Learning Architecture Research Articles

Effect of different weight initialization strategies on transfer learning for plant disease detection

AbstractThe weight initialization technique for transfer learning refers to the practice of using pretrained models that can be modified to solve new problems, instead of starting the training process from scratch. In this study, six different transfer learning weight initialization strategies were proposed for plant disease detection: scratch (i.e., random initialization), pretrained model on cross‐domain (ImageNet), model trained on related domain (ISIC 2019), model trained on related domain (ISIC 2019) with cross‐domain (ImageNet) weights, model trained on same domain (PlantVillage), and model trained on same domain (PlantVillage) with cross‐domain weights (ImageNet). Weights from each strategy were transferred to a target dataset (Plant Pathology 2021). These strategies were implemented using eight deep learning architectures. It was observed that transferring from any strategy led to an average acceleration of convergence ranging from 33.88% to 73.16% in mean loss and an improvement of 8.72%–42.12% in mean F1‐score compared to the scratch strategy. Moreover, although smaller and less comprehensive than ImageNet, transferring information from the same domain or related domain proved to be competitive compared to transferring from ImageNet. This indicates that ImageNet, which is widely favoured in the literature, may not necessarily represent the most optimal transfer source for the given context. In addition, to identify which strategies have significant differences, a post hoc analysis using Tukey's HSD test was conducted. Finally, the classifications made by the proposed models were visualized using Grad‐CAM to provide a qualitative understanding of how different weight initialization strategies affect the focus areas of the models.

Personalized federated learning for improving radar based precipitation nowcasting on heterogeneous areas

AbstractThe increasing generation of data in different areas of life, such as the environment, highlights the need to explore new techniques for processing and exploiting data for useful purposes. In this context, artificial intelligence techniques, especially through deep learning models, are key tools to be used on the large amount of data that can be obtained, for example, from weather radars. In many cases, the information collected by these radars is not open, or belongs to different institutions, thus needing to deal with the distributed nature of this data. In this work, the applicability of a personalized federated learning architecture, which has been called adapFL, on distributed weather radar images is addressed. To this end, given a single available radar covering 400 km in diameter, the captured images are divided in such a way that they are disjointly distributed into four different federated clients. The results obtained with adapFL are analyzed in each zone, as well as in a central area covering part of the surface of each of the previously distributed areas. The ultimate goal of this work is to study the generalization capability of this type of learning technique for its extrapolation to use cases in which a representative number of radars is available, whose data can not be centralized due to technical, legal or administrative concerns. The results of this preliminary study indicate that the performance obtained in each zone with the adapFL approach allows improving the results of the federated learning approach, the individual deep learning models and the classical Continuity Tracking Radar Echoes by Correlation approach.

Semi-Supervised Mixture Learning for Graph Neural Networks With Neighbor Dependence.

A graph neural network (GNN) is a powerful architecture for semi-supervised learning (SSL). However, the data-driven mode of GNNs raises some challenging problems. In particular, these models suffer from the limitations of incomplete attribute learning, insufficient structure capture, and the inability to distinguish between node attribute and graph structure, especially on label-scarce or attribute-missing data. In this article, we propose a novel framework, called graph coneighbor neural network (GCoNN), for node classification. It is composed of two modules: GCoNN Γ and GCoNN Γ° . GCoNN Γ is trained to establish the fundamental prototype for attribute learning on labeled data, while GCoNN Γ° learns neighbor dependence on transductive data through pseudolabels generated by GCoNN Γ . Next, GCoNN Γ is retrained to improve integration of node attribute and neighbor structure through feedback from GCoNN Γ° . GCoNN tends to convergence iteratively using such an approach. From a theoretical perspective, we analyze this iteration process from a generalized expectation-maximization (GEM) framework perspective which optimizes an evidence lower bound (ELBO) by amortized variational inference. Empirical evidence demonstrates that the state-of-the-art performance of the proposed approach outperforms other methods. We also apply GCoNN to brain functional networks, the results of which reveal response features across the brain which are physiologically plausible with respect to known language and visual functions.

