| | SLO | ENG | Piškotki in zasebnost

Večja pisava | Manjša pisava

Iskanje po katalogu digitalne knjižnice Pomoč

Iskalni niz: išči po
išči po
išči po
išči po
* po starem in bolonjskem študiju

Opcije:
  Ponastavi


1 - 10 / 25
Na začetekNa prejšnjo stran123Na naslednjo stranNa konec
1.
2.
New suptech tool of the predictive generation for insurance companies : the case of the European market
Timotej Jagrič, Daniel Zdolšek, Robert Horvat, Iztok Kolar, Niko Erker, Jernej Merhar, Vita Jagrič, 2023, izvirni znanstveni članek

Opis: Financial innovation, green investments, or climate change are changing insurers’ business ecosystems, impacting their business behaviour and financial vulnerability. Supervisors and other stakeholders are interested in identifying the path toward deterioration in the insurance company’s financial health as early as possible. Suptech tools enable them to discover more and to intervene in a timely manner. We propose an artificial intelligence approach using Kohonen’s self-organizing maps. The dataset used for development and testing included yearly financial statements with 4058 observations for European composite insurance companies from 2012 to 2021. In a novel manner, the model investigates the behaviour of insurers, looking for similarities. The model forms a map. For the obtained groupings of companies from different geographical origins, a common characteristic was discovered regarding their future financial deterioration. A threshold defined using the solvency capital requirement (SCR) ratio being below 130% for the next year is applied to the map. On the test sample, the model correctly identified on average 86% of problematic companies and 79% of unproblematic companies. Changing the SCR ratio level enables differentiation into multiple map sections. The model does not rely on traditional methods, or the use of the SCR ratio as a dependent variable but looks for similarities in the actual insurer’s financial behaviour. The proposed approach offers grounds for a Suptech tool of predictive generation to support early detection of the possible future financial distress of an insurance company.
Ključne besede: European insurance market, suptech, supervision, financial deterioration identification, neural networks
Objavljeno v DKUM: 25.03.2024; Ogledov: 79; Prenosov: 9
.pdf Celotno besedilo (2,55 MB)
Gradivo ima več datotek! Več...

3.
UAV Thermal Imaging for Unexploded Ordnance Detection by Using Deep Learning
Milan Bajić, Jr., Božidar Potočnik, 2023, izvirni znanstveni članek

Opis: A few promising solutions for thermal imaging Unexploded Ordnance (UXO) detection were proposed after the start of the military conflict in Ukraine in 2014. At the same time, most of the landmine clearance protocols and practices are based on old, 20th-century technologies. More than 60 countries worldwide are still affected by explosive remnants of war, and new areas are contaminated almost every day. To date, no automated solutions exist for surface UXO detection by using thermal imaging. One of the reasons is also that there are no publicly available data. This research bridges both gaps by introducing an automated UXO detection method, and by publishing thermal imaging data. During a project in Bosnia and Herzegovina in 2019, an organisation, Norwegian People's Aid, collected data about unexploded ordnances and made them available for this research. Thermal images with a size of 720 x 480 pixels were collected by using an Unmanned Aerial Vehicle at a height of 3 m, thus achieving a very small Ground Sampling Distance (GSD). One of the goals of our research was also to verify if the explosive war remnants' detection accuracy could be improved further by using Convolutional Neural Networks (CNN). We have experimented with various existing modern CNN architectures for object identification, whereat the YOLOv5 model was selected as the most promising for retraining. An eleven-class object detection problem was solved primarily in this study. Our data were annotated semi-manually. Five versions of the YOLOv5 model, fine-tuned with a grid-search, were trained end-to-end on randomly selected 640 training and 80 validation images from our dataset. The trained models were verified on the remaining 88 images from our dataset. Objects from each of the eleven classes were identified with more than 90% probability, whereat the Mean Average Precision (mAP) at a 0.5 threshold was 99.5%, and the mAP at thresholds from 0.5 to 0.95 was 87.0% up to 90.5%, depending on the model's complexity. Our results are comparable to the state-of-the-art, whereat these object detection methods have been tested on other similar small datasets with thermal images. Our study is one of the few in the field of Automated UXO detection by using thermal images, and the first that solves the problem of identifying more than one class of objects. On the other hand, publicly available thermal images with a relatively small GSD will enable and stimulate the development of new detection algorithms, where our method and results can serve as a baseline. Only really accurate automatic UXO detection solutions will help to solve one of the least explored worldwide life-threatening problems.
Ključne besede: unmanned aerial vehicle, unexploded ordnance, thermal imaging, UXOTi_NPA dataset, convolutional neural networks, deep learning
Objavljeno v DKUM: 12.02.2024; Ogledov: 171; Prenosov: 8
.pdf Celotno besedilo (16,94 MB)
Gradivo ima več datotek! Več...

