1. A brief review on benchmarking for large language models evaluation in healthcareLeona Cilar Budler, Hongyu Chen, Aokun Chen, Maxim Topaz, Wilson Tam, Jiang Bian, Gregor Štiglic, 2025, pregledni znanstveni članek Opis: This paper reviews benchmarking methods for evaluating large language models (LLMs) in healthcare settings. It highlights the importance of rigorous benchmarking to ensure LLMs' safety, accuracy, and effectiveness in clinical applications. The review also discusses the challenges of developing standardized benchmarks and metrics tailored to healthcare-specific tasks such as medical text generation, disease diagnosis, and patient management. Ethical considerations, including privacy, data security, and bias, are also addressed, underscoring the need for multidisciplinary collaboration to establish robust benchmarking frameworks that facilitate LLMs' reliable and ethical use in healthcare. Evaluation of LLMs remains challenging due to the lack of standardized healthcare-specific benchmarks and comprehensive datasets. Key concerns include patient safety, data privacy, model bias, and better explainability, all of which impact the overall trustworthiness of LLMs in clinical settings. Ključne besede: artificial intelligence, benchmarking, chatbots, healthcare, large language models, natural language processing Objavljeno v DKUM: 12.05.2025; Ogledov: 0; Prenosov: 0
Povezava na datoteko |
2. Improving personalized meal planning with large language models: identifying and decomposing compound ingredientsLeon Kopitar, Leon Bedrač, Larissa Jane Strath, Jiang Bian, Gregor Štiglic, 2025, izvirni znanstveni članek Opis: Background/Objectives: Identifying and decomposing compound ingredients within meal plans presents meal customization and nutritional analysis challenges. It is essential for accurately identifying and replacing problematic ingredients linked to allergies or intolerances and helping nutritional evaluation. Methods: This study explored the effectiveness of three large language models (LLMs)—GPT-4o, Llama-3 (70B), and Mixtral (8x7B), in decomposing compound ingredients into basic ingredients within meal plans. GPT-4o was used to generate 15 structured meal plans, each containing compound ingredients. Each LLM then identified and decomposed these compound items into basic ingredients. The decomposed ingredients were matched to entries in a subset of the USDA FoodData Central repository using API-based search and mapping techniques. Nutritional values were retrieved and aggregated to evaluate accuracy of decomposition. Performance was assessed through manual review by nutritionists and quantified using accuracy and F1-score. Statistical significance was tested using paired t-tests or Wilcoxon signed-rank tests based on normality. Results: Results showed that large models—both Llama-3 (70B) and GPT-4o—outperformed Mixtral (8x7B), achieving average F1-scores of 0.894 (95% CI: 0.84–0.95) and 0.842 (95% CI: 0.79–0.89), respectively, compared to an F1-score of 0.690 (95% CI: 0.62–0.76) from Mixtral (8x7B). Conclusions: The open-source Llama-3 (70B) model achieved the best performance, outperforming the commercial GPT-4o model, showing its superior ability to consistently break down compound ingredients into precise quantities within meal plans and illustrating its potential to enhance meal customization and nutritional analysis. These findings underscore the potential role of advanced LLMs in precision nutrition and their application in promoting healthier dietary practices tailored to individual preferences and needs. Ključne besede: artificial intelligence, food analysis, LLM, Ilama, GPT, mixtral, ingredient identification, ingredient decomposition, personalized nutrition, meal customization, nutritional analysis, dietary planning Objavljeno v DKUM: 08.05.2025; Ogledov: 0; Prenosov: 1
Celotno besedilo (684,70 KB) Gradivo ima več datotek! Več... |