TOOL CONDITION MONITORING IN MACHINING USING ROBUST RESIDUAL NEURAL NETWORKS

Tool condition monitoring (TCM) aims to improve process efficiency, quality and tool maintenance costs by monitoring critical variables such as tool wear. This study proposes a deep learning (DL) architecture based on process-informed robust residual networks (Robust-ResNet) to predict tool wear in milling processes using time series of internal computer numerical control (CNC) signals. The Robust-ResNet architecture uses skip connections to move through multiple convolutional layers, avoiding the vanishing gradient problem of other neural network algorithms. The study includes an evaluation of the binding of process information as input to the architecture and an attention mechanism between skips to make more robust predictions. The proposed architecture has been trained and optimised using an open access data set of face milling time series. In this particular case, AC and DC signals have been used together with the corresponding tool wear values. The results of this study demonstrate the benefits of using deep learning techniques in the prediction of tool wear using internal signals provided by the CNC itself. The implementation of the proposed architecture is expected to help reduce maintenance costs, improve product quality and increase production efficiency in milling manufacturing processes.

Deep SVBRDF Acquisition and Modelling: A Survey

AbstractHand in hand with the rapid development of machine learning, deep learning and generative AI algorithms and architectures, the graphics community has seen a remarkable evolution of novel techniques for material and appearance capture. Typically, these machine‐learning‐driven methods and technologies, in contrast to traditional techniques, rely on only a single or very few input images, while enabling the recovery of detailed, high‐quality measurements of bi‐directional reflectance distribution functions, as well as the corresponding spatially varying material properties, also known as Spatially Varying Bi‐directional Reflectance Distribution Functions (SVBRDFs). Learning‐based approaches for appearance capture will play a key role in the development of new technologies that will exhibit a significant impact on virtually all domains of graphics. Therefore, to facilitate future research, this State‐of‐the‐Art Report (STAR) presents an in‐depth overview of the state‐of‐the‐art in machine‐learning‐driven material capture in general, and focuses on SVBRDF acquisition in particular, due to its importance in accurately modelling complex light interaction properties of real‐world materials. The overview includes a categorization of current methods along with a summary of each technique, an evaluation of their functionalities, their complexity in terms of acquisition requirements, computational aspects and usability constraints. The STAR is concluded by looking forward and summarizing open challenges in research and development toward predictive and general appearance capture in this field. A complete list of the methods and papers reviewed in this survey is available at computergraphics.on.liu.se/star_svbrdf_dl/.

Crowdsourced Federated Learning Architecture with Personalized Privacy Preservation

Crowdsourced Federated Learning Architecture with Personalized Privacy Preservation

Aspect-based sentiment analysis: A dual-task learning architecture using imbalanced maximized-area under the curve proximate support vector machine and reinforcement learning

Aspect-based sentiment analysis (ABSA) deals with understanding sentiments about specific attributes. Aspect term extraction (ATE) plays an essential role in aspect sentiment classification (ASC) by specifying the main aspects for sentiment accuracy. This paper introduces a dual-task framework for ATE and ASC models in which, the ATE model uses the imbalanced maximized-area under the curve proximate support vector machine (ImAUC-PSVM) method to mitigate the imbalance problem and the output of ATE serves as the reward in the training process of ASC supported by RL (reinforcement learning). The ImAUC-PSVM uses the conventional PSVM (proximate support vector machine) strength for sentiment tasks and the AUC (area under the curve) optimization for imbalance addressing. The ASC model employs multiple kernels for word embedding, with a transductive long short-term memory (TLSTM) for sentence representation that considers samples near the test point for better model refinement. The policy of the RL-based training strategy of the ASC model is represented by a multilayer perceptron (MLP) that categorizes sentences into positive, negative and neutral sentiments. In this RL context, aspect terms serve as rewards in the training process. The scope loss function (SLF) is also integrated within RL to rectify dataset imbalances. To optimize ATE and ASC models, an artificial bee colony (ABC) technique is incorporated for hyperparameter optimization. Experiments and evaluations conducted on English (Restaurant and Laptop) and Hindi datasets show the superiority of the proposed model over existing models of the literature. Moreover, transfer learning (TL) is applied to a Twitter dataset to assess its effect on ATE. The F-measures of the ATE model are obtained as 90.12 and 84.10 and the accuracies of the ASC model are calculated as 84.10 and 82.02 on Restaurant and Laptop datasets, respectively.

Higher Order Polynomial Transformer for Fine-Grained Freezing of Gait Detection.

