Faculty of Computer Science and Engineering
Permanent URI for this communityhttps://repository.ukim.mk/handle/20.500.12188/5
The Faculty of Computer Science and Engineering (FCSE) within UKIM is the largest and most prestigious faculty in the field of computer science and technologies in Macedonia, and among the largest
faculties in that field in the region.
The FCSE teaching staff consists of 50 professors and 30 associates. These include many “best in field” personnel, such as the most referenced scientists in Macedonia and the most influential professors in the ICT industry in the Republic of Macedonia.
Browse
56 results
Search Results
- Some of the metrics are blocked by yourconsent settings
Item type:Publication, Routing, Modulation and Spectrum Allocation in Elastic Optical Networks(IEEE, 2018-11); Elastic optical networks are seen as an efficient solution for future optical networks. In these networks, the routing, modulation and spectrum allocation problem consists of finding an optimal network utilization by jointly addressing all three aspects. We examine two different approaches for solving this problem dynamically in a network with a variable traffic. The first approach uses several candidate shortest paths and allows dynamic spectral (re)allocation, while the second approach employs a genetic algorithm for finding the most appropriate path. Our results showed that the second approach could often provide lower average blocking probability and a more efficient spectrum utilization than the first one, particularly for networks with shorter links, indicating that it is a potential solution for future high capacity transport network. - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Multilayer Link Prediction in Online Social Networks(IEEE, 2018-11); ; At present, people are communicating using various online social networks, each exhibiting different topology and structure. Link prediction is an important and difficult task in graph mining with a goal to predict the evolution of a social network using its topological features. However, nowadays it becomes even more important to predict the evolution of an interwoven (mutiplex) network structure by using network features from its constituent parts, i.e. in our case, different online social networks. In this work, we are using certain features from Twitter and Foursquare social networks to predict the link formation in both networks. The results show that the prediction rate depends not only on different node pair features and machine learning algorithms, but also on the properties of the target network on which we predict the link formation. We show that when predicting links in the Foursquare network the best obtained accuracy is above 90%, whereas when predicting the evolution of the Twitter network it is around 87%. We argue that the higher accuracy for link prediction in the Foursquare network is due to its locality nature. Finally, both results show an improvement in the prediction accuracy compared to the existing approaches for this dataset presented in the literature. - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Cuisine classification using recipe's ingredients(IEEE, 2018-05); ; ; The purpose of this paper is to explore the linkage between recipe's ingredients and identification of a cuisine. This has been tackled as a problem of cuisine classification. We will examine various approaches (different machine learning algorithms) for recipes classification based on the recipe's ingredients. The output will be the recommendation of the classification methodology, i.e. what kind of preprocessing can be done to improve the classification and the performance of several classifiers on the dataset we will be using. - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Web tool for graph embeddings representation techniques evaluation(IEEE, 2019-05); ; Graphs are important structures used in wide range of applications techniques for graph analysis proposed in the literature. At recent time, graph embeddings can be found as one of the most popular techniques used for graph analysis. There are plenty graph embedding techniques, but the choice of the most appropriate one is a challenging task which depends on the type of the graph and the intended application. The aim of this paper is to present a web-based tool with a simple user interface that enables performance comparison of different models for graph transformation. The system allows graph embeddings models comparison for three different tasks - node classification, edge prediction and graph reconstruction. Additionally, the users can compare specific algorithms performance for different values of their hyperparameters. This supports and accelerates the process of selecting the most appropriate model for graph transformation and enables the use of these models by a larger number of people in different scientific disciplines. - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Comparative Analysis of Network Embeddings for Functional Annotation in Protein Interaction Networks(IEEE, 2020-09-28); ; ; One of the major problems in bioinformatics is the computational prediction of functions for the large number of sequenced proteins which will facilitate the expensive and long process of wet lab verification. Protein-protein interaction networks (PINs) are considered as one of the richest sources of information for solving this problem. PINs can be represented as graphs, where the nodes are the proteins with their functions as node labels and the edges are their physical interactions. In this paper embedding vectors are created to represent the nodes of the graph which are later used as the input data for a classification model. This is a graph node classification problem and because of the property of proteins to have multiple functions, it is also a multi-label problem. The classification model used is linear SVM, while the embeddings are built with 4 algorithms, HOPE, SDNE, GF and node2vec and then a comparative analysis is done on the results. Hamming loss is used as an evaluation metrics, because of the multi-label problem. Based on the comparative evaluation, recommendation for using a specific network embedding in specific scenarios is given. - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Enhancing and simulating the RIP routing protocol in cloud(IEEE, 2014-11); - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Design of the BCJR decoding algorithm with reduced space complexity(IEEE, 2015-05)Given an M-state (recursive) convolutional encoder, we show that, in theory, computing the forward alpha probabilities of the BCJR decoding algorithm can be done with 2M memory elements. Building on this idea we propose new design with reduced space complexity for the original BCJR algorithm. Initial experiments with rate-1/2 1025-bit-long Turbo Codes show possibility for compressing the size of memory for the alpha probabilities for about 97%. - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Max-log-MAP decoding with reduced memory complexity(IEEE, 2015-09); ; Given an M-state (recursive) convolutional encoder and information sequence of length n, the space complexity of unoptimized Bahl-Cocke-Jelinek-Raviv (BCJR) decoder is considered to be O(nm). However, if BCJR's forward alpha coefficients are continuously recomputed instead of stored in memory, it can be shown that the space complexity will drop to O(m). In this paper we start from these observations and present a technique for memory reduction in the Max-Log-MAP algorithm. We test our design on a rate-1/2 1025-bit-long Turbo Code and show considerable memory saving. - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Sequential Register Renaming(IEEE, 2020-09-28)Register renaming unit is a bottleneck in the superscalar cores because it limits the number of instructions and the number of threads that may concurrently be processed. We propose a register renaming unit with linear complexity with respect to the number of instructions simultaneously renamed. The proposed renaming unit renames source operands in a sequential manner following the program order of the instructions. We show that in worst case sequential register renaming may follow contemporary trends with respect to the number of instructions and the number of threads that may be simultaneously renamed. - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Decoding of LTE Turbo Codes Initialized with the Two Recursive Convolutional Codes(IEEE, 2020-09-28)Turbo codes were the first error-correcting codes that demonstrated reliable communications near the channel capacity with practically feasible hardware. Due to their excellent error-correcting capability, they are part of many modern communication technologies, like 3G, 4G, LTE, etc. An LTE turbo encoder is a rate-1/3 systematic encoder made of a first 8-state recursive convolutional encoder and a second 8-state recursive convolutional encoder. Recursive convolutional encoders are identical, coupled in parallel concatenation scheme, and connected with pseudo-random QPP interleaver. The decoding of LTE Turbo codes is iterative. One iteration starts with decoding of the first convolutional code and continues with decoding the second convolutional code. In order to achieve higher decoding speeds, recursive convolutional codes are decoded with the MAX-Log-MAP algorithm. In this paper, we decode LTE turbo codes with two decoding processes. In one of the decoding processes iterations start with the first convolution code, wherein in the other decoding process iterations start with the second convolution code. We combine the results of both decoding processes to improve turbo decoding process. In channel coding in many communications systems, the speed of decoding is the main design goal. Proposed decoding technique may be used to reduce the number of iterations of LTE turbo decoders, thus improving the decoding speed.