4.
Accuracy is not enough: optimizing for a fault detection delay
Matej Šprogar, Domen Verber, 2023, izvirni znanstveni članek

Opis: This paper assesses the fault-detection capabilities of modern deep-learning models. It highlights that a naive deep-learning approach optimized for accuracy is unsuitable for learning fault-detection models from time-series data. Consequently, out-of-the-box deep-learning strategies may yield impressive accuracy results but are ill-equipped for real-world applications. The paper introduces a methodology for estimating fault-detection delays when no oracle information on fault occurrence time is available. Moreover, the paper presents a straightforward approach to implicitly achieve the objective of minimizing fault-detection delays. This approach involves using pseudo-multi-objective deep optimization with data windowing, which enables the utilization of standard deep-learning methods for fault detection and expanding their applicability. However, it does introduce an additional hyperparameter that needs careful tuning. The paper employs the Tennessee Eastman Process dataset as a case study to demonstrate its findings. The results effectively highlight the limitations of standard loss functions and emphasize the importance of incorporating fault-detection delays in evaluating and reporting performance. In our study, the pseudo-multi-objective optimization could reach a fault-detection accuracy of 95% in just a fifth of the time it takes the best naive approach to do so.
Ključne besede: artificial neural networks, deep learning, fault detection, accuracy, multi-objective optimization
Objavljeno v DKUM: 30.11.2023; Ogledov: 252; Prenosov: 11
.pdf Celotno besedilo (478,93 KB)
Gradivo ima več datotek! Več...

5.
Development of an open-source framework for automatic alignment of KORUZA free-space optical communication system : magistrsko delo
Nejc Klemenčič, 2022, magistrsko delo

Opis: This thesis aims to implement an open-source framework for the automatic alignment and tracking of the KORUZA v2 Pro free-space optical solution. Free-space optical systems are explored and current optical alignment and tracking solutions are analyzed. We use Neural Network-based object detection approaches to complement the essential collection of framework functionality. We train a Neural Network to detect KORUZA v2 Pro units with data gathered from currently deployed links. The out-of-the-box solution for automatic alignment and tracking can be freely modified and extended.
Ključne besede: free-space optics, automatic alignment, neural networks, object detection, open-source framework
Objavljeno v DKUM: 14.03.2022; Ogledov: 662; Prenosov: 76
.pdf Celotno besedilo (4,68 MB)

6.
Nature-inspired algorithms for hyperparameter optimization : magistrsko delo
Filip Glojnarić, 2019, magistrsko delo

Opis: This master thesis is focusing on the utilization of nature-inspired algorithms for hyperparameter optimization, how they work and how to use them. We present some existing methods for hyperparameter optimization as well as propose a novel method that is based on six different nature-inspired algorithms: Firefly algorithm, Grey Wolf Optimizer, Particle Swarm Optimization, Genetic algorithm, Differential Evolution, and Hybrid Bat algorithm. We also show the optimization results (set of hyperparameters) for each algorithm and we present the plots of the accuracy for each combination and handpicked one. In discussion of the results, we provide the answers on our research questions as well as propose ideas for future work.
Ključne besede: artificial intelligence, artificial neural networks, machine learning, nature-inspired algorithms, evolutionary algorithms
Objavljeno v DKUM: 09.12.2019; Ogledov: 1905; Prenosov: 114
.pdf Celotno besedilo (969,13 KB)

7.
FORMULATION, PREPARATION AND CHARACTERIZATION OF NANOEMULSIONS FOR PARENTERAL NUTRITION : doctoral disertation
Dušica Mirković, 2019, doktorska disertacija

