Catalogue Search | MBRL
Are you sure you want to remove the book from the shelf?
{{itemTitle}}
488
result(s) for
"Jamali, Ali"
Sort by:
Swin Transformer and Deep Convolutional Neural Networks for Coastal Wetland Classification Using Sentinel-1, Sentinel-2, and LiDAR Data
The use of machine learning algorithms to classify complex landscapes has been revolutionized by the introduction of deep learning techniques, particularly in remote sensing. Convolutional neural networks (CNNs) have shown great success in the classification of complex high-dimensional remote sensing imagery, specifically in wetland classification. On the other hand, the state-of-the-art natural language processing (NLP) algorithms are transformers. Although the transformers have been studied for a few remote sensing applications, the integration of deep CNNs and transformers has not been studied, particularly in wetland mapping. As such, in this study, we explore the potential and possible limitations to be overcome regarding the use of a multi-model deep learning network with the integration of a modified version of the well-known deep CNN network of VGG-16, a 3D CNN network, and Swin transformer for complex coastal wetland classification. Moreover, we discuss the potential and limitation of the proposed multi-model technique over several solo models, including a random forest (RF), support vector machine (SVM), VGG-16, 3D CNN, and Swin transformer in the pilot site of Saint John city located in New Brunswick, Canada. In terms of F-1 score, the multi-model network obtained values of 0.87, 0.88, 0.89, 0.91, 0.93, 0.93, and 0.93 for the recognition of shrub wetland, fen, bog, aquatic bed, coastal marsh, forested wetland, and freshwater marsh, respectively. The results suggest that the multi-model network is superior to other solo classifiers from 3.36% to 33.35% in terms of average accuracy. Results achieved in this study suggest the high potential for integrating and using CNN networks with the cutting-edge transformers for the classification of complex landscapes in remote sensing.
Journal Article
Land use land cover mapping using advanced machine learning classifiers
Due to the recent climate changes such as floods and droughts, there is a need for Land Use Land Cover (LULC) mapping to monitor environmental changes that have effects on ecology, policy management, health and disaster management. As such, in this study, two well-known machine learning classifiers, namely, Support Vector Machine (SVM) and Random Forest (RF), are used for land cover mapping. In addition, two advanced deep learning algorithms, namely, the GAMLP and FSMLP, that are based on the Multi-layer Perceptron (MLP) function are developed in MATLAB programming language. The GAMLP uses a Genetic Algorithm (GA) to optimise parameters of the MLP function and, on the other hand, the FSMLP uses a derivative-free function for optimisation of the MLP function parameters. Three different scenarios using Landsat-8 imagery with spatial resolutions of 30 and 15 m are defined to investigate the effects of data pre-processing on the final predicted LULC map. Results based on the statistical indices, including overall accuracy (OA) and kappa index, show that the developed MLP-based algorithms have relatively high accuracies with higher than 98% correct classification. Besides the statistical indices, final LULC maps are interpreted visually where the GAMLP and FSMLP give the best results for the pre-processed Landsat-8 imagery with a spatial resolution of 15 m, but they have the worst outcomes for the unprocessed Landsat-8 imagery compared to SVM and RF classifiers visually and statistically.
Journal Article
How to Learn More? Exploring Kolmogorov–Arnold Networks for Hyperspectral Image Classification
Convolutional neural networks (CNNs) and vision transformers (ViTs) have shown excellent capability in complex hyperspectral image (HSI) classification. However, these models require a significant number of training data and are computational resources. On the other hand, modern Multi-Layer Perceptrons (MLPs) have demonstrated a great classification capability. These modern MLP-based models require significantly less training data compared with CNNs and ViTs, achieving state-of-the-art classification accuracy. Recently, Kolmogorov–Arnold networks (KANs) were proposed as viable alternatives for MLPs. Because of their internal similarity to splines and their external similarity to MLPs, KANs are able to optimize learned features with remarkable accuracy, in addition to being able to learn new features. Thus, in this study, we assessed the effectiveness of KANs for complex HSI data classification. Moreover, to enhance the HSI classification accuracy obtained by the KANs, we developed and proposed a hybrid architecture utilizing 1D, 2D, and 3D KANs. To demonstrate the effectiveness of the proposed KAN architecture, we conducted extensive experiments on three newly created HSI benchmark datasets: QUH-Pingan, QUH-Tangdaowan, and QUH-Qingyun. The results underscored the competitive or better capability of the developed hybrid KAN-based model across these benchmark datasets over several other CNN- and ViT-based algorithms, including 1D-CNN, 2DCNN, 3D CNN, VGG-16, ResNet-50, EfficientNet, RNN, and ViT.
