Iskanje po katalogu digitalne knjižnice

Opis: Machine learning (ML) has demonstrated remarkable ability to uncover hidden patterns in data. However, the presence of biases and discrimination originating from the data itself and, consequently, emerging in the ML outcomes, remains a pressing concern. With the exponential growth of unstructured data, such as images, fairness has become increasingly critical, as neural network (NN) models may inadvertently learn and perpetuate societal and historical biases. To address this challenge, we propose a fairness-aware loss function that iteratively prioritizes the worst-performing sensitive group during NN training. This approach aims to balance treatment quality across sensitive groups, achieving fairer image classification outcomes while incurring only a slight compromise in overall performance. Our method, evaluated on the FairFace dataset, demonstrates significant improvements in fairness metrics while maintaining comparable overall quality. These trade-offs highlight that the minor decrease in overall quality is justified by the improvement in fairness of the models.
Ključne besede: fairness, search-basimage classification, machine learning, supervised learnign, neural networks
Objavljeno v DKUM: 23.04.2025; Ogledov: 0; Prenosov: 2
Celotno besedilo (2,01 MB)

Opis: The design of neural network architecture is becoming more difficult as the complexity of the problems we tackle using machine learning increases. Many variables influence the performance of a neural model, and those variables are often limited by the researcher's prior knowledge and experience. In our master's thesis, we will focus on becoming familiar with evolutionary neural network design, anomaly detection techniques, and a deeper knowledge of autoencoders and their potential for application in unsupervised learning. Our practical objective will be to build a neural architecture search based on swarm intelligence, and construct an autoencoder architecture for anomaly detection in the MNIST dataset.
Ključne besede: neural architecture search, machine learning, swarm intelligence
Objavljeno v DKUM: 18.10.2021; Ogledov: 1269; Prenosov: 113
Celotno besedilo (3,18 MB)
Gradivo ima več datotek! Več...

Opis: Purpose: This paper examines the distribution of elderly missing persons in the city of Zagreb per variables (age, sex of the missing person, social and marital status, employment, place and duration of disappearance, method of finding the missing person, its outcome, reasons for disappearance, etc.) needed for the design of the plan of the search for the missing person. Design/Methods/Approach: The paper uses data collected through a questionnaire with 417 variables and processed with certain basic statistical methods. The collected data are examined on a sample of 170 elderly missing persons in the city of Zagreb, and through characteristics of methods, measures and actions, together with the methods of search for elderly persons in cities, with special reference to Zagreb. Findings: The statistically established behavioural modes of certain groups of elderly people are analysed: persons with Alzheimer’s disease and persons with suicidal risk. The stated findings largely assist in planning and conducting the search for those persons, especially for missing person profiling and his/her possible movement since disappearance. The success of the search for the missing person presumes knowledge about the stated specifics by all searchers. It especially applies to the police in the formal, due to its sole duty and competence for the search for missing persons, as well as in the real sense where they conduct search measures and actions together with other participants adapted to real situations of disappearance of elderly persons. Research Limitations / Implications: The research has been limited by data available from the police bulletin of daily events, but this limitation was partially removed by a direct insight into the police files containing the case history. During the research, the legal and ethical regulations regarding personal data protection were strictly followed. Originality/Value: There is a number of researches on police search for the missing persons (UK, USA, Australia etc.), but few of them deal with elderly persons. This paper introduces a new offensive approach to the search for the missing person on a strategic level (missing person profiling with regard to specific variables, possible movement and routes, “wanderings”). New police procedures are suggested on a tactical level (e.g., stopping a person who disappeared in a certain area), but they include traditional police methods (car & regular patrols, police officers on bicycles, police search dogs, collecting information, conducting interviews, terrain search etc.). Introduction of the principle of partnership search for the missing person. Research results should be useful to police practice and their education.
Ključne besede: missing persons, police search, elderly persons, Alzheimer's disease, Croatia
Objavljeno v DKUM: 16.04.2020; Ogledov: 1070; Prenosov: 70
Celotno besedilo (366,35 KB)
Gradivo ima več datotek! Več...

Opis: V magistrski nalogi smo preučili smotrnost integracije iskalnega pogona ElasticSearch v sistem e-Arhiv Pošte Slovenije, ki v obstoječi konfiguraciji uporablja iskalni pogon Microsoft SQL Server Full Text Search. V ta namen smo primerjali oba iskalna pogona, in sicer z vidika hitrosti procesiranja iskalnih zahtevkov, hitrosti indeksiranja polno-tekstovnih podatkov in hitrosti procesiranja iskalnih zahtevkov v času, ko se izvaja indeksiranje polno-tekstovnih podatkov. Preizkus obeh iskalnih pogonov je potekal na isti infrastrukturi z identično strojno konfiguracijo. Za doseganje večje veljavnost in zanesljivosti raziskave smo uporabili podatke iz testnega sistema e-Arhiv Pošte Slovenije. Rezultati so pokazali, da je iskalni pogon ElasticSearch z vidika hitrosti iskanja in indeksiranja polno-tekstovnih podatkov boljša izbira od iskalnega pogona Microsoft SQL Server Full Text Search in da predstavlja zelo dobrega kandidata za integracijo v sistem e-Arhiv Pošte Slovenije.
Ključne besede: Polno-tekstovno iskanje, ElasticSearch, Microsoft SQL Server Full Text Search, polno-tekstovni indeks
Objavljeno v DKUM: 21.12.2018; Ogledov: 2669; Prenosov: 223
Celotno besedilo (3,68 MB)

Opis: Background and Purpose: We are living in a time of accelerated globalization, which has a far-reaching impact for youth employment. Therefore, the paper presents the problem of the transition of tertiary educated young people into the working environment in terms of the importance of education and gaining work experience during their studies. The main purpose of the study was to analyze the attitudes and thinking of students regarding the selection of studies and work experience that young people are gaining during their studies with student work in connection with their job prospects. Design/Methodology/Approach: The data used in our research come from a broader research on Slovenian Youths (SY) dating back to 2013. Among all the respondents of SY we selected those who classified themselves as students aged between 19 and 24, who filled out the part of the questionnaire pertaining to employment. The research hypotheses were tested with the analysis of variance (ANOVA) and independent samples t-test. Results: We found that among students who chose to study for different reasons there were differences in the belief that they will immediately find employment after graduating. We also found that between those students who supported themselves during their studies and those who only studied there were differences in individual statements regarding their transition from education to employment. We concluded that work during their studies, especially in the field of the profession they were studying for, had a positive impact on the thinking of students pertaining to the transition toward employment. Conclusion: The analysis of employment issues faced by young people with tertiary education was supported by empirical data, while in conclusion suggestions for improvement were presented. The article also deepens the understanding of the problems of youth employment in a broader perspective and the understanding of the observed developments in society.
Ključne besede: youth employment, tertiary education, work experience, job search
Objavljeno v DKUM: 10.10.2018; Ogledov: 1988; Prenosov: 393
Celotno besedilo (468,08 KB)
Gradivo ima več datotek! Več...