Hrčak - Portal of scientific journals of Croatia
Not a member yet
321498 research outputs found
Sort by
UMJETNA INTELIGENCIJA KAO ALAT ZA RAZVOJ POSLOVNIH STRATEGIJA I STVARANJE PREDNOSTI ZA VISOKOTEHNOLOŠKE STARTUPOVE
No full textThis study examines whether the strategic integration of artificial intelligence (AI) into managerial decision-making is associated with the development of sustainable competitive advantages. Addressing a gap in the literature on AI adoption in resource-constrained startup ecosystems, the research focuses on startups operating in the Baltic countries. Using a mixed-methods approach that combines survey data, financial analysis, and expert interviews, the study develops an original composite index to capture AI-driven competitive advantages across innovation, operational efficiency, and customer value. The empirical results demonstrate that deeper and more strategic AI integration is positively associated with stronger competitive advantages. The findings further reveal that this relationship is non-linear and significantly strengthened by access to financial resources. Overall, the study highlights that AI generates competitive benefits for startups only when implemented strategically and supported by adequate funding and human capital, offering important implications for startup founders, investors, and policymakers. Given the cross-sectional nature of the data, the reported relationships should be interpreted as associative rather than strictly causal.Ovo istraživanje ispituje je li strateška integracija umjetne inteligencije (UI) u menadžersko odlučivanje povezana s razvojem održivih konkurentskih prednosti. Uzimajući u obzir nedostatak u literaturi o primjeni umjetne inteligencije u startup ekosustavima s ograničenim resursima, istraživanje se usredotočuje na startupove u baltičkim zemljama. Primjenom mješovitog istraživačkog pristupa, koji kombinira anketne podatke, financijsku analizu i intervjue sa stručnjacima, studija razvija kompozitni indeks za mjerenje konkurentskih prednosti temeljenih na umjetnoj inteligenciji u okviru dimenzija inovacija, operativnih učinkovitosti i vrijednosti za korisnike. Empirijski rezultati pokazuju da je dublja i strateška integracija umjetne inteligencije pozitivno povezana s jačim konkurentskim prednostima. Nalazi dodatno otkrivaju da je taj odnos nelinearan i značajno ojačan dostupnošću financijskih resursa. U cjelini, istraživanje naglašava da UI donosi konkurentske koristi startupovima samo kada se implementira strateški i uz odgovarajuću financijsku potporu i ljudski kapital, nudeći važne implikacije za osnivače startupova, investitore i donositelje politika. S obzirom na presječni karakter podataka, utvrđeni odnosi trebaju se tumačiti kao asocijativni, a ne strogo uzročni
Spectral expansion for impulsive dynamic Dirac system on the whole line
No full textIn this study, we consider an impulsive dynamic Dirac system on the whole line. A spectral function of this system is constructed. We establish a Parseval equality and expansion formula in terms of the spectral function
Mechanical Recycling of Polyamide Trimmer Line: ASolutionoraWasteof Time?
No full textU ovom radu provedeno je mehaničko recikliranje niti za košnju u laboratorijskom dvopužnom ekstruderu. Procijenjen je utjecaj mehaničkog recikliranja na strukturu, toplinska svojstva i toplinsku postojanost materijala od kojeg su izrađene niti za košnju. Primjenom infracrvene spektroskopije s Fourierovom transformacijom (FT-IR) utvrđeno je da su niti izrađene od
poli(ε-kaprolaktama), odnosno poliamida 6 (PA6). Ekstrudiranje nije utjecalo na strukturu PA6. Toplinska svojstva niti za košnju određena su primjenom diferencijalne pretražne kalorimetrije (DSC). Rezultati su ukazali na to da je ekstrudiranje imalo utjecaj na pakiranje i slaganje lanaca, odnosno uzrokovalo početak kristalizacije PA6 iz taljevine pri višoj temperaturi. Utjecaj mehaničkog recikliranja na toplinsku postojanost niti za košnju istraživan je primjenom termogravimetrijske analize (TG). Temeljem vrijednosti karakterističnih parametara (Tonset i Tmax) može se zaključiti da mehaničko recikliranje ne narušava toplinsku postojanost PA6. Mehaničko recikliranje može biti potencijalno rješenje zbrinjavanja rabljenih niti za košnju.U ovom radu provedeno je mehaničko recikliranje niti za košnju u laboratorijskom dvopužnom ekstruderu. Procijenjen je utjecaj mehaničkog recikliranja na strukturu, toplinska svojstva i toplinsku postojanost materijala od kojeg su izrađene niti za košnju. Primjenom infracrvene spektroskopije s Fourierovom transformacijom (FT-IR) utvrđeno je da su niti izrađene od
poli(ε-kaprolaktama), odnosno poliamida 6 (PA6). Ekstrudiranje nije utjecalo na strukturu PA6. Toplinska svojstva niti za košnju određena su primjenom diferencijalne pretražne kalorimetrije (DSC). Rezultati su ukazali na to da je ekstrudiranje imalo utjecaj na pakiranje i slaganje lanaca, odnosno uzrokovalo početak kristalizacije PA6 iz taljevine pri višoj temperaturi. Utjecaj mehaničkog recikliranja na toplinsku postojanost niti za košnju istraživan je primjenom termogravimetrijske analize (TG). Temeljem vrijednosti karakterističnih parametara (Tonset i Tmax) može se zaključiti da mehaničko recikliranje ne narušava toplinsku postojanost PA6. Mehaničko recikliranje može biti potencijalno rješenje zbrinjavanja rabljenih niti za košnju
