1. Weakly-supervised multilingual medical NER for symptom extraction for low-resource languagesRigon Sallauka, Umut Arioz, Matej Rojc, Izidor Mlakar, 2025, izvirni znanstveni članek Opis: Patient-reported health data, especially patient-reported outcomes measures, are vital for improving clinical care but are often limited by memory bias, cognitive load, and inflexible questionnaires. Patients prefer conversational symptom reporting, highlighting the need for robust methods in symptom extraction and conversational intelligence. This study presents a weakly-supervised pipeline for training and evaluating medical Named Entity Recognition (NER) models across eight languages, with a focus on low-resource settings. A merged English medical corpus, annotated using the Stanza i2b2 model, was translated into German, Greek, Spanish, Italian, Portuguese, Polish, and Slovenian, preserving the entity annotations medical problems, diagnostic tests, and treatments. Data augmentation addressed the class imbalance, and the fine-tuned BERT-based models outperformed baselines consistently. The English model achieved the highest F1 score (80.07%), followed by German (78.70%), Spanish (77.61%), Portuguese (77.21%), Slovenian (75.72%), Italian (75.60%), Polish (75.56%), and Greek (69.10%). Compared to the existing baselines, our models demonstrated notable performance gains, particularly in English, Spanish, and Italian. This research underscores the feasibility and effectiveness of weakly-supervised multilingual approaches for medical entity extraction, contributing to improved information access in clinical narratives—especially in under-resourced languages.
Ključne besede: low-resource languages, machine translation, medical entity extraction, NER, NLP, patient-reported outcomes, weakly-supervised learning
Objavljeno v DKUM: 19.05.2025; Ogledov: 0; Prenosov: 4
 Celotno besedilo (338,94 KB) |
2. Using machine learning and natural language processing for unveiling similarities between microbial dataLucija Brezočnik, Tanja Žlender, Maja Rupnik, Vili Podgorelec, 2024, izvirni znanstveni članek Opis: Microbiota analysis can provide valuable insights in various fields, including diet and nutrition, understanding health and disease, and in environmental contexts, such as understanding the role of microorganisms in different ecosystems. Based on the results, we can provide targeted therapies, personalized medicine, or detect environmental contaminants. In our research, we examined the gut microbiota of 16 animal taxa, including humans, as well as the microbiota of cattle and pig manure, where we focused on 16S rRNA V3-V4 hypervariable regions. Analyzing these regions is common in microbiome studies but can be challenging since the results are high-dimensional. Thus, we utilized machine learning techniques and demonstrated their applicability in processing microbial sequence data. Moreover, we showed that techniques commonly employed in natural language processing can be adapted for analyzing microbial text vectors. We obtained the latter through frequency analyses and utilized the proposed hierarchical clustering method over them. All steps in this study were gathered in a proposed microbial sequence data processing pipeline. The results demonstrate that we not only found similarities between samples but also sorted groups’ samples into semantically related clusters. We also tested our method against other known algorithms like the Kmeans and Spectral Clustering algorithms using clustering evaluation metrics. The results demonstrate the superiority of the proposed method over them. Moreover, the proposed microbial sequence data pipeline can be utilized for different types of microbiota, such as oral, gut, and skin, demonstrating its reusability and robustness.
Ključne besede: machine learning, NLP, hierarchical clustering, microbial data, microbiome, n-grame
Objavljeno v DKUM: 04.09.2024; Ogledov: 38; Prenosov: 13
Celotno besedilo (4,48 MB) |
3. A VAN-Based Multi-Scale Cross-Attention Mechanism for Skin Lesion Segmentation NetworkShuang Liu, Zeng Zhuang, Yanfeng Zheng, Simon Kolmanič, 2023, izvirni znanstveni članek Opis: With the rise of deep learning technology, the field of medical image segmentation has undergone rapid development. In recent years, convolutional neural networks (CNNs) have brought many achievements and become the consensus in medical image segmentation tasks. Although many neural networks based on U-shaped structures and methods, such as skip connections have achieved excellent results in medical image segmentation tasks, the properties of convolutional operations limit their ability to effectively learn local and global features. To address this problem, the Transformer from the field of natural language processing (NLP) was introduced to the image segmentation field. Various Transformer-based networks have shown significant performance advantages over mainstream neural networks in different visual tasks, demonstrating the huge potential of Transformers in the field of image segmentation. However, Transformers were originally designed for NLP and ignore the multidimensional nature of images. In the process of operation, they may destroy the 2D structure of the image and cannot effectively capture low-level features. Therefore, we propose a new multi-scale cross-attention method called M-VAN Unet, which is designed based on the Visual Attention Network (VAN) and can effectively learn local and global features. We propose two attention mechanisms, namely MSC-Attention and LKA-Cross-Attention, for capturing low-level features and promoting global information interaction. MSC-Attention is designed for multi-scale channel attention, while LKA-Cross-Attention is a cross-attention mechanism based on the large kernel attention (LKA). Extensive experiments show that our method outperforms current mainstream methods in evaluation metrics such as Dice coefficient and Hausdorff 95 coefficient.
