithms, new check relations arise resulting in more local computations. These local computations are found using computer aided search. To improve performance, special parity (extra) sub-matrix multiplications (PSMMs) are generated (two of them) at the expense of increasing communication/computation cost of the system. Our preliminary results demonstrate that the proposed method outperforms a Strassen-like algorithm with two copies and secures a very close performance to three copy version using only 2 PSMMs, reducing the total number of compute nodes by around 24% i.e., from 21 to 16.Article Citation - WoS: 4Citation - Scopus: 7On the Distribution Modeling of Heavy-Tailed Disk Failure Lifetime in Big Data Centers(IEEE, 2021) Arslan, Şuayb Şefik; Arslan, Şefik Şuayb; Zeydan, Engin; 02.02. Department of Computer Engineering; 02. Faculty of Engineering; 01. MEF UniversityIt has become commonplace to observe frequent multiple disk failures in big data centers in which thousands of drives operate simultaneously. Disks are typically protected by replication or erasure coding to guarantee a predetermined reliability. However, in order to optimize data protection, real life disk failure trends need to be modeled appropriately. The classical approach to modeling is to estimate the probability density function of failures using nonparametric estimation techniques such as kernel density estimation (KDE). However, these techniques are suboptimal in the absence of the true underlying density function. Moreover, insufficient data may lead to overfitting. In this article, we propose to use a set of transformations to the collected failure data for almost perfect regression in the transform domain. Then, by inverse transformation, we analytically estimated the failure density through the efficient computation of moment generating functions, and hence, the density functions. Moreover, we developed a visualization platform to extract useful statistical information such as model-based mean time to failure. Our results indicate that for other heavy-tailed data, the complex Gaussian hypergeometric distribution and classical KDE approach can perform best if the overfitting problem can be avoided and the complexity burden is overtaken. On the other hand, we show that the failure distribution exhibits less complex Argus-like distribution after performing the Box–Cox transformation up to appropriate scaling and shifting operations.Conference Object Citation - WoS: 1Citation - Scopus: 1Hata Düzeltme Çıktı Kodları: Genel Bakış, Zorluklar ve Gelecek Yönelimler(IEEE, 2019) Arslan, Şuayb Şefik; Arslan, Şefik Şuayb; Güney, Osman B.; 02.02. Department of Computer Engineering; 02. Faculty of Engineering; 01. MEF UniversityÇok sınıflı sınıflandırma problemini çözmenin en etkili yollarından biri, bir grup akıllıca tasarlanmıs ikili sınıflandırıcı kullanarak, sınıflandırıcı sonuçlarını belli bir kritere göre bir araya getirmektir. Hata Düzeltme Çıktı Kodları (HDÇK) birden fazla ikili sınıflandırma yoluyla is bölümü saglayan basarılı tekniklerden biridir. Bu çalışmamızın amacı modern HDÇK tiplerine kısa bir giris yapmak, ikili sınıflandırma sonuçlarını birlestiren çesitli kod çözme yöntemleri ve zorlukları, avantajları ve dezavantajlarını ortaya koyan karsılastırmalı bir çalısma sunmaktır. Ayrıca HDÇK tekniğinin birkaç önemli uygulaması, MNIST veri seti üzerindeki performansı ve gelecekteki egilimlerin bazıları sunulmaktadır.Conference Object Citation - WoS: 2Citation - Scopus: 2Distributed Matrix Multiplication With Mds Array Bp-Xor Codes for Scaling Clusters(IEEE, 2019) Arslan, Şefik Şuayb; 02.02. Department of Computer Engineering; 02. Faculty of Engineering; 01. MEF UniversityThis study presents a novel coded computation technique for distributed matrix-matrix product computation at a massive scale that outperforms well known previous strategies in terms of total execution time. Our method achieves this performance by distributing the encoding operation over the cluster (slave) nodes at the expense of increased master-slave communication. The product computation is performed using MDS array Belief Propagation (BP)-decodable codes based on pure XOR operations. In addition, our scheme is configurable and suited for modern compute node architectures equipped with multiple processing units organized in a hierarchical manner. Assuming the number of backup nodes being sublinear in the size of the product, we shall demonstrate that the proposed scheme achieves order-optimal computation from an end-to-end latency perspective while ensuring acceptable communication requirements that can be addressed by today's high speed network link infrastructures.Conference Object Kernel Density Estimation for Optimal Detection in All-Bit Mlc Flash Memories(IEEE, 2019) Arslan, Şefik Şuayb; Ashraf, Reza A.; Pusane, Ali E.; Ashrafi, Reza A.; 02.02. Department of Computer Engineering; 02. Faculty of Engineering; 01. MEF UniversityNAND flash memories have recently become the main component of large-scale non-volatile storage systems. Recent studies have shown that various error sources degrade the Multi-level cell (MLC) memory performance, including intercell interference, retention error, and random telegraph noise. Accurate integration of these error sources into the analytical model to optimally derive the governing probability distributions and consequently the detection thresholds to minimize error rates lie at the heart of MLC research. Utilizing static derivations will not address the detection problem, as aforementioned error sources exhibit a strong non-stationary behavior. In this paper, a novel low-complexity implementation of a non-parametric learning mechanism, kernel density estimation, shall be used to periodically estimate the underlying probability distributions and hence approximate the optimal detection performance for time-varying all-bit-line MLC flash channel.