3. Reasons for Facebook usage : data from 46 countriesMarta Kowal, Piotr Sorokowski, Agnieszka Sorokowska, Małgorzata Dobrowolska, Katarzyna Pisanski, Anna Oleszkiewicz, Toivo Aavik, Grace Akello, Charlotte Alm, Naumana Amjad, Maja Zupančič, Tina Kavčič, Bojan Musil, Nejc Plohl, Afifa Anjum, Kelly Asao, Chiemezie Atama, Derya Atamtürk Duyar, Richard Ayebare, Mons Bendixen, Aicha Bensafia, Boris Bizumic, Mahmoud Boussena, David M. Buss, Marina Butovskaya, Seda Can, Katarzyna Cantarero, Antonin Carrier, Hakan Cetinkaya, Daniel Conroy-Beam, Marco A. C. Varella, Rosa M. Cueto, Marcin Czub, Daria Dronova, Seda Dural, Izzet Duyar, Berna Ertugrul, Agustín Espinosa, Ignacio Estevan, Carla S. Esteves, Tomasz Frackowiak, Jorge Contreras-Graduño, Farida Guemaz, Ivana Hromatko, Chin-Ming Hui, Iskra Herak, Jas L. Jaafar, Feng Jiang, Konstantinos Kafetsios, Leif Edward Ottesen Kennair, Nicolas Kervyn, Nils C. Köbis, András Láng, Georgina R. Lennard, Ernesto León, Torun Lindholm, Giulia Lopez, Mohammad Madallh Alhabahba, Alvaro Mailhos, Zoi Manesi, Rocío Martínez, Sarah L. McKerchar, Norbert Meskó, Girishwar Misra, Hoang Moc Lan, Conal Monaghan, Emanuel C. Mora, Alba Moya Garófano, George Nizharadze, Elisabeth Oberzaucher, Mohd S. Omar Fauzee, Ike E. Onyishi, Baris Özener, Ariela F. Pagani, Vilmante Pakalniskiene, Miriam Parise, Farid Pazhoohi, Mariia Perun, Annette Pisanski, Camelia Popa, Pavol Prokop, Muhammad Rizwan, Mario Sainz, Svjetlana Salkicević, Ruta Sargautyte, Susanne Schmehl, Oksana Senyk, Rizwana Shaikh, Shivantika Sharad, Franco Simonetti, Meri Tadinac, Truong Thi Khanh Ha, Trinh Thi Linh, Karina Ugalde González, Nguyen Van Luot, Christin-Melanie Vauclair, Luis D. Vega, Gyesook Yoo, Stanislava Yordanova Stoyanova, Zainab F. Zadeh, 2020, drugi znanstveni članki Ključne besede: online social networks, Facebook, cross cultural psychology, cross cultural differences, human sex differences, age differences, motives Objavljeno v DKUM: 27.01.2025; Ogledov: 0; Prenosov: 8
Celotno besedilo (259,63 KB) Gradivo ima več datotek! Več... |
4. Tilt correction toward building detection of remote sensing imagesKang Liu, Zhiyu Jiang, Mingliang Xu, Matjaž Perc, Xuelong Li, 2021, izvirni znanstveni članek Opis: Building detection is a crucial task in the field of remote sensing, which can facilitate urban construction planning, disaster survey, and emergency landing. However, for large-size remote sensing images, the great majority of existing works have ignored the image tilt problem. This problem can result in partitioning buildings into separately oblique parts when the large-size images are partitioned. This is not beneficial to preserve semantic completeness of the building objects. Motivated by the above fact, we first propose a framework for detecting objects in a large-size image, particularly for building detection. The framework mainly consists of two phases. In the first phase, we particularly propose a tilt correction (TC) algorithm, which contains three steps: texture mapping, tilt angle assessment, and image rotation. In the second phase, building detection is performed with object detectors, especially deep-neural-network-based methods. Last but not least, the detection results will be inversely mapped to the original large-size image. Furthermore, a challenging dataset named Aerial Image Building Detection is contributed for the public research. To evaluate the TC method, we also define an evaluation metric to compute the cost of building partition. The experimental results demonstrate the effects of the proposed method for building detection. Ključne besede: building detection, cost of building partition, deep neural network, remote sensing, tilt correction Objavljeno v DKUM: 26.09.2024; Ogledov: 0; Prenosov: 1
Celotno besedilo (8,62 MB) Gradivo ima več datotek! Več... |
5. Recycling of acetate and ammonium from digestate for single cell protein production by a hybrid electrochemical-membrane fermentation processDanfei Zeng, Yufeng Jiang, Carina Schneider, Yanyan Su, Claus Hélix-Nielsen, Yifeng Zhang, 2023, izvirni znanstveni članek Ključne besede: resource reclamation, single cell protein, electrodialysis, forward osmosis, Saccharomyces cerevisiae Objavljeno v DKUM: 10.05.2024; Ogledov: 180; Prenosov: 20