Ključne besede: CNNs, deep learning, medical image processing, NLP, semantic segmentation
Objavljeno v DKUM: 14.03.2024; Ogledov: 496; Prenosov: 315
Celotno besedilo (1,46 MB)
Gradivo ima več datotek! Več... |
4. Klasifikacija besedila s prenosnim učenjem : magistrsko deloJure Žerak, 2020, magistrsko delo Opis: Magistrsko delo ima namen preizkusiti metodo prenosnega učenja na obdelavi naravnega jezika in jo primerjati s klasičnimi metodami učenja nevronskih mrež, metodo LSTM. V delu sta uporabljena opisna metoda za teoretični in eksperiment za praktični del dela. V slednjem smo ugotovili, da je metoda prenosnega učenja na majhni količini podatkov bolj točna od klasičnih metod, vendar za to potrebuje več časa. Delo primerja prednaučeni model Bert in klasično metodo LSTM, zato je priporočljivo primerjati rezultate tudi z drugimi prednaučenimi modeli in klasičnimi metodami.
Ključne besede: nevronske mreže, prenosno učenje, NLP, PyTorch, LSTM
Objavljeno v DKUM: 01.12.2020; Ogledov: 932; Prenosov: 121
Celotno besedilo (1,99 MB) |
5. Pomen nevrolingvističnega programiranja za odličnost v komunikacijiOlga Gorkič, 2016, diplomsko delo Opis: Komunikacija je sestavni del našega življenja. Ni mogoče ne komunicirati, se je pa mogoče dobrega sporazumevanja, izmenjavanja misli in vzpostavljanja odnosov naučiti. V pričujočem diplomskem delu ugotavljamo, kaj je učinkovita komunikacija, se dotaknemo nastanka in virov nevrolingvističnega programiranja in predstavimo temelje, na katerih je nevrolingvistično programiranje nastalo. Predstavimo uporabne veščine in znanja, s pomočjo katerih spoznavamo sebe in svet okoli nas ter ustvarjamo okoliščine, ki nam omogočajo graditi in vedno znova izboljševati odnose tako na osebnem kot tudi na poslovnem področju. Sledi spoznavanje strategij, ki so načini, kako si organiziramo naše misli in vedenje. Prav prepoznavanje naših misli in vedenja ter ugotavljanje načinov našega zaznavanja sveta je pot, ki nas popelje v ozaveščanje naših želja in posledično v učinkovito postavljanje in doseganje ciljev. Nevrolingvistično programiranje nam ponuja vrsto modelov, ki nam omogočajo, da prepoznamo, kaj hočemo in kako bomo to dosegli, in nekatere izmed njih v tem delu tudi podrobneje predstavljamo. V nadaljevanju spoznamo, kaj je modeliranje, opredelimo ključne elemente in pogoje modeliranja ter spoznamo sam proces modeliranja. Praktično spoznamo modeliranje na primeru, ko modeliramo harmonijo kot notranje stanje človeka.
Ključne besede: nevrolingvistično programiranje, komuniciranje, zaznavanje, NLP strategije, dobro oblikovani cilji, NLP modeliranje
Objavljeno v DKUM: 24.05.2016; Ogledov: 2015; Prenosov: 239
Celotno besedilo (8,22 MB) |
6. |
7. Optimiranje jeklenih nosilcev s pomočjo NLPKen Višnar, 2015, delo diplomskega projekta/projektno delo Opis: Projektna naloga prikazuje princip reševanja problemov s pomočjo matematičnega programiranja - optimizacije in zapisom le te s programskim jezikom GAMS. Za primer prostoležečega jeklenega nosilca je bil modeliran program, v katerem je izvedena optimizacija z upoštevanjem vseh parametrov. Minimirala se je funkcija teže konstrukcije. Rezultat nas je popeljal do stroškov obravnavanega nosilca.
Ključne besede: gradbeništvo, jekleni nosilci, optimiranje, NLP, GAMS
Objavljeno v DKUM: 23.09.2015; Ogledov: 1637; Prenosov: 151
Celotno besedilo (1,21 MB) |
8. Cost optimal project schedulingUroš Klanšek, Mirko Pšunder, 2008, izvirni znanstveni članek Opis: This paper presents the cost optimal project scheduling. The optimization was performed by the nonlinear programming approach, NLP The nonlinear total project cost objective function is subjected to the rigorous system of the activity preceden- ce relationship constraints, the activity duration constraints and the project duration constraints. The set of activity precedence relationship constraints was defined to comprise Finish-to-Start, Start-to-Start, Start-to-Finish and Finish-to-Finish precedence relationships between activities. The activity duration constraints determine relationships between minimum, maximum and possible duration of the project activities. The project duration constraints define the maximum feasible project duration. A numerical example is presented at the end of the paper in order to present the applicability of the proposed approach.
