Deep Neural Network Techniques Research Articles

Adaptive deep neural networks for solving corner singular problems

Deep neural networks (DNNs) for numerical solutions to partial differential equations (PDEs) have exhibited their remarkable merits of meshless methods, dimensionless features, and nonlinear approximation powers. The majority of conventional DNNs are mainly concentrated on the low- or high-dimensional PDE with smooth solutions. In singular problems the DNN does not behave so efficiently as in the smooth problem because of low regularities of the solution. In this paper, we propose a novel adaptive DNN for the PDE with corner singular solutions. In the corner singular problem gradients of solution vary rapidly around some particular vertices or lines. This poses a difficulty in development of DNN solvers. We design adaptive DNN techniques for the corner singularity, which incorporates (a) an adaptive loss function that assigns radial weights to local losses, (b) an adaptive activation function that varies the choice of activation functions from neuron to neuron, and (c) an adaptive sampling strategy that selectively focuses on sampling points near the singular locations. The adaptive DNN with these adaptive schemes greatly enhances approximation accuracy and computation efficiency. This is verified by a comprehensive series of numerical experiments of 2D and 3D corner singular problems. The comparison analyzes with the existing DNN solvers are also made to demonstrate effectiveness of the new method. The L2 relative errors of the proposed adaptive DNN are reduced by an order of magnitude with respect to several conventional DNN methods.

Quantifying soil erosion and influential factors in Guwahati's urban watershed using statistical analysis, machine and deep learning

Soil erosion is a complex environmental issue influenced by rapid climate change, resource exploitation, and soil degradation etc. These factors have triggered global acceleration of soil erosion, primarily due to rapid transformation of topographical features and landscape composition. Guwahati, a thriving financial hub in northeast India, witnessing significant landscape change on both the banks of the Brahmaputra river therefore becomes disaster-prone zones. Hence, the objective of the present study is to identify soil erosion factors and assess its impact using statistical, machine learning, and deep learning techniques. It employs Revised Universal Soil Loss Equation (RUSLE) model for soil erosion estimation, furthermore analyzing physical attributes such as morphometrics, topography, drainage networks, and land use fragmentation indicators. Partial Least Squares Regression (PLSR), Random Forest (RF) sensitivity analysis, and Deep Neural Network (DNN) techniques are used in the study. The RUSLE model showed a significant range of soil erosion rates in the study area, spanning from 168.16 to 188.60 tonnes/hectare/year. Particularly, Silsako, Bharalu, North Guwahati, and Foreshore experiences the most severe soil loss. Amongst all influential factors contributing to soil erosion, the most important key parameters are rainfall, drainage density, landscape fragmentation components (such as cohesion index, edge density, and Shannon diversity index), along with stream frequency and basin relief, as indicated by the RF and DNN models. Furthermore, the PLSR analysis assigned linear weights to variables, highlighting the effectiveness of 14 out of 15 independent predictors derived from basin characteristics in accurately estimating soil erosion. This study provides important quantitative insights through rigorous scientific analysis, enabling well-informed decisions in urban watershed management within the Brahmaputra region. Furthermore, it enhances understanding of the area's urgent needs, societal implications, and environmental conditions related to soil erosion.

Comparative Analysis of Decision Tree (DT) & Deep Neural Network (DNN) Techniques to Diagnose the Stator Winding Fault in the Permanent Magnet Synchronous Motor (PMSM)

Every industry is moving towards automation, and Electrical motors like DC and AC play a significant role in the overall automation scheme. Electrical motors, however, are prone to multiple manufacturing defects, and untimed wear and tear. Automated detection of faults, while the motor is in operations, can prevent automation breakdowns, and unforeseen accidents. The majority of faults in electrical motors are because of overheating of windings, bearing misalignment, shaft mismatch, vibrations, noise, etc. A few hours of Electric Motor running can generate a lot of sensor data measuring temperature, torque etc.. Manual analysis of such data can be extremely time consuming. Machine Learning can be used to perform automated analysis of data and real-time fault detection as suggested by many experts. In this research, we have used a few machine learning techniques to train the model on top of sensor data and use the model to diagnose the electric motor faults, occurring because of heat in the stator winding part of the permanent magnet synchronous motor (pmsm). We have specifically compared Decision Tree (DT) and Deep Neural Network (DNN) techniques on real-life datasets, and compared their accuracy. As per the final results, Decision Tree (DT) and Deep Neural Network (DNN) have 93.82% & 97% accuracy respectively.