Učinci izvoza na regionalni razvoj Hrvatske
No full textThe aim of this paper is to analyze the role of exports in the regional development of Croatia. Specifically, the paper employs a panel econometric model to assess the impact of county-level exports on GDP, GDP per capita, and GDP per capita growth during the period 2004-2022. In addition to the main variable of interest, exports by county, several control variables are included, namely investment, employment, human capital (measured through education), and tourism, all of which also influence regional economic performance. Diagnostic tests were conducted to examine the statistical properties and validity of the models used. Based on the results, the fixed effects (FE) model was identified as the most appropriate method. Alongside the econometric analysis, the paper also employs descriptive statistics to outline the main features of Croatian exports at the county level. The findings suggest that exports, investment, employment, and education all have a significant influence on economic growth and income per capita at the regional level. The impact of exports was found to be both statistically significant and positive, indicating that foreign trade plays a crucial role in county-level economic development. Despite these important insights, the analysis also has certain limitations. First, there may be an endogeneity issue within the models. Second, while key macroeconomic variables are included, other potentially important regional development factors such as innovation, EU funding and institutional quality should be considered for future research. Third, the structural heterogeneity among counties (e.g., industrial versus tourism-based regions) could affect the consistency of the model. The main scientific contribution of this paper lies in its empirical confirmation that exports also play a vital role in enhancing economic performance and living standards at the regional level in Croatia.Cilj ovog rada je analizirati ulogu izvoza u regionalnom razvoju Hrvatske. Konkretnije, u radu će se koristiti panel ekonometrijski model kako bi se ocijenio utjecaj izvoza županija na njihov BDP, BDP per capita i rast BDP- per capita u periodu 2004-2022. Osim varijable od interesa, a to je izvoz po županijama, koristit će se i kontrolne varijable koje također utječu na regionalne ekonomske pokazatelje, a to su investicije, zaposlenost, ljudski kapital ili obrazovanje i turizam. U radu su provedeni dijagnostički testovi kako bi se utvrdila statistička svojstva i valjanost korištenih metoda te je utvrđeno da je prikladnije koristiti model s fiksnim efektima (FE). Osim ekonometrijske panel analize, u radu se deskriptivnom statistikom opisuju osnovne karakteristike hrvatskog izvoza po županijama. U konačnici, rezultati analize su pokazali da faktori poput izvoza, investicija, zaposlenosti i obrazovanja utječu na gospodarski rast i dohodak po stanovniku na razini županija. Učinak izvoza se pokazao statistički značajnim i pozitivnim što sugerira da vanjska trgovina igra ključnu ulogu u ekonomskom razvoju županija. Iako rezultati pružaju važne uvide, analiza u ovom radu ima i nekoliko ograničenja. Prvo, u modelima je moguć problem endogenosti. Drugo, potrebno je uključiti ne samo ključne makroekonomske varijable nego i ostale potencijalno bitne čimbenike regionalnog rata poput inovacija, EU fondova ili kvalitete institucija. Treće, u modelima je moguć problem heterogenosti jer se županije strukturno razlikuju među sobom (npr. turističke i industrijske županije). Glavni znanstveni doprinos ovog rada se temelji na empirijskoj potvrdi da izvoz i na regionalnoj razini doprinosi ekonomskom rastu i povećanju životnog standarda u Hrvatskoj
AI Tools in Programming Education: Student Perspectives and Usage Trends
No full textThis research investigates the expanding use of generative AI (Mahmoud et al., 2025) tools among computer programming students. Building on a 2023 study, a 2024 survey of 182 undergraduate and graduate students explored usage patterns, perceived effectiveness, and attitudes towards their academic use. Thematic analysis revealed a significant increase in AI adoption, with students frequently using tools for code generation, debugging, and explanation. While students value the speed and efficiency AI offers, they express significant concerns about output accuracy, over-reliance, and the impact on developing independent reasoning skills. Notably, attitudes shifted from favouring full allowance of AI tools to partial allowance for programming assignments, with distinct usage patterns observed across different study levels. These findings underscore the urgent need for educational institutions to establish clear guidelines. A balanced approach is required to integrate AI tools constructively while mitigating risks to academic integrity and deep learning
A Deep Learning-Based Framework for Robust Facial Keypoint Localization in Unconstrained Conditions