Ključne besede: project management, scheduling, optimization, nonlinear programming, NLP
Objavljeno v DKUM: 10.07.2015; Ogledov: 1863; Prenosov: 427
Celotno besedilo (444,81 KB)
Gradivo ima več datotek! Več... |
9. Soil-nail wall stability analysis using ANFISPrimož Jelušič, Bojan Žlender, 2013, izvirni znanstveni članek Opis: The safety-factor optimization for a soil-nail wall is presented. The optimization is performed using the non-linear programming (NLP) approach. For this purpose, the NLP optimization model OPTINC was developed. The safety factor and the optimal inclination of the soil nails from the horizontal direction depend on the design of the soil-nail wall. Based on these results the ANFIS-INC model was developed for the prediction of the optimal inclination of the soil nail for any design of soil-nail wall. Additionally, an ANFIS-SF model was developed to predict the safety factor for different inclinations of the wall, the slope angle of the terrain, the length of the nails, and the hole diameter. It was found that the inclination of the soil nail should be adjusted to the inclination of wall, the length of nail, the slope angle of the terrain and the hole diameter. With increasing inclination of the wall, the length of the soil nail and the hole diameter, the safety factor is increasing. On the other hand, the safety factor is decreasing with the increasing slope angle of the terrain. The use of nonlinear programming and an Adaptive Network based Fuzzy Inference System allows a comprehensive analysis of the geotechnical problems.
Ključne besede: soil-nail, wall stability, optimization, NLP, ANFIS
Objavljeno v DKUM: 10.07.2015; Ogledov: 2064; Prenosov: 83
Celotno besedilo (284,95 KB)
Gradivo ima več datotek! Več... |
10. Uporabnost nevro-lingvističnega programiranja pri preiskovanju kaznivih dejanj : diplomsko deloAdmir Husić, 2015, diplomsko delo Opis: V diplomski nalogi so predstavljeni temelji nevro-lingvističnega programiranja (NLP) in uporabnost NLP pri preiskovanju kaznivih dejanj.
Kriminaliteta je zaradi svojih posledic vseprisoten, škodljiv in resen problem skoraj vsake družbe. V današnjem času se kljub hitremu napredku znanosti in razvoju družbe še zmeraj soočamo z veliko mero kriminalitete. Bliskovit razvoj različnih znanosti, orodij in orožij, olajšuje in omogoča nove oblike kriminala in njegovega izvrševanja.
Za uspešno zoperstavljanje in boj proti kriminaliteti in storilcem, morajo policisti posedovati določene komunikacijske kompetence in veščine. Opravljanje dela policista zahteva veliko interakcije z različnimi posamezniki, med katere sodijo žrtve, oškodovanci, priče in potencialni storilci. Vsakdo od naštetih ima drugačno vlogo in cilj sodelovanja s policisti. Ena od nalog policistov je zbiranje informacij in dokazov, ki bodo primerni in zakoniti za uspešno rešitev primera. Za opravljanje nalog policije je potrebna uspešna komunikacija.
Poznavanje NLP policistom omogoča drugačen in nov pogled na komunikacijo in procese, ki se odvijajo pri posamezniku v fazi komunikacije. NLP ponuja prednosti, vendar s seboj nosi tudi slabosti, omejitve in pasti. Najbolj pereča problematika NLP so njegove trditve in teorije, ki niso znanstveno in empirično podprte, ponujajo pa rezultate v praksi. Slabost NLP je utrjeno prepričanje laične javnosti, da vzorec očesnih premikov deluje kot sredstvo za odkrivanje laži. Opravljene raziskave ne nudijo konkretnih in dokončnih rezultatov, zato je to področje, ki mu je v prihodnosti potrebno nameniti več pozornosti.
Diplomsko delo je teoretično, začelo se bo s predstavitvijo pojma in temeljev NLP. V nadaljevanju bo predstavljeno preiskovanje kaznivega dejanja in uporabnost NLP pri preiskovanju. Sledilo bo preverjanje in razlaga hipotez diplomskega dela, ki smo jih uvodoma postavili.
Ključne besede: kazniva dejanja, preiskovanje, zaslišanje, informativni razgovori, nevrolingvistično programiranje, NLP, diplomske naloge
Objavljeno v DKUM: 23.02.2015; Ogledov: 3169; Prenosov: 294
Celotno besedilo (570,06 KB) |