An optimized efficient combinatorial learning using deep neural network and statistical techniques

Research work is to discover the rapid requirement of Artificial Intelligence and Statistics in medical research. Objective is to design a diagnostic prediction system that can detect and predict diseases at an early stage from clinical data sets. Some of major diseases leading reasons of death globally are heart disease and cancer. There are different kinds of cancer, in this study we focused on breast cancer and heart disease. Prediction of these diseases at a very early stage is curable and preventive diagnosis can control death rate. Designed two Artificial Intelligence systems for prediction of above-mentioned diseases using statistics and Deep neural networks (i) Combinatorial Learning (CLSDnn) and (ii) an optimized efficient Combinatorial Learning (eCLSDnn). To evaluate the performance of the proposed system conducted experiments on three different data sets, in which two data sets are of breast cancer namely, Wisconsin-data set of UCI Machine Learning repository and AI for Social Good: Women Coders’ Bootcamp data set and Cleveland heart disease data set of UCI Machine Learning repository. The proposed architectures of binary classification are validated for 70%–30% data splitting and on K-fold cross validation. Recognition of Malignant cancerous tumors CLSDnn model achieved maximum accuracy of 98.53% for Wisconsin data set, 95.32% for AI for Social Good: Women Coders’ data set and 96.72% for Cleveland data set. Recognition of Malignant cancerous tumors eCLSDnn model achieved 99.36% for Wisconsin data set, 97.12% for AI for Social Good: Women Coders’ data set and 99.56% for the Cleveland heart disease data set.

Real-time MRI lungs images revealing using Hybrid feedforward Deep Neural Network and Convolutional Neural Network

This research focused on Real-time MRI lung images that were revealed using three grade processes by manipulating nanophotonics components, mapping by deep learning, machine learning, and pattern recognition. This research is Solving Magnetic resonance imaging of interstitial lung diseases with Hybrid feedforward Deep Neural Network (ffDNN) and Convolutional Neural Network (CNN) architecture. The feedforward deep neural network (ffDNN) and Convolutional Neural Network (CNN) techniques are used to Solving Magnetic resonance imaging of interstitial lung diseases on the nanophotonics components, deep learning, and machine learning Platform. The Proposed semiconductor monolithic integration approach employed for bio-Magnetic resonance imaging characterization using photonic crystal “Symptomatic Image Revealing” details of the resonant monolithic. The proposed machine-learning-based approach revealed characterizing multi-parameter design space of nanophotonic components using Nano-optic imagers. The Pattern Recognition for MRI was performed for lower dimensionality. Finally, the Hybrid feedforward Deep Neural Network (ffDNN) and Convolutional Neural Network (CNN) architecture for calculating the height and size of scatterers using the inverse design of the meta-optical structure. The temporal resolution assessment of image data pixel size 280x360 hyperspectral imaging temporal resolution is 25, and magnetic resonance imaging temporal resolution is 50. The Image distribution shows that phase shift and transmission are 2.78 degrees and at 95%. The result for the inverse design using CNN returns the efficient inverse design of test data that can be designed according to the required pressure distribution. Wavelength 1000 nanometer to 1600 machine learning method absorbance 40% and ffDNN absorbance 33%.