Journal Article
A Cloud-Based Framework for Large-Scale Monitoring of Ocean Plastics Using Multi-Spectral Satellite Imagery and Generative Adversarial Network
Marine debris is considered a threat to the inhabitants, as well as the marine environments. Accumulation of marine debris, besides climate change factors, including warming water, sea-level rise, and changes in oceans’ chemistry, are causing the potential collapse of the marine environment’s health. Due to the increase of marine debris, including plastics in coastlines, ocean and sea surfaces, and even in deep ocean layers, there is a need for developing new advanced technology for the detection of large-sized marine pollution (with sizes larger than 1 m) using state-of-the-art remote sensing and machine learning tools. Therefore, we developed a cloud-based framework for large-scale marine pollution detection with the integration of Sentinel-2 satellite imagery and advanced machine learning tools on the Sentinel Hub cloud application programming interface (API). Moreover, we evaluated the performance of two shallow machine learning algorithms of random forest (RF) and support vector machine (SVM), as well as the deep learning method of the generative adversarial network-random forest (GAN-RF) for the detection of ocean plastics in the pilot site of Mytilene Island, Greece. Based on the obtained results, the shallow algorithms of RF and SVM achieved an overall accuracy of 88% and 84%, respectively, with available training data of plastic debris. The GAN-RF classifier improved the detection of ocean plastics of the RF method by 8%, achieving an overall accuracy of 96% by generating several synthetic ocean plastic samples.
Journal Article
A Dual-Modal Mixture-of-Experts Attention U-Net (DMoE-AttU-Net) for Change Detection Using Heterogeneous Optical and SAR Remote Sensing Images
Binary change detection (BCD) using heterogeneous optical and SAR imagery faces challenges due to modality-specific noise and the lack of adaptive fusion strategies. Existing methods often fail to suppress SAR speckle noise and accurately localize fine boundaries. This study proposes a novel deep architecture, termed Dual-Modal Mixture-of-Experts Attention U-Net (DMoE-AttU-Net), featuring (i) dual-stream encoders for modality-specific feature extraction, (ii) a mixture-of-experts (MoE) module in the SAR stream with a gating network for dynamic fusion, (iii) Squeeze-and-Excitation (SE) and spatial attention mechanisms in the decoder, and (iv) hierarchical skip connections for multi-scale fusion. Unlike existing multimodal change detection frameworks that apply uniform feature fusion, the proposed architecture introduces a modality-aware design in which the MoE mechanism is selectively applied to the SAR stream, enabling adaptive suppression of speckle noise while preserving complementary optical information. These components collectively enhance change localization and reduce noise-induced artifacts. The proposed model achieved a mean IoU of 0.855 and a kappa coefficient of 0.836 on three optical–SAR datasets, outperforming state-of-the-art methods in both accuracy and spatial consistency.
Journal Article
Internet of Things intrusion detection systems: a comprehensive review and future directions
The Internet of Things (IoT) is a paradigm that connects objects to the Internet as a whole and enables them to work together to achieve common objectives, such as innovative home automation. Potential attackers see the scattered and open IoT service structure as an appealing target for cyber-attacks. So, security cannot be dealt with independently. Security must be designed and built-in to every layer of the IoT system. IoT security concerns not only network and data security but also human health and life attacks. Therefore, the development of the loT system to provide security through resistance to attacks is a de facto requirement to make the loT safe and operational. Protecting these things is very important for system security. Plus, it is important to integrate the Intrusion Detection System (IDS) with IoT systems. IDS intends to track and analyze network traffic from different resources and detect malicious activities. It is a significant part of cybersecurity technology. In short, IDS is a process used to detect malicious activities against victims by several methods. Besides, the method of Systematic Literature Review (SLR) is used to classify, review, and incorporate results from all similar research that answers one or more IDS research topics and perform a detailed empirical research analysis on IDS techniques. Furthermore, depending on the detection technique, we classify IDS approaches in IoT as signature-based, anomaly-based, specification-based, and hybrid. Also, for the IDS approaches, the authors give a parametric comparison. The benefits and drawbacks of the chosen mechanisms are then addressed. Eventually, there is an analysis of open problems as well as potential trend directions.
