Iskanje po katalogu digitalne knjižnice

Opis: In this paper we present NiaAML, an AutoML framework that we have developed for creating machine learning pipelines and hyperparameter tuning. The composition of machine learning pipelines is presented as an optimization problem that can be solved using various stochastic, population-based, nature-inspired algorithms. Nature-inspired algorithms are powerful tools for solving real-world optimization problems, especially those that are highly complex, nonlinear, and involve large search spaces where traditional algorithms may struggle. They are applied widely in various fields, including robotics, operations research, and bioinformatics. This paper provides a comprehensive overview of the software architecture, and describes the main tasks of NiaAML, including the automatic composition of classification and regression pipelines. The overview is supported by an practical illustrative example.
Ključne besede: AutoML, classification, nature-inspired algorithms, optimization
Objavljeno v DKUM: 19.01.2026; Ogledov: 0; Prenosov: 1
Celotno besedilo (1,43 MB)
Gradivo ima več datotek! Več...

Opis: Learning and implementing algorithms is a fundamental but challenging aspect of Computer Science education. One of the key tools used in teaching algorithms is pseudocode, which serves as an abstract representation of the logic behind a given algorithm. This study explores the educational value of the FLoCIC (Few Lines of Code for Image Compression) algorithm, which is designed to teach lossless image compression through algorithmic implementation, particularly within the context of multimedia data. Image compression represents a typical multimedia task that combines algorithmic thinking with practical problem-solving. By analysing questionnaire responses (N = 121) from undergraduate and graduate students, this study identifies critical challenges in pseudocode-based learning, including understanding complex algorithmic components and debugging recursive functions. This paper highlights the influence of prior knowledge in areas such as data structures, compression, and algorithms in general on the success of students in completing the task, with graduate students demonstrating stronger results compared to undergraduates. The study analyses the role of external resources and online code repositories, further revealing their utility in supporting implementation efforts but highlighting the need for a fundamental understanding of the algorithm for successful implementation. The findings highlight the importance of promoting conceptual understanding and practical problem-solving skills to improve student learning in algorithmic tasks.
Ključne besede: FLoCIC, computer science, algorithms, pseudocode, coding, generative AI, multimedia, education
Objavljeno v DKUM: 02.10.2025; Ogledov: 0; Prenosov: 13
Celotno besedilo (1,65 MB)

Opis: The problem of finding aperiodic low auto-correlation binary sequences (LABS) presents a significant computational challenge, particularly as the sequence length increases. Such sequences have important applications in communication engineering, physics, chemistry, and cryptography. This paper introduces a dual-step algorithm for long binary sequences with high merit factors. The first step employs a parallel algorithm utilizing skew-symmetry and restriction classes to generate sequence candidates with merit factors above a predefined threshold. The second step uses a priority queue algorithm to refine these candidates further, searching the entire search space unrestrictedly. By combining GPU-based parallel computing and dual-step optimization, our approach has successfully identified best-known binary sequences for all lengths ranging from 450 to 527, with the exception of length 518, where the previous best-known merit factor value was matched with a different sequence. This hybrid method significantly outperforms traditional exhaustive and stochastic search methods, offering an efficient solution for finding long sequences with good merit factors.
Ključne besede: binary sequences, Golay's merit factor, autocorrelation, algorithms
Objavljeno v DKUM: 30.05.2025; Ogledov: 0; Prenosov: 13
Celotno besedilo (968,30 KB)

Opis: Reporting the empirical results of swarm and evolutionary computation algorithms is a challenging task with many possible difficulties. These difficulties stem from the stochastic nature of such algorithms, as well as their inability to guarantee an optimal solution in polynomial time. This research deals with measuring the performance of stochastic optimization algorithms, as well as the confidence intervals of the empirically obtained statistics. Traditionally, the arithmetic mean is used for measuring average performance, but we propose quantiles for measuring average, peak and bad-case performance, and give their interpretations in a relevant context for measuring the performance of the metaheuristics. In order to investigate the differences between arithmetic mean and quantiles, and to confirm possible benefits, we conducted experiments with 7 stochastic algorithms and 20 unconstrained continuous variable optimization problems. The experiments showed that median was a better measure of average performance than arithmetic mean, based on the observed solution quality. Out of 20 problem instances, a discrepancy between the arithmetic mean and median happened in 6 instances, out of which 5 were resolved in favor of median and 1 instance remained unresolved as a near tie. The arithmetic mean was completely inadequate for measuring average performance based on the observed number of function evaluations, while the 0.5 quantile (median) was suitable for that task. The quantiles also showed to be adequate for assessing peak performance and bad-case performance. In this paper, we also proposed a bootstrap method to calculate the confidence intervals of the probability of the empirically obtained quantiles. Considering the many advantages of using quantiles, including the ability to calculate probabilities of success in the case of multiple executions of the algorithm and the practically useful method of calculating confidence intervals, we recommend quantiles as the standard measure of peak, average and bad-case performance of stochastic optimization algorithms.
Ključne besede: algorithmic performance, experimental evaluation, metaheuristics, quantile, confidence interval, stochastic algorithms, evolutionary computation, swarm intelligence, experimental methodology
Objavljeno v DKUM: 28.03.2025; Ogledov: 0; Prenosov: 16
Celotno besedilo (490,48 KB)
Gradivo ima več datotek! Več...

Opis: Drawing is a well-established method to improve the mechanical properties of wet-spun fibers, as it orients the polymer chains, increases the chain density, and homogenizes the microstructure. This work aims to investigate how drawing variables, such as the draw ratio, drawing speed, and temperature affect the elastic modulus (E) and the strain at break (εB) of biobased wet-spun fibers constituted by neat polylactic acid (PLA) and a PLA/poly(dodecamethylene 2,5-furandicarboxylate) (PDoF) (80/20 wt/wt) blend. Drawing experiments were conducted with a design of experiment (DOE) approach following a 24 full factorial design. The results of the quasi-static tensile tests on the drawn fibers, analyzed by the analysis of variance (ANOVA) and modeled through the response surface methodology (RSM), highlight that the presence of PDoF significantly lowers E, which instead is maximized if the temperature and draw ratio are both low. On the other hand, εB is enhanced when the drawing is performed at a high temperature. Finally, a genetic algorithm was implemented to find the optimal combination of drawing parameters that maximize both E and εB. The resulting Pareto curve highlights that the temperature influences the mechanical results only for neat PLA fibers, as the stiffness increases by drawing at lower temperatures, while optimal Pareto points for PLA/PDoF fibers are mainly determined by the draw ratio and the draw rate.
Ključne besede: fibers, poly(lactic acid), furanoate polyesters, drawing, response surface methodology, genetic algorithms
Objavljeno v DKUM: 24.03.2025; Ogledov: 0; Prenosov: 8
Celotno besedilo (1,23 MB)
Gradivo ima več datotek! Več...