| | 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 - 9 / 9
Na začetekNa prejšnjo stran1Na naslednjo stranNa konec
1.
Cross-Hole GPR for Soil Moisture Estimation Using Deep Learning
Blaž Pongrac, Dušan Gleich, Marko Malajner, Andrej Sarjaš, 2023, izvirni znanstveni članek

Opis: This paper presents the design of a high-voltage pulse-based radar and a supervised data processing method for soil moisture estimation. The goal of this research was to design a pulse-based radar to detect changes in soil moisture using a cross-hole approach. The pulse-based radar with three transmitting antennas was placed into a 12 m deep hole, and a receiver with three receive antennas was placed into a different hole separated by 100 m from the transmitter. The pulse generator was based on a Marx generator with an LC filter, and for the receiver, the high-frequency data acquisition card was used, which can acquire signals using 3 Gigabytes per second. Used borehole antennas were designed to operate in the wide frequency band to ensure signal propagation through the soil. A deep regression convolutional network is proposed in this paper to estimate volumetric soil moisture using time-sampled signals. A regression convolutional network is extended to three dimensions to model changes in wave propagation between the transmitted and received signals. The training dataset was acquired during the period of 73 days of acquisition between two boreholes separated by 100 m. The soil moisture measurements were acquired at three points 25 m apart to provide ground truth data. Additionally, water was poured into several specially prepared boreholes between transmitter and receiver antennas to acquire additional dataset for training, validation, and testing of convolutional neural networks. Experimental results showed that the proposed system is able to detect changes in the volumetric soil moisture using Tx and Rx antennas.
Ključne besede: ground penetrating radar, cross-hole, L-band, deep learning, convolutional neural network, soil moisture estimation
Objavljeno v DKUM: 03.04.2024; Ogledov: 184; Prenosov: 19
.pdf Celotno besedilo (3,22 MB)
Gradivo ima več datotek! Več...

2.
A graph pointer network-based multi-objective deep reinforcement learning algorithm for solving the traveling salesman problem
Jeewaka Perera, Shih-Hsi Liu, Marjan Mernik, Matej Črepinšek, Miha Ravber, 2023, izvirni znanstveni članek

Opis: Traveling Salesman Problems (TSPs) have been a long-lasting interesting challenge to researchers in different areas. The difficulty of such problems scales up further when multiple objectives are considered concurrently. Plenty of work in evolutionary algorithms has been introduced to solve multi-objective TSPs with promising results, and the work in deep learning and reinforcement learning has been surging. This paper introduces a multi-objective deep graph pointer network-based reinforcement learning (MODGRL) algorithm for multi-objective TSPs. The MODGRL improves an earlier multi-objective deep reinforcement learning algorithm, called DRL-MOA, by utilizing a graph pointer network to learn the graphical structures of TSPs. Such improvements allow MODGRL to be trained on a small-scale TSP, but can find optimal solutions for large scale TSPs. NSGA-II, MOEA/D and SPEA2 are selected to compare with MODGRL and DRL-MOA. Hypervolume, spread and coverage over Pareto front (CPF) quality indicators were selected to assess the algorithms’ performance. In terms of the hypervolume indicator that represents the convergence and diversity of Pareto-frontiers, MODGRL outperformed all the competitors on the three well-known benchmark problems. Such findings proved that MODGRL, with the improved graph pointer network, indeed performed better, measured by the hypervolume indicator, than DRL-MOA and the three other evolutionary algorithms. MODGRL and DRL-MOA were comparable in the leading group, measured by the spread indicator. Although MODGRL performed better than DRL-MOA, both of them were just average regarding the evenness and diversity measured by the CPF indicator. Such findings remind that different performance indicators measure Pareto-frontiers from different perspectives. Choosing a well-accepted and suitable performance indicator to one’s experimental design is very critical, and may affect the conclusions. Three evolutionary algorithms were also experimented on with extra iterations, to validate whether extra iterations affected the performance. The results show that NSGA-II and SPEA2 were greatly improved measured by the Spread and CPF indicators. Such findings raise fairness concerns on algorithm comparisons using different fixed stopping criteria for different algorithms, which appeared in the DRL-MOA work and many others. Through these lessons, we concluded that MODGRL indeed performed better than DRL-MOA in terms of hypervolumne, and we also urge researchers on fair experimental designs and comparisons, in order to derive scientifically sound conclusions.
Ključne besede: multi-objective optimization, traveling salesman problems, deep reinforcement learning
Objavljeno v DKUM: 28.03.2024; Ogledov: 137; Prenosov: 19
.pdf Celotno besedilo (7,89 MB)
Gradivo ima več datotek! Več...