Freezing of Gait (FoG) is a common symptom of Parkinson's disease (PD), manifesting as a brief, episodic absence, or marked reduction in walking, despite a patient's intention to move. Clinical assessment of FoG events from manual observations by experts is both time-consuming and highly subjective. Therefore, machine learning-based FoG identification methods would be desirable. In this article, we address this task as a fine-grained human action recognition problem based on vision inputs. A novel deep learning architecture, namely, higher order polynomial transformer (HP-Transformer), is proposed to incorporate pose and appearance feature sequences to formulate fine-grained FoG patterns. In particular, a higher order self-attention mechanism is proposed based on higher order polynomials. To this end, linear, bilinear, and trilinear transformers are formulated in pursuit of discriminative fine-grained representations. These representations are treated as multiple streams and further fused by a cross-order fusion strategy for FoG detection. Comprehensive experiments on a large in-house dataset collected during clinical assessments demonstrate the effectiveness of the proposed method, and an area under the receiver operating characteristic (ROC) curve (AUC) of 0.92 is achieved for detecting FoG.

A Deep Learning Architectures for Kidney Disease Classification

Deep learning has become an extremely powerful tool for complex tasks such as image classification and segmentation. The medical industry often lacks high-quality, balanced datasets, which can be a challenge for deep learning algorithms that need sufficiently large amounts of data to train and increase their performance. This is especially important in the context of kidney issues such as for stones, cysts and tumors. We used deep learning models for this study to classify or detect several types of kidney diseases. We use different classification models, such as VGG-19, (CNNs) Convolutional Neural Networks, ResNet- 101, VGG-16, ResNet-50, and DenseNet-169, which can be enhanced through techniques such as classification, segmentation, and transfer learning. These algorithms can help improve model accuracy by allowing them to learn from multiple datasets. This technique has the potential to revolutionize the diagnosis and treatment of kidney problems as it enables more accurate and effective classification of CT-scan images. This may ultimately lead to better patient outcomes and improved overall health outcomes. INDEX TERMS Kidney disease, deep learning, image classification, convolutional neural networks (CNNs), segmentation, transfer learning, CT-scan images.

Enhanced ResNet50 Model for Aqueous Solubility Prediction of Drug Compounds Using Deep Learning Techniques

Background: Aqueous Solubility (AS) is a critical factor in drug discovery (DD), directly influencing a drug’s bioavailability and overall efficacy. Accurate prediction of AS remains a challenge despite the advancement in machine learning techniques, which are essential for improving the pharmacokinetics and formulation of new compounds. Methods: This study determines an enhanced ResNet50 deep learning architecture for predicting AS in drug compounds. Deep-net models with 8, 16, and 20-layer ResNet50 Convolutional Neural Network (CNN) architectures were developed. A dataset of 9,532 drug compounds, represented by molecular footprints, was used to train the models. The training process utilized a ten-fold cross-validation technique to optimize the model's predictive performance. Results: The 20-layer ResNet50 model outperformed human experts and shallower models, achieving an R² value of 0.423 and an RMSE of 0.678. The model also demonstrated an impressive ASP accuracy rate of 90.6%, significantly surpassing the predictions made by human experts and simpler neural network models. Conclusion: This study demonstrates that deeper-net architectures, particularly the 20-layer ResNet50 model, offer superior performance in predicting AS. These deep learning models provide a reliable and efficient solution for improving solubility predictions, crucial for advancing drug discovery efforts.

RetNet30: A Novel Stacked Convolution Neural Network Model for Automated Retinal Disease Diagnosis

ABSTRACTAutomated diagnosis of retinal diseases holds significant promise in enhancing healthcare efficiency and patient outcomes. However, existing methods often lack the accuracy and efficiency required for timely disease detection. To address this gap, we introduce RetNet30, a novel stacked convolutional neural network (CNN) designed to revolutionize automated retinal disease diagnosis. RetNet30 combines a custom‐built 30‐layer CNN with a fine‐tuned Inception V3 model, integrating these sub‐models through logistic regression to achieve superior classification performance. Extensive evaluations on retinal image datasets such as DRIVE, STARE, CHASE_DB1, and HRF demonstrate significant improvements in accuracy, sensitivity, specificity, and area under the ROC curve (AUROC) when compared to conventional approaches. By leveraging advanced deep learning architectures, RetNet30 not only enhances diagnostic precision but also generalizes effectively across diverse datasets, establishing a new benchmark in retinal disease classification. This novel approach offers a highly efficient and reliable solution for early disease detection and patient management, addressing the limitations of manual examination methods. Through rigorous quantitative and qualitative assessments, our proposed method demonstrates its potential to significantly impact medical image analysis and improve healthcare outcomes. RetNet30 marks a major step forward in automated retinal disease diagnosis, showcasing the future of AI‐driven advancements in ophthalmology.