No full textFacial keypoint localization plays a critical role in facial recognition, security monitoring, and human-computer interaction. Traditional methods rely heavily on handcrafted features, making them sensitive to occlusions, lighting variations, and pose changes. This study proposes a deep learning-based framework integrating lightweight convolutional neural networks (CNNs) and Conditional Random Fields (CRFs) to improve keypoint detection and localization accuracy under unconstrained conditions. A fast connected convolutional layer is introduced in a cascaded network structure, significantly reducing feature space information loss and enhancing geometric relationship modeling. The results showed that the proposed face detection model had a small cumulative error value, with a feature recognition accuracy of over 0.9, and an average accuracy of over 90 for all classes under three different image conditions. The proposed localization model had smaller error values and a much lower error rate than other algorithms under various segmented image differences, effectively considering data feature differences and achieving higher localization accuracy. The proposed deep learning model can effectively achieve the fusion of output features and improve the effectiveness of facial keypoint localization
Chaos Mapping and Marine Predators Algorithm-Based Deep Learning Framework for Intrusion Detection in IIoT Networks
No full textThe Industrial Internet of Things (IIoT) extends IoT applications to industrial environments, driving significant enhancements in operational efficiency. However, this evolution brings heightened cybersecurity risks, posing challenges to the protection of IIoT systems. To address these issues, this study introduces a novel Chaos Mapping and Marine Predators Algorithm-based Deep Learning Intrusion Detection System (CMPADL-IDS). The proposed model employs a two-stage process: feature selection using Chaos Mapping and the Marine Predators Algorithm (CMPA) and anomaly detection using a Long Short-Term Memory Autoencoder (LSTM-AE). The CMPA effectively identifies optimal features by leveraging chaotic systems and MPA's intelligent optimization capabilities. For enhanced performance, Bayesian Optimization (BO) is employed to fine-tune LSTM-AE hyperparameters, optimizing detection accuracy and computational efficiency. The framework was tested on the ToN-IoT dataset and managed to reach an average accuracy of 98.40%, a precision of 80.30%, a recall of 77.80%, an F1-score of 78.62% and an AUC score of 88.80%. The evaluation proves that using the suggested feature selection and anomaly detection techniques improves IIoT network security more than existing methods
A PointPillars-based 3D point cloud object detector of USVs for small target detection in dynamic aquatic environments
Full text linkLiDAR, a crucial sensor for Unmanned Surface Vehicles (USVs), allows for precise 3D modelling but encounters challenges in real-time target detection due to sparse point clouds. Current 3D point cloud detectors struggle to effectively capture fine-grained details and dynamic water surface features, while high-performance models often rely on custom operators, making deployment more complicated. Additionally, current water surface datasets lack the resolution necessary for small target detection. To tackle these issues, this study enhances the PointPillars model with the Voxel-Guided Label Assignment (VGLA) strategy, improving feature extraction through adaptive label assignment. A high-resolution point cloud dataset focused on small aquatic objects has also been developed based on 128-beam LiDAR. The proposed PointPillars-VGLA achieves 3D AP scores of 89.50%, 83.70%, and 75.20%, as well as BEV AP scores of 95.20%, 91.00%, and 86.70% across three target categories. Ablation experiments confirm the effectiveness of the VGLA module, with accuracy gains of up to 2.27% over CenterPoint. Deployed on the Jetson AGX Orin with TensorRT, the model achieves real-time inference at 30 FPS, enabling efficient detection and tracking in dynamic aquatic environments
Development of collision-case-based testing scenarios for validating autonomous ship collision-avoidance algorithms
Full text linkThe criticality of collision-avoidance technology for ensuring safe navigation of autonomous ships necessitates diverse testing scenarios that reflect complex maritime environments. However, previous testing scenarios, often based on virtual trajectories or simplified encounters, have shown limitations in adequately representing real-world conditions. This study proposes a novel framework for developing collision-avoidance testing scenarios based on actual collision cases. The framework consists of three stages: collision case collection, trajectory extraction, and scenario development. Relevant data were extracted from selected cases, and the trajectories of ships influencing the collision were combined to reconstruct the circumstances at the time of the incident. Encounter situations were then diversified by altering the roles and positions of own and target ships, and finally systematically categorised into a structured testing set. Unlike previous testing scenarios, the developed scenarios exhibit distinctive characteristics derived from actual collision cases, including situations where navigation rules cannot be strictly applied, dynamic encounters, speed variations, and environmental conditions. By reflecting real maritime environments, these scenarios provide a solid basis for validating and improving collision-avoidance algorithms. The proposed framework is expected to contribute not only to the advancement of autonomous-ship technology but also to the enhancement of maritime safety