3.
A VAN-Based Multi-Scale Cross-Attention Mechanism for Skin Lesion Segmentation Network
Shuang 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: 434; Prenosov: 305
.pdf Celotno besedilo (1,46 MB)
Gradivo ima več datotek! Več...

4.
Despeckling of SAR Images Using Residual Twin CNN and Multi-Resolution Attention Mechanism
Blaž Pongrac, Dušan Gleich, 2023, izvirni znanstveni članek

Opis: The despeckling of synthetic aperture radar images using two different convolutional neural network architectures is presented in this paper. The first method presents a novel Siamese convolutional neural network with a dilated convolutional network in each branch. Recently, attention mechanisms have been introduced to convolutional networks to better model and recognize features. Therefore, we propose a novel design for a convolutional neural network using an attention mechanism for an encoder–decoder-type network. The framework consists of a multiscale spatial attention network to improve the modeling of semantic information at different spatial levels and an additional attention mechanism to optimize feature propagation. Both proposed methods are different in design but they provide comparable despeckling results in subjective and objective measurements in terms of correlated speckle noise. The experimental results are evaluated on both synthetically generated speckled images and real SAR images. The methods proposed in this paper are able to despeckle SAR images and preserve SAR features.
Ključne besede: synthetic aperture radar, speckle, speckle suppression, despeckling, deep learning, convolutional neural network
Objavljeno v DKUM: 21.02.2024; Ogledov: 191; Prenosov: 12
.pdf Celotno besedilo (13,70 MB)
Gradivo ima več datotek! Več...

5.
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: 265; Prenosov: 17
.pdf Celotno besedilo (16,94 MB)
Gradivo ima več datotek! Več...

6.
7.
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: 311; Prenosov: 23
.pdf Celotno besedilo (478,93 KB)
Gradivo ima več datotek! Več...

8.
Reinforcement learning relocation assignment in the multiple-deep storage system
Jakob Marolt, Bojan Rosi, Tone Lerher, 2022, objavljeni znanstveni prispevek na konferenci

Opis: The field of reinforcement learning shows promising results in recent publications for solving complex combinatorial problems. In this paper, an agent is trained to tackle the relocation problem that appears in various logistics systems. The relocation problem was formulated as an array that is accessible only from the top side. The agent has to relocate blocking SKUs and thus enable access to the SKU with the highest dispatch priority. We utilised the Deep Q-learning Network (DQN) to train the agent. Two case studies are presented – one with 1.2 ∙ 10 and another with 2 ∙ 10 possible states. The results display that the agents made a significantly better decision toward the learning process’s end than at the beginning
Ključne besede: relocation problem, reinforcement learning, DQN, multiple-deep, storage system
Objavljeno v DKUM: 14.03.2023; Ogledov: 490; Prenosov: 13
.pdf Celotno besedilo (659,02 KB)

9.
Deep Learning on Low Power Embedded Devices Using RISC-V Cores with an Extended Instruction Set : master's thesis
Jure Vreča, 2020, magistrsko delo

Opis: This thesis explores the possibility of running neural networks on microcontrollers and how to optimize their performance using instruction set extensions. Microcontrollers are seen as too weak to run neural networks. We challenge this view and show that stripped-down neural networks can run and be useful for some applications. We used an open-source microcontroller called PULPino to run our neural network. The benefit of various instructions and optimizations for minimizing energy consumption to run deep learning algorithms was evaluated. Hardware loops, loop unrolling, and the dot-product unit were implemented and tested. We developed an FPGA-based testing system to evaluate our hardware. We also developed a deep learning library and a test neural network for our hardware. We wrote two versions of the deep learning library. One version is the reference code, and the other is the optimized code that uses the dot product unit. Using the testing system, we tested the performance of the two versions. The synthesis was run to determine the power and energy consumption. We also tried out various optimizations to see if the performance could be improved. Using instruction set extensions and algorithmic optimizations we reduced the clock cycle count by 72% for the convolutional layers and by 78% for fully-connected layers. This reduced power consumption by 73%. We compare our results with related research.
Ključne besede: deep learning, embedded system, instruction set, RISC-V
Objavljeno v DKUM: 03.11.2020; Ogledov: 1466; Prenosov: 154
.pdf Celotno besedilo (2,69 MB)

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