Deep Learning for Multilabel Classification of Coral Reef Conditions in the Indo‐Pacific Using Underwater Photo Transect Method

ABSTRACTBecause coral reef ecosystems face threats from human activities and climate change, coral reef conservation programmes are implemented worldwide. Monitoring coral health provides references for guiding conservation activities. However, current labour‐intensive methods result in a backlog of unsorted images, highlighting the need for automated classification. Few studies have simultaneously utilized accurate labels along with updated algorithms and datasets. This study aimed to create a dataset representing common coral reef conditions and associated stressors in the Indo‐Pacific. Concurrently, it assessed existing classification algorithms and proposed a new multilabel method for automatically detecting coral reef conditions and extracting ecological information. A dataset containing over 20,000 high‐resolution coral images of different health conditions and stressors was constructed based on the field survey. Seven representative deep learning architectures were tested on this dataset, and their performance was quantitatively evaluated using the F1 metric and the match ratio. Based on this evaluation, a new method utilizing the ensemble learning approach was proposed. The proposed method accurately classified coral reef conditions as healthy, compromised, dead and rubble; it also identified corresponding stressors, including competition, disease, predation and physical issues. This method can help develop the coral image archive, guide conservation activities and provide references for decision‐making for reef managers and conservationists. The proposed ensemble learning approach outperforms others on the dataset, showing state‐of‐the‐art (SOTA) performance. Future research should improve its generalizability and accuracy to support global coral reef conservation efforts.

Hand-Crafted Features With a Simple Deep Learning Architecture for Sensor-Based Human Activity Recognition

Hand-Crafted Features With a Simple Deep Learning Architecture for Sensor-Based Human Activity Recognition

Improving vertebral diagnosis in computed tomography scans: a clinically oriented attention-driven asymmetric convolution network for segmentation

ObjectiveVertebral segmentation in computed tomography (CT) images remains an essential issue in medical image analysis, stemming from the variability of vertebral shapes, high complex deformations, and the inherent ambiguity in CT scans. The purpose of this study is to develop advanced methods to effectively address this challenging task. MethodsWe proposed an attention-driven asymmetric convolution deep learning (AACDL) framework for vertebral segmentation. Specifically, our approach involves constructing a novel asymmetric convolutional U-shaped deep learning architecture to enhance feature extraction capabilities by increasing its depth for capturing richer spatial details. Further, we construct a pyramid global context module that integrates global context information through pyramid pooling to boost segmentation accuracy particularly in smaller anatomical regions. Sequential channel and spatial attention mechanisms are also implemented within the network to enable it to automatically concentrate on learning the most salient features and regions across different dimensions. ResultsThe performance precision of our network was rigorously assessed using a suite of four benchmark metrics: the Dice coefficient, mean intersection over union (mIoU), Precision rate, and F1-score. When compared against the ground truth (GT), our model delivered outstanding scores, attaining a Dice coefficient of 82.79%, an mIoU of 90.72%, a Precision rate of 90.19%, and an F1-score of 90.09%, each reflecting the commendable accuracy and reliability of our network's segmentation output. ConclusionThe proposed AACDL method successfully realizes accurate segmentation of vertebral CT images, thereby demonstrating significant potential for clinical applications with its robust performance metrics. Its ability to handle the complexities associated with vertebral segmentation paves the way for enhanced diagnostic and treatment planning processes in healthcare settings.

An optimised deep learning approach for alzheimer’s disease classification

<p>Alzheimer’s disease (AD) is a progressive and incurable brain disorder. It starts out subtly and gets worse with time. 60 to 70 percent of dementia cases are brought on by this illness. An Alzheimer’s patient is diagnosed every two seconds, according to research. The complexity of the brain makes it often very challenging to identify in elderly people. In the area of medical imaging, deep learning is growing. Several deep learning techniques that attempted to identify and categorise the magnetic resonance imaging (MRI) brain images into four stages of AD will be compared in this work. 6400 MRI brain images were extracted from a dataset and divided into training, validation, and testing datasets. In our research on twelve deep learning architectures, inceptionV3 has given the best results with 99.56% and 97.75% accuracy on train and validation, respectively, and on test data, the model has achieved an accuracy of 95.81%. We trained the models using optimised ImageNet weights, which resulted in higher accuracy across all twelve models.</p>

Automated Classification System Based on YOLO Architecture for Body Condition Score in Dairy Cows.