Smart surveillance using deep learning

<span>Smart surveillance systems play an important role in security today. The goal of security systems is to protect users against fires, car accidents, and other forms of violence. The primary function of these systems is to offer security in residential areas. In today’s culture, protecting our homes is critical. Surveillance, which ranges from private houses to large corporations, is critical in making us feel safe. There are numerous machine learning algorithms for home security systems; however, the deep learning convolutional neural network (CNN) technique outperforms the others. The Keras, Tensorflow, Cv2, Glob, Imutils, and PIL libraries are used to train and assess the detection method. A web application is used to provide a user-friendly environment. The flask web framework is used to construct it. The flash-mail, requests, and telegram application programming interface (API) apps are used in the alerting approach. The surveillance system tracks abnormal activities and uses machine learning to determine if the scenario is normal or not based on the acquired image. After capturing the image, it is compared with the existing dataset, and the model is trained using normal events. When there is an anomalous event, the model produces an output from which the mean distance for each frame is calculated.</span>

Improving Polarimetric Radar-Based Drop Size Distribution Retrieval and Rain Estimation Using a Deep Neural Network

Abstract Raindrop size distributions (DSD) and rain rate have been estimated from polarimetric radar data using different approaches with the accuracy depending on the errors both in the radar measurements and the estimation methods. Herein, a deep neural network (DNN) technique was utilized to improve the estimation of the DSD and rain rate by mitigating these errors. The performance of this approach was evaluated using measurements from a two-dimensional video disdrometer (2DVD) at the Kessler Atmospheric and Ecological Field Station in Oklahoma as ground truth with the results compared against conventional estimation methods for the period 2006–17. Physical parameters (mass-/volume-weighted diameter and liquid water content), rain rate, and polarimetric radar variables (including radar reflectivity and differential reflectivity) were obtained from the DSD data. Three methods—physics-based inversion, empirical formula, and DNN—were applied to two different temporal domains (instantaneous and rain-event average) with three diverse error assumptions (fitting, measurement, and model errors). The DSD retrievals and rain estimates from 18 cases were evaluated by calculating the bias and root-mean-squared error (RMSE). DNN produced the best performance for most cases, with up to a 5% reduction in RMSE when model errors existed. DSD and rain estimated from a nearby polarimetric radar using the empirical and DNN methods were well correlated with the disdrometer observations; the rain-rate estimate bias of the DNN was significantly reduced (3.3% in DNN vs 50.1% in empirical). These results suggest that DNN has advantages over the physics-based and empirical methods in retrieving rain microphysics from radar observations.

Deep neural network techniques for monaural speech enhancement and separation: state of the art analysis

AbstractDeep neural networks (DNN) techniques have become pervasive in domains such as natural language processing and computer vision. They have achieved great success in tasks such as machine translation and image generation. Due to their success, these data driven techniques have been applied in audio domain. More specifically, DNN models have been applied in speech enhancement and separation to perform speech denoising, dereverberation, speaker extraction and speaker separation. In this paper, we review the current DNN techniques being employed to achieve speech enhancement and separation. The review looks at the whole pipeline of speech enhancement and separation techniques from feature extraction, how DNN-based tools models both global and local features of speech, model training (supervised and unsupervised) to how they address label ambiguity problem. The review also covers the use of domain adaptation techniques and pre-trained models to boost speech enhancement process. By this, we hope to provide an all inclusive reference of all the state of art DNN based techniques being applied in the domain of speech separation and enhancement. We further discuss future research directions. This survey can be used by both academic researchers and industry practitioners working in speech separation and enhancement domain.

