Computer Science
Loading...
ISSN 1508-2806
e-ISSN: 2300-7036
Issue Date
2022
Volume
Vol. 23
Number
No. 2
Description
Journal Volume
Computer Science
Vol. 23 (2022)
Projects
Pages
Articles
Finding frequent items: novel method for improving Apriori algorithm
(Wydawnictwa AGH, 2022) Karimtabar, Noorollah; Fard, Mohammad Javad Shayegan
In this paper, we use an intelligent method for improving the Apriori algorithm in order to extract frequent itemsets. PAA (the proposed Apriori algorithm) pursues two goals: first, it is not necessary to take only one data item at each step – in fact, all possible combinations of items can be generated at each step, and second, we can scan only some transactions instead of scanning all of the transactions to obtain a frequent itemset. For performance evaluation, we conducted three experiments with the traditional Apriori, BitTableFI, TDM-MFI, and MDC-Apriori algorithms. The results exhibited that the algorithm execution time was significantly reduced due to the significant reduction in the number of transaction scans to obtain the itemset. As in the first experiment, the time that was spent to generate frequent items underwent a reduction of 52% as compared to the algorithm in the first experiment. In the second experiment, the amount of time that was spent was equal to 65%, while in the third experiment, it was equal to 46%.
Assessment of nature-inspired algorithms for text feature selection
(Wydawnictwa AGH, 2022) Çoban, Önder
This paper provides a comprehensive assessment of basic feature selection (FS) methods that have originated from nature-inspired (NI) meta-heuristics, two well-known filter-based FS methods are also included for comparison. The performances of the considered methods are compared on four balanced highdimensional and real-world text data sets regarding the accuracy, the number of selected features, and computation time. This study differs from existing studies in terms of the extent of experimental analyses that were performed under different circumstances where the classifier, feature model, and term-weighting scheme were different. The results of the extensive experiments indicated that basic NI algorithms produce slightly different results than filter-based methods for the text FS problem. However, filter-based methods often provide better results by using lower numbers of features and computation times.
Set representation for rule-generation algorithms
(Wydawnictwa AGH, 2022) Kharkongor, Carynthia; Nath, Bhabesh
The task of mining association rules has become one of the most widely used discovery pattern methods in knowledge discovery in databases (KDD). One such task is to represent an item set in the memory. The representation of the item set largely depends on the type of data structure that is used for storing them. Computing the process of mining an association rule impacts the memory and time requirements of the item set. With the constant increase of the dimensionality of data and data sets, mining such a large volume of data sets will be difficult since all of these item sets cannot be placed in the main memory. As the representation of an item set greatly affects the efficiency of the rule-mining association, a compact and compressed representation of the item set is needed. In this paper, a set representation is introduced that is more memory- and cost-efficient. Bitmap representation takes 1 byte for an element, but a set representation uses 1 bit. The set representation is being incorporated in the Apriori algorithm. Set representation is also being tested for different rule-generation algorithms. The complexities of these different rule-generation algorithms that use set representation are being compared in terms of memory and time of execution.
Human gesture recognition using hidden Markov models and sensor Fusion
(Wydawnictwa AGH, 2022) Emmanuel, Domínguez Ramón; Raquel, Díaz Hernández; Leopoldo, Altamirano Robles
Considering the continued drive of human needs along with the constant improvement of technology, it is convenient to develop techniques that can enhance communication between computers and humans in the most intuitive ways possible. The possibility of automatically recognizing human gestures using artificial vision (among other kinds of sensors) allows us to explore a whole range of applications to control and interact with environments. Nowadays, most approaches for gesture recognition using sensors agree in the use of vision, myography, and movement devices that are applied to robotic, medical, and industrial applications. In the context of this work, we study the principles of using both vision and body contact sensing applied to the automatic classification of a human gesture set. For this, two different approaches have been evaluated: feed-forward neural networks, and hidden Markov models. These models have been studied and implemented for recognizing up to eight different human hand gestures that are commonly applied in collaborative robotics tasks.
Single-shot determination of differential gene network on multiple disease subtypes
(Wydawnictwa AGH, 2022) Sadhu, Arnab; Bhattacharyya, Balaram; Mukhopadhyay, Tathagato
A differential gene expressional network determines the prominent genes under altered phenotypes. The traditional approach requires $n(n-2)/2$ comparisons for $n$ phenotypes. We present a direct method for determining a differentia network under multiple phenotypes. We explore the non-discrete nature of gene expression as a pattern in a fuzzy rough set. An edge between a pair of genes represents a positive region of a fuzzy similarity relationship upon a phenotypic change. We apply a weight-ranking formula and obtain a directed ranked network, we label this as a phenotype interwoven network. Those nodes with large in-degree connectivity bubble up as significant genes under respective phenotypic changes. We tested the method on six diseases and achieved good corroboration with the results of previous studies in the two-step approach. The subgraphs of the isolated genes achieved good significance upon validation through an information theoretic approach. The top-ranking genes determined in all of our case studies are in consonance with the findings of the respective wet-lab tests.