1. |
2. Automated speech analysis in depressive disorder: enhancing diagnosis and monitoringAljaž Neuberg, 2025, magistrsko delo Opis: This paper investigates the automatic recognition of depression by integrating acoustic, linguistic and emotional features extracted from clinical interviews in the DAIC-WOZ dataset. A total of six classical machine learning classifiers such as Decision Tree, Random Forest, SVM, Gradient Boosting, AdaBoost and XGBoost were systematically evaluated under different class balancing methods (such as SMOTE, SMOTETomek and Random Undersampling) and feature selection strategies. The best model, a decision tree classifier with SMOTE-based balancing and a feature selection technique, achieved a weighted F1 score and accuracy of 0.78 with only eight selected features. These features included all three modalities, demonstrating the
added benefit of a multimodal approach. The results suggest that even relatively simple models, when supported by careful preprocessing and dimensionality reduction, can provide accurate and interpretable predictions. This work emphasizes the importance of feature engineering and balancing techniques in clinical machine learning tasks and lays the foundation for future research on scalable and explainable depression detection systems.
Ključne besede: Depression, Classification, Machine Learning, Data Balancing, Feature Selection
Objavljeno v DKUM: 03.11.2025; Ogledov: 0; Prenosov: 2
 Celotno besedilo (4,28 MB) |
3. Uporaba metod strojnega učenja za klasifikacijo nalog po prioritetah v IT projektihTatyana Unuchak, Mirjana Kljajić Borštnar, Yauhen Unuchak, 2025, izvirni znanstveni članek Opis: Določanje prioritet in razvrščanje nalog še vedno predstavlja izziv pri učinkovitem vodenju projektov. Obstaja veliko klasičnih pristopov
za določanje prioritet. Vendar so te tehnike delovno intenzivne, subjektivne in neprilagodljive. V prispevku obravnavamo pristope za
samodejno določanje prioritet nalog v IT projektih, ki temeljijo na strojnem učenju. Raziskujemo, kako lahko z uporabo metod strojnega učenja pomagamo projektnim vodjem pri učinkovitejšem razvrščanju nalog v IT projektih. V ta namen smo na množici več kot
1000000 zapisov projektnih nalog razvili klasifikacijski model za samodejno določanje prioritet. Problem, ki smo ga obravnavali, je
večrazredni, pri tem je večina primerov, označenih z najvišjo prioriteto, kar predstavlja izziv pri modeliranju kot tudi pri učinkovitosti
upravljanja IT projektov. Preskusili smo različne algoritme ter različne pristope, s ciljem izboljšanja rezultatov klasifikacije. Pokazali
smo, da je naloge smiselno razvrstiti v manjše skupine prioritet, kar prispeva k večji natančnosti klasifikacijskega modela in preglednosti prioritet nalog, slednje pa lahko olajša upravljanje IT projektov.
Ključne besede: IT project management, machine learning, task prioritization, multiclass classification, data imbalance
Objavljeno v DKUM: 28.08.2025; Ogledov: 0; Prenosov: 3
Celotno besedilo (2,25 MB)
Gradivo ima več datotek! Več... |
4. Classification of finger movements through optimal EEG channel and feature selectionMurside Degirmenci, Yilmaz Kemal Yuce, Matjaž Perc, Yalcin Isler, 2025, izvirni znanstveni članek Ključne besede: classification, finger movements, EEG, feature selection, applied physics
Objavljeno v DKUM: 22.07.2025; Ogledov: 0; Prenosov: 7
Celotno besedilo (1,08 MB)
Gradivo ima več datotek! Več... |
5. Evaluating Proprietary and Open-Weight Large Language Models as Universal Decimal Classification Recommender SystemsMladen Borovič, Eftimije Tomovski, Tom Li Dobnik, Sandi Majninger, 2025, izvirni znanstveni članek Opis: Manual assignment of Universal Decimal Classification (UDC) codes is time-consuming and inconsistent as digital library collections expand. This study evaluates 17 large language models (LLMs) as UDC classification recommender systems, including ChatGPT variants (GPT-3.5, GPT-4o, and o1-mini), Claude models (3-Haiku and 3.5-Haiku), Gemini series (1.0-Pro, 1.5-Flash, and 2.0-Flash), and Llama, Gemma, Mixtral, and DeepSeek architectures. Models were evaluated zero-shot on 900 English and Slovenian academic theses manually classified by professional librarians. Classification prompts utilized the RISEN framework, with evaluation using Levenshtein and Jaro–Winkler similarity, and a novel adjusted hierarchical similarity metric capturing UDC’s faceted structure. Proprietary systems consistently outperformed open-weight alternatives by 5–10% across metrics. GPT-4o achieved the highest hierarchical alignment, while open-weight models showed progressive improvements but remained behind commercial systems. Performance was comparable between languages, demonstrating robust multilingual capabilities. The results indicate that LLM-powered recommender systems can enhance library classification workflows. Future research incorporating fine-tuning and retrieval-augmented approaches may enable fully automated, high-precision UDC assignment systems.