Journal Article
Swin Transformer for Complex Coastal Wetland Classification Using the Integration of Sentinel-1 and Sentinel-2 Imagery
The emergence of deep learning techniques has revolutionized the use of machine learning algorithms to classify complicated environments, notably in remote sensing. Convolutional Neural Networks (CNNs) have shown considerable promise in classifying challenging high-dimensional remote sensing data, particularly in the classification of wetlands. State-of-the-art Natural Language Processing (NLP) algorithms, on the other hand, are transformers. Despite the fact that transformers have been utilized for a few remote sensing applications, they have not been compared to other well-known CNN networks in complex wetland classification. As such, for the classification of complex coastal wetlands in the study area of Saint John city, located in New Brunswick, Canada, we modified and employed the Swin Transformer algorithm. Moreover, the developed transformer classifier results were compared with two well-known deep CNNs of AlexNet and VGG-16. In terms of average accuracy, the proposed Swin Transformer algorithm outperformed the AlexNet and VGG-16 techniques by 14.3% and 44.28%, respectively. The proposed Swin Transformer classifier obtained F-1 scores of 0.65, 0.71, 0.73, 0.78, 0.82, 0.84, and 0.84 for the recognition of coastal marsh, shrub, bog, fen, aquatic bed, forested wetland, and freshwater marsh, respectively. The results achieved in this study suggest the high capability of transformers over very deep CNN networks for the classification of complex landscapes in remote sensing.
Journal Article
Proposing a sequential iterative method based on a per-unit model for rotor shape optimization in permanent magnet assisted synchronous reluctance motors
Purpose
During the design process of synchronous reluctance motors (SynRMs), one crucial step, after its main dimensioning, is optimizing the rotor geometry for maximum average torque and minimum torque ripple. However, because of the complexity of rotor flux-barrier layers geometry, the number of rotor geometrical parameters is high and this step could be quite complex and time-consuming. To obtain a good performance, one needs a robust algorithm to optimize the rotor geometry. The purpose of this paper is to present a sequential iterative method for rotor shape optimization in SynRMs based on the per-unit rotor model to maximize the average torque and minimize the torque ripple.
Design/methodology/approach
In the presented method, at first, rotor geometrical parameters are classified into several groups based on their geometrical similarities, and then optimization is done on these individual groups iteratively. The method starts with an arbitrary feasible rotor geometry and proceeds to optimize it. Because the method’s performance depends on initial rotor geometry, different cases are studied to investigate the convergence and robustness of the method. The MATLAB software is used to implement the optimization algorithm, and the ANSYS Maxwell software is used for the finite element analysis.
Findings
The performance of the proposed method is studied on a three-phase 0.75 kW-1,500 rpm permanent magnet assisted SynRM. The results show that the method improves the average torque while reducing the torque ripple. Even if the method starts with an inappropriate initial rotor geometry, it is robust enough and converges within an acceptable number of iterations.
Originality/value
The value of this paper is in introducing a per-unit rotor model. When the authors optimize the rotor geometry for a specific motor rating, it can be scaled up or down for other ratings with little effort. In this work, the number of rotor poles is four and the number of rotor flux-barrier layers per pole is three. Other combinations could be analyzed in future studies.
Journal Article
An adaptive method for calculation of iron losses in switched reluctance motors using a minimum number of magnetostatic finite element simulations
Purpose
This paper aims to present an adaptive method based on finite element analysis to calculate iron losses in switched reluctance motors (SRMs). Calculation of iron losses by analytical formulas has limited accuracy. On the other hand, its estimation in rotating electrical machines through fully dynamic simulations with a fine time-step is time-consuming. However, in the initial design process, a quick and sufficiently accurate method, i.e. a value close to that of iron losses, is always welcome. The method presented in this paper is a semi-analytical approach. The main problem is that iron losses depend on d B/d t. Therefore, the accuracy of the calculation of iron losses depends on the accuracy of the calculation of the first derivative of the flux density waveform. When adopting a magnetostatic model to estimate the iron losses, an important question arises: by how many magnetostatic simulations can the iron losses be estimated within the desired accuracy? In the proposed algorithm, the aim is not to accurately calculate the value of iron losses in SRMs. The objective is to find a numerical error criterion to calculate iron losses in SRMs with a minimum number of magnetostatic simulations.
Design/methodology/approach
A finite element solver is developed by authors in MATLAB to solve the 2 D nonlinear magnetostatic problem using the Newton–Raphson method. A parametric program is developed to create geometry and mesh. The proposed method is implemented in MATLAB using the developed solver. Counterpart simulations are done in the ANSYS Maxwell software to validate the accuracy of the results generated by the developed solver.
Findings
The performance of the proposed method is studied on a 12/8 (500 W) SRM. Three scenarios are studied. The first one is the calculation of iron losses by uniform refinement, and the second one is by adaptive refinement, and the last one is by adaptive refinement started by particular initial points (switching points). According to the results, the proposed method substantially reduces the number of magnetostatic simulations without sacrificing accuracy.
Originality/value
The main novelty of this paper is introducing an error criterion to find the minimum number of magnetostatic simulations that are needed to calculate iron losses with the desired accuracy.
Journal Article