Celotno besedilo (1,71 MB) Gradivo ima več datotek! Več... |
6. A new method for testing the anti-permeability strength of clay failure under a high water pressureFu-wei Jiang, Ming-tang Lei, Xiao-zhen Jiang, 2015, izvirni znanstveni članek Opis: It is difficult to judge the failure of clay seepage under a high water pressure.This paper presents a new method to assess clay failure based on the anti-permeability strength, which is the critical water pressure to destroy the clay. An experiment is designed to test the value that avoids the problem of the time-consuming, traditional method to test clay seepage deformation. The experimental system and the process of testing are introduced in this paper. With a self-designed experimental system and method, 18 groups of sample were tested. The results show that the clay thickness and the seepage paths influence the anti-permeability strength. It also indicates that water infiltrates into the clay under the condition that its pressure exceeds a minimum value (P0). Ključne besede: clay failure, seepage deformation, anti-permeability strength, high water pressure Objavljeno v DKUM: 15.06.2018; Ogledov: 1410; Prenosov: 171
Celotno besedilo (249,24 KB) Gradivo ima več datotek! Več... |
7. Emergence of target waves in paced populations of cyclically competing speciesLuo-Luo Jiang, Tao Zhou, Matjaž Perc, Xin Huang, Bing-Hong Wang, 2009, izvirni znanstveni članek Opis: We investigate the emergence of target waves in a cyclic predator-prey model incorporating a periodic current of the three competing species in a small area situated at the center of a square lattice. The periodic current acts as a pacemaker, trying to impose its rhythm on the overall spatiotemporal evolution of the three species. We show that the pacemaker is able to nucleate target waves that eventually spread across the whole population, whereby three routes leading to this phenomenon can be distinguished depending on the mobility of the three species and the oscillation period of the localized current. First, target waves can emerge due to the synchronization between the periodic current and oscillations of the density of the three species on the spatial grid. The second route is similar to the first, the difference being that the synchronization sets in only intermittently. Finally, the third route toward target waves is realized when the frequency of the pacemaker is much higher than that characterizing the oscillations of the overall density of the three species. By considering the mobility and frequency of the current as variable parameters, we thus provide insights into the mechanisms of pattern formation resulting from the interplay between local and global dynamics in systems governed by cyclically competing species. Ključne besede: cyclical interactions, target waves, spatial games, diversity Objavljeno v DKUM: 30.06.2017; Ogledov: 1555; Prenosov: 399
Celotno besedilo (2,81 MB) Gradivo ima več datotek! Več... |
8. Spreading of cooperative behaviour across interdependent groupsLuo-Luo Jiang, Matjaž Perc, 2013, izvirni znanstveni članek Opis: Recent empirical research has shown that links between groups reinforce individuals within groups to adopt cooperative behaviour. Moreover, links between networks may induce cascading failures, competitive percolation, or contribute to efficient transportation. Here we show that there in fact exists an intermediate fraction of links between groups that is optimal for the evolution of cooperation in the prisoners dilemma game. We consider individual groups with regular, random, and scale-free topology, and study their different combinations to reveal that an intermediate interdependence optimally facilitates the spreading of cooperative behaviour between groups. Excessive between-group links simply unify the two groups and make them act as one, while too rare between-group links preclude a useful information flow between the two groups. Interestingly, we find that between-group links are more likely to connect two cooperators than in-group links, thus supporting the conclusion that they are of paramount importance. Ključne besede: social dilemma, cooperation, public goods, biased utility, interdependent networks, statistical physics of social systems Objavljeno v DKUM: 23.06.2017; Ogledov: 1795; Prenosov: 426