Transfer Learning with Deep Neural Network toward the Prediction of Wake Flow Characteristics of Containerships

In this study, deep neural network (DNN) and transfer learning (TL) techniques were employed to predict the viscous resistance and wake distribution based on the positions of flow control fins (FCFs) applied to containerships of various sizes. Both methods utilized data collected through computational fluid dynamics (CFD) analysis. The position of the flow control fin (FCF) and hull form information were utilized as input data, and the output data included viscous resistance coefficients and components of propeller axial velocity. The base DNN model was trained and validated using a source dataset from a 1000 TEU containership. The grid search cross-validation technique was employed to optimize the hyperparameters of the base DNN model. Then, transfer learning was applied to predict the viscous resistance and wake distribution for containerships of varying sizes. To enhance the accuracy of feature prediction with a limited amount of data, learning rate optimization was conducted. Transfer learning involves retraining and reconfiguring the base DNN model, and the accuracy was verified based on the fine-tuning method of the learning model. The results of this study can provide hull designers for containerships with performance evaluation information by predicting wake distribution, without relying on CFD analysis.

Sentiment analysis for predicting stress among workers and classification utilizing CNN: Unveiling the mechanism

Stress is one of the main factors that affect the lives of many individuals' lives. Stressful work situations cause significant levels of stress in many employees. Extreme stress can cause significant problems at work and put workers' productivity, health, and safety at risk. Therefore, it is crucial to inform the person about his or her risky lifestyle and forewarn him or her before a serious issue arises. This study offers a novel optimized deep neural network technique, ABC-ABO-CNN, to forecast worker stress. Individuals are significantly impacted by stress, especially in high-stress job contexts. Identifying and measuring employee stress levels is critical to maintaining productivity, health, and safety. Traditional methods like self-assessment questionnaires have limitations, leading to biased results. The suggested approach uses a Convolutional Neural Network (CNN) in conjunction with the African buffalo and Artificial Bee Colony optimization algorithms to overcome these drawbacks. Physiological information such as heart rate, BMI, pressure, glucose level, and skin reaction is used to detect stress. The next step is to categorize stress levels using sentiment analysis. When the suggested mechanism's performance is compared to that of traditional methods, it is shown to have greater accuracy, stress detection rate, and precision values.

Discovery of small molecule modulators targeting β-catenin interactions using molecular docking and molecular dynamics strategy

The rapid activation of signaling pathways is the primary factor responsible for the progression of serious malignancies. Particularly, overexpression of β-catenin is one of the preventative oncogenic biomarkers that is widely present in many tumor types. Therefore, we have designed an integrated pharmacophore strategy to discover the potential inhibitor against β-catenin. Primarily, the ligand-, energy-optimized and the receptor cavity-based pharmacophore hypothesis AAHHR, ADR and ANNNRRR were constructed to screen the active molecules from the database. In order to improve the screening efficacy, the resultant molecules were scrutinized by integrating the results obtained through docking and MM/GBSA calculations. Further, the common hit molecules were subjected to a pharmacokinetic and toxicity study which resulted in two potent β-catenin inhibitors: BAS03014832 and BAS01077671. The existence of thiazole and pyrazole moiety is responsible for the anti-neoplastic activity of these molecules. Finally, a deep neural network technique was used to assess the inhibitory action of these compounds across 66 colorectal cancer cell lines. Overall, the results of our study portray a pathway for the experimental biologist in developing selective β-catenin inhibitors.

Intelligent Fault Diagnosis of Rolling Element Bearings Based on Modified AlexNet.

The reliable and safe operation of industrial systems needs to detect and diagnose bearing faults as early as possible. Intelligent fault diagnostic systems that use deep learning convolutional neural network (CNN) techniques have achieved a great deal of success in recent years. In a traditional CNN, the fully connected layer is located in the final three layers, and such a layer consists of multiple layers that are all connected. However, the fully connected layer of the CNN has the disadvantage of too many training parameters, which makes the model training and testing time longer and incurs overfitting. Additionally, because the working load is constantly changing and noise from the place of operation is unavoidable, the efficiency of intelligent fault diagnosis techniques suffers great reductions. In this research, we propose a novel technique that can effectively solve the problem of traditional CNN and accurately identify the bearing fault. Firstly, the best pre-trained CNN model is identified by considering the classification's success rate for bearing fault diagnosis. Secondly, the selected CNN model is modified to effectively reduce the parameter quantities, overfitting, and calculating time of this model. Finally, the best classifier is identified to make a hybrid model concept to achieve the best performance. It is found that the proposed technique performs well under different load conditions, even in noisy environments, with variable signal-to-noise ratio (SNR) values. Our experimental results confirm that this proposed method is highly reliable and efficient in detecting and classifying bearing faults.