Body condition score (BCS) is a common tool used to assess the welfare of dairy cows and is based on scoring animals according to their external appearance. If the BCS of dairy cows deviates from the required value, it can lead to diseases caused by metabolic problems in the animal, increased medication costs, low productivity, and even the loss of dairy cows. BCS scores for dairy cows on farms are mostly determined by observation based on expert knowledge and experience. This study proposes an automatic classification system for BCS determination in dairy cows using the YOLOv8x deep learning architecture. In this study, firstly, an original dataset was prepared by dividing the BCS scale into five different classes of Emaciated, Poor, Good, Fat, and Obese for images of Holstein and Simmental cow breeds collected from different farms. In the experimental analyses performed on the dataset prepared in this study, the BCS values of 102 out of a total of 126 cow images in the test set were correctly classified using the proposed YOLOv8x deep learning architecture. Furthermore, an average accuracy of 0.81 was achieved for all BCS classes in Holstein and Simmental cows. In addition, the average area under the precision-recall curve was 0.87. In conclusion, the BCS classification system for dairy cows proposed in this study may allow for the accurate observation of animals with rapid declines in body condition. In addition, the BCS classification system can be used as a tool for production decision-makers in early lactation to reduce the negative energy balance.

Complementary Analysis of Deep Learning Architectures for Brain Tumor Classification Using MRI Images

Complementary Analysis of Deep Learning Architectures for Brain Tumor Classification Using MRI Images

Deep hybrid approach with sequential feature extraction and classification for robust malware detection

Malware attacks have escalated significantly with an increase in the number of internet users and connected devices. With the increasingly different types of malware released by hackers, designing new and competitive techniques to detect advanced malware is essential. In the proposed research, we have developed a multi-level feature extraction technique using deep learning architectures and a classification model to classify malware families. The essential features from the malware images are extracted using the Gated Recurrent Unit in the first step, which are further fed to a Convolutional Neural Network model for extracting the final feature vector. The multi-level feature selection is followed by classification into various malware families using Cost-sensitive Boot Strapped Weighted Random Forest (CSBW-RF). The proposed approach gave promising results of 99.58 % accuracy in distinguishing the 25 different malware families on the Mallmg dataset. This hybrid model gave significantly better performance scores for classifying visually similar malware families. The generalizability of the proposed model is benchmarked with the popular Microsoft Big 2015 dataset and has achieved comparatively higher performance scores than many existing models. This benchmarking demonstrates the robustness and scalability of our approach. The use of cost-sensitive learning and bootstrapping techniques also contributed to the model’s ability to generalize well to new and unseen data. These enhancements ensure that our model can be effectively applied in diverse real-world scenarios, maintaining high performance across different environments and malware types. This research can contribute to detecting malware attacks and can be integrated in threat monitoring systems. The successful application of this hybrid model indicates its potential for deployment in real-world cybersecurity environments, providing a strong defense against evolving malware threats.

ResGAT: Residual Graph Attention Networks for molecular property prediction

Molecular property prediction is an important step in the drug discovery pipeline. Numerous computational methods have been developed to predict a wide range of molecular properties. While recent approaches have shown promising results, no single architecture can comprehensively address all tasks, making this area persistently challenging and requiring substantial time and effort. Beyond traditional machine learning and deep learning architectures for regular data, several deep learning architectures have been designed for graph-structured data to overcome the limitations of conventional methods. Utilizing graph-structured data in quantitative structure–activity relationship (QSAR) modeling allows models to effectively extract unique features, especially where connectivity information is crucial. In our study, we developed residual graph attention networks (ResGAT), a deep learning architecture for molecular graph-structured data. This architecture is a combination of graph attention networks and shortcut connections to address both regression and classification problems. It is also customizable to adapt to various dataset sizes, enhancing the learning process based on molecular patterns. When tested multiple times with both random and scaffold sampling strategies on nine benchmark molecular datasets, QSAR models developed using ResGAT demonstrated stability and competitive performance compared to state-of-the-art methods.

A Federated Learning Architecture for Blockchain DDoS Attacks Detection

A Federated Learning Architecture for Blockchain DDoS Attacks Detection