Celotno besedilo (665,03 KB) Gradivo ima več datotek! Več... |
9. Motor unit characteristics after targeted muscle reinnervationTamás Kapelner, Ning Jiang, Aleš Holobar, Ivan Vujaklija, Aidan Roche, Dario Farina, Oskar Aszmann, 2016, izvirni znanstveni članek Opis: Targeted muscle reinnervation (TMR) is a surgical procedure used to redirect nerves originally controlling muscles of the amputated limb into remaining muscles above the amputation, to treat phantom limb pain and facilitate prosthetic control. While this procedure effectively establishes robust prosthetic control, there is little knowledge on the behavior and characteristics of the reinnervated motor units. In this study we compared the m. pectoralis of five TMR patients to nine able-bodied controls with respect to motor unit action potential (MUAP) characteristics. We recorded and decomposed high-density surface EMG signals into individual spike trains of motor unit action potentials. In the TMR patients the MUAP surface area normalized to the electrode grid surface (0.25 ± 0.17 and 0.81 ± 0.46, p < 0.001) and the MUAP duration (10.92 ± 3.89 ms and 14.03 ± 3.91 ms, p < 0.01) were smaller for the TMR group than for the controls. The mean MUAP amplitude (0.19 ± 0.11 mV and 0.14 ± 0.06 mV, p = 0.07) was not significantly different between the two groups. Finally, we observed that MUAP surface representation in TMR generally overlapped, and the surface occupied by motor units corresponding to only one motor task was on average smaller than 12% of the electrode surface. These results suggest that smaller MUAP surface areas in TMR patients do not necessarily facilitate prosthetic control due to a high degree of overlap between these areas, and a neural information—based control could lead to improved performance. Based on the results we also infer that the size of the motor units after reinnervation is influenced by the size of the innervating motor neuron. Ključne besede: target muscle reinnervation, motor unit, controlling muscles Objavljeno v DKUM: 19.06.2017; Ogledov: 1639; Prenosov: 395
Celotno besedilo (1,51 MB) Gradivo ima več datotek! Več... |
10. If cooperation is likely punish mildly: insights from economic experiments based on the snowdrift gameLuo-Luo Jiang, Matjaž Perc, Attila Szolnoki, 2013, izvirni znanstveni članek Opis: Punishment may deter antisocial behavior. Yet to punish is costly, and the costs often do not offset the gains that are due to elevated levels of cooperation. However, the effectiveness of punishment depends not only on how costly it is, but also on the circumstances defining the social dilemma. Using the snowdrift game as the basis, we have conducted a series of economic experiments to determine whether severe punishment is more effective than mild punishment. We have observed that severe punishment is not necessarily more effective, even if the cost of punishment is identical in both cases. The benefits of severe punishment become evident only under extremely adverse conditions, when to cooperate is highly improbable in the absence of sanctions. If cooperation is likely, mild punishment is not less effective and leads to higher average payoffs, and is thus the much preferred alternative. Presented results suggest that the positive effects of punishment stem not only from imposed fines, but may also have a psychological background. Small fines can do wonders in motivating us to chose cooperation over defection, but without the paralyzing effect that may be brought about by large fines. The later should be utilized only when absolutely necessary. Ključne besede: public goods, punishment, economic experiments, snowdrift game Objavljeno v DKUM: 19.06.2017; Ogledov: 1200; Prenosov: 409
Celotno besedilo (522,94 KB) Gradivo ima več datotek! Več... |