Ključne besede: universal decimal classification, large language models, conversational systems, recommender systems, prompt engineering, zero-shot classification, hierarchical similarity
Objavljeno v DKUM: 21.07.2025; Ogledov: 0; Prenosov: 11
Celotno besedilo (447,50 KB)
Gradivo ima več datotek! Več... |
6. AI model for industry classification based on website dataTimotej Jagrič, Aljaž Herman, 2024, izvirni znanstveni članek Opis: This paper presents a broad study on the application of the BERT (Bidirectional Encoder Representations from Transformers) model for multiclass text classification, specifically focusing on categorizing business descriptions into 1 of 13 distinct industry categories. The study involved a detailed fine-tuning phase resulting in a consistent decrease in training loss, indicative of the model’s learning efficacy. Subsequent validation on a separate dataset revealed the model’s robust performance, with classification accuracies ranging from 83.5% to 92.6% across different industry classes. Our model showed a high overall accuracy of 88.23%, coupled with a robust F1 score of 0.88. These results highlight the model’s ability to capture and utilize the nuanced features of text data pertinent to various industries. The model has the capability to harness real-time web data, thereby enabling the utilization of the latest and most up-to-date information affecting to the company’s product portfolio. Based on the model’s performance and its characteristics, we believe that the process of relative valuation can be drastically improved.
Ključne besede: industry classification, BERT transformer, business descriptions, multiclass text classification, AI
Objavljeno v DKUM: 01.07.2025; Ogledov: 0; Prenosov: 10
Celotno besedilo (1,01 MB)
Gradivo ima več datotek! Več... |
7. Fostering fairness in image classification through awareness of sensitive dataIvona Colakovic, Sašo Karakatič, 2025, izvirni znanstveni članek 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: 6
Celotno besedilo (2,01 MB) |
8. EEG-based finger movement classification with intrinsic time-scale decompositionMurside Degirmenci, Yilmaz Kemal Yuce, Matjaž Perc, Yalcin Isler, 2024, izvirni znanstveni članek Ključne besede: brain-computer interfaces, electroencephalogram, feature reduction, machine learning, finger movements classification, time series analysis
Objavljeno v DKUM: 16.04.2025; Ogledov: 0; Prenosov: 3
Celotno besedilo (1,44 MB)
Gradivo ima več datotek! Več... |
9. Parkinson’s disease non-motor subtypes classification in a group of Slovenian patients : actuarial vs. data-driven approachTimotej Petrijan, Jan Zmazek, Marija Menih, 2023, izvirni znanstveni članek Opis: Background and purpose: The aim of this study was to examine the risk factors, prodromal symptoms, non-motor symptoms (NMS), and motor symptoms (MS) in different Parkinson’s disease (PD) non-motor subtypes, classified using newly established criteria and a data-driven approach.
Methods: A total of 168 patients with idiopathic PD underwent comprehensive NMS and MS examinations. NMS were assessed by the Non-Motor Symptom Scale (NMSS), Montreal Cognitive Assessment (MoCA), Hamilton Depression Scale (HAM-D), Hamilton Anxiety Rating Scale (HAM-A), REM Sleep Behavior Disorder Screening Questionnaire (RBDSQ), Epworth Sleepiness Scale (ESS), Starkstein Apathy Scale (SAS) and Fatigue Severity Scale (FSS). Motor subtypes were classified based on Stebbins’ method. Patients were classified into groups of three NMS subtypes (cortical, limbic, and brainstem) based on the newly designed inclusion criteria. Further, data-driven clustering was performed as an alternative, statistical learning-based classification approach. The two classification approaches were compared for consistency.
Results: We identified 38 (22.6%) patients with the cortical subtype, 48 (28.6%) with the limbic, and 82 (48.8%) patients with the brainstem NMS PD subtype. Using a data-driven approach, we identified five different clusters. Three corresponded to the cortical, limbic, and brainstem subtypes, while the two additional clusters may have represented patients with early and advanced PD. Pearson chi-square test of independence revealed that a priori classification and cluster membership were significantly related to one another with a large effect size (χ2(8) = 175.001, p < 0.001, Cramer’s V = 0.722). The demographic and clinical profiles differed between NMS subtypes and clusters.
Conclusion: Using the actuarial and clustering approach, marked differences between individual NMS subtypes were found. The newly established criteria have potential as a simplified tool for future clinical research of NMS subtypes of Parkinson’s disease.
Ključne besede: Parkinson’s disease, non-motor symptoms subtypes, a priori classification, cluster analysis
Objavljeno v DKUM: 07.04.2025; Ogledov: 0; Prenosov: 9
Celotno besedilo (1,37 MB)
Gradivo ima več datotek! Več... |
10. An overview of the current state of cell viability assessment methods using OECD classificationEneko Madorran, Miha Ambrož, Jure Knez, Monika Sobočan, 2025, pregledni znanstveni članek Opis: Over the past century, numerous methods for assessing cell viability have been
developed, and there are many different ways to categorize these methods accordingly. We
have chosen to use the Organisation for Economic Co-operation and Development (OECD)
classification due to its regulatory importance. The OECD categorizes these methods
into four groups: non-invasive cell structure damage, invasive cell structure damage, cell
growth, and cellular metabolism. Despite the variety of cell viability methods available,
they can all be categorized within these four groups, except for two novel methods based
on the cell membrane potential, which we added to the list. Each method operates on
different principles and has its own advantages and disadvantages, making it essential
for researchers to choose the method that best fits their experimental design. This review
aims to assist researchers in making this decision by describing these methods regarding
their potential use and providing direct references to the cell viability assessment methods.
Additionally, we use the OECD classification to facilitate potential regulatory use and to
highlight the need for adding a new category to their list.
Ključne besede: cell viability, cell-based methods, in vitro toxicology, OECD cell viability classification
Objavljeno v DKUM: 13.02.2025; Ogledov: 0; Prenosov: 8
Celotno besedilo (2,77 MB)
Gradivo ima več datotek! Več... |