Opis: The aim of this doctoral research was to develop and optimize parenteral nanoemulsions as well as the total parenteral nutrition (TPN) admixture containing a nanoemulsion obtained in the course of the optimization process (hereinafter referred to as optimal nanoemulsion), and to examine their physicochemical and biological quality as well. In addition, the quality of the prepared nanoemulsions was compared with the quality of the industrial nanoemulsion (Lipofundin® MCT/LCT 20%), and, in the end, the TPN admixture initially prepared was also compared with the admixture into which the industrial emulsion was incorporated. Parenteral nanoemulsions that were considered in this dissertation were prepared by the high-pressure homogenization method. This method is the most widely applied method for the production of nanoemulsions due to the shortest length of homogenization time, the best-obtained homogeneity of the product and the smallest droplet diameter. For the nanoemulsion formulation, preparation and optimization purposes, by using, firstly, the concept of the computer-generated fractional design, and, after that, the full experimental design, the assessment of both direct effects of different formulation and process parameters (the oil phase type, the emulsifier type and concentration, a number of homogenization cycles and the pressure under which homogenization was carried out) as well as the effects of their interactions on the characteristics of prepared nanoemulsions was performed. Monitoring the nanoemulsion physical and chemical stability parameters was carried out immediately after their preparation, and then after 10, 30 and 60 days. It included the visual inspection, the measurement of the droplet diameter (the mean and volume droplet diameter), the polydispersity index, the ζ-potential, the pH value, the electrical conductivity, and the peroxide number. After the preparation and after 60 days, the biological evaluation (the sterility test and the endotoxic test) of the prepared nanoemulsions was carried out. As far as the characterization of the TPN admixture is concerned, it included practically the same parameters. The dynamics of monitoring the characteristics of the TPN admixture was determined on the basis of practical needs of hospitalized patients (0h, 24h and 72h). The scope and comprehensiveness of this issue indicated the need to divide the doctoral dissertation into three basic stages. The first stage was preliminary. Using the 24-1 fractional factorial design, nanoemulsions for the parenteral nutrition were prepared. They contained either a combination of soybean and fish oil, or a combination of medium chain triglycerides and fish oil. In addition, the type and the amount of an emulsifier used, a number of high-pressure homogenization cycles, and the homogenization pressure, were also varied. The measurement of the above-mentioned parameters for the industrial nanoemulsion was parallely carried out (Lipofundin® MCT/LCT 20%). The objective of this part of the research was to identify critical numerical factors having the most significant effect on the characteristics that define the prepared parenteral nanoemulsions. Parameters that were singled out as the result of this stage of the research (the emulsifier concentration and a number of homogenization cycles) were used as independent variables in the second stage of the research.
Ključne besede: nanoemulsions, total parenteral nutrition admixtures, high pressure homogenization, design of experiments, optimization, analysis of variance, artificial neural networks
Objavljeno v DKUM: 07.06.2019; Ogledov: 11805; Prenosov: 17
.pdf Celotno besedilo (2,82 MB)

8.
9.
The accuracy of the germination rate of seeds based on image processing and artificial neural networks
Uroš Škrubej, Črtomir Rozman, Denis Stajnko, 2015, izvirni znanstveni članek

Opis: This paper describes a computer vision system based on image processing and machine learning techniques which was implemented for automatic assessment of the tomato seed germination rate. The entire system was built using open source applications Image J, Weka and their public Java classes and linked by our specially developed code. After object detection, we applied artificial neural networks (ANN), which was able to correctly classify 95.44% of germinated seeds of tomato (Solanum lycopersicum L.).
Ključne besede: image processing, artificial neural networks, seeds, tomato
Objavljeno v DKUM: 14.11.2017; Ogledov: 1449; Prenosov: 444
.pdf Celotno besedilo (353,43 KB)
Gradivo ima več datotek! Več...

10.
Analysis of neural network responses in calibration of microsimulation traffic model
Irena Ištoka Otković, Damir Varevac, Matjaž Šraml, 2015, izvirni znanstveni članek

Opis: Microsimulation models are frequently used in traffic analysis. Various optimization methods are used in calibration, and the one method that has shown success is neural networks. This paper shows the responses of neural networks during calibration of a microsimulation traffic model. We analyzed two calibration methods by applying neural networks and comparing their neural network learning (according to their achieved correlation and the mean error of prediction) and their generalization ability (comparison of generalization results was analyzed in two steps). The best correlation between the microsimulation results and neural network prediction was 88.3%, achieved for the traveling time prediction, on which the first calibration method is based.
Ključne besede: microsimulation traffic models, calibration, response of neural networks, traveling time, queue parameters
Objavljeno v DKUM: 02.08.2017; Ogledov: 1246; Prenosov: 374
.pdf Celotno besedilo (1,07 MB)
Gradivo ima več datotek! Več...

Iskanje izvedeno v 0.2 sek.
Na vrh
Logotipi partnerjev Univerza v Mariboru Univerza v Ljubljani Univerza na Primorskem Univerza v Novi Gorici