Deep neural network technique for automated detection of ADHD and CD using ECG signal

Background and objectiveAttention Deficit Hyperactivity problem (ADHD) is a common neurodevelopment problem in children and adolescents that can lead to long-term challenges in life outcomes if left untreated. Also, ADHD is frequently associated with Conduct Disorder (CD), and multiple research have found similarities in clinical signs and behavioral symptoms between both diseases, making differentiation between ADHD, ADHD comorbid with CD (ADHD+CD), and CD a subjective diagnosis. Therefore, the goal of this pilot study is to create the first explainable deep learning (DL) model for objective ECG-based ADHD/CD diagnosis as having an objective biomarker may improve diagnostic accuracy. MethodsThe dataset used in this study consist of ECG data collected from 45 ADHD, 62 ADHD+CD, and 16 CD patients at the Child Guidance Clinic in Singapore. The ECG data were segmented into 2 s epochs and directly used to train our 1-dimensional (1D) convolutional neural network (CNN) model. ResultsThe proposed model yielded 96.04% classification accuracy, 96.26% precision, 95.99% sensitivity, and 96.11% F1-score. The Gradient-weighted class activation mapping (Grad-CAM) function was also used to highlight the important ECG characteristics at specific time points that most impact the classification score. ConclusionIn addition to achieving model performance results with our suggested DL method, Grad-CAM's implementation also offers vital temporal data that clinicians and other mental healthcare professionals can use to make wise medical judgments. We hope that by conducting this pilot study, we will be able to encourage larger-scale research with a larger biosignal dataset. Hence allowing biosignal-based computer-aided diagnostic (CAD) tools to be implemented in healthcare and ambulatory settings, as ECG can be easily obtained via wearable devices such as smartwatches.

Deep multi-task learning for nephropathy diagnosis on immunofluorescence images

Background and ObjectiveAs an advanced technique, immunofluorescence (IF) is one of the most widely-used medical image for nephropathy diagnosis, due to its ease of acquisition with low cost. In practice, the clinically collected IF images are commonly corrupted by blurs at different degrees, mainly because of the inaccurate focus at the acquisition stage. Although deep neural network (DNN) methods achieve the great success in nephropathy diagnosis, their performance dramatically drops over the blurred IF images. This significantly limits the potential of leveraging the advanced DNN techniques in real-world nephropathy diagnosis scenarios. MethodsThis paper first establishes two IF databases with synthetic blurs (IFVB) and real-world blurs (Real-IF) for nephropathy diagnosis, respectively, including 1,659 patients and 6,521 IF images with various degrees of blurs. According to the analysis on these two databases, we propose a deep hierarchical multi-task learning based nephropathy diagnosis (DeepMT-ND) method to bridge the gap between the low-level vision and high-level medical tasks. Specifically, DeepMT-ND simultaneously handles the main task of automatic nephropathy diagnosis, as well as the auxiliary tasks of image quality assessment (IQA) and de-blurring. ResultsExtensive experiments show the superiority of our DeepMT-ND in terms of diagnosis accuracy and generalization ability. For instance, our method performs better than nephrologists with at least 15.4% and 6.5% accuracy improvements in IFVB and Real-IF, respectively. Meanwhile, our method also achieves comparable performance in two auxiliary tasks of IQA and de-blurring on blurred IF images. ConclusionsIn this paper, we propose a new DeepMT-ND method for nephropathy diagnosis on blurred IF images. The proposed hierarchical multi-task learning framework provides the new scope to narrow the gap between the low-level vision and high-level medical tasks, and will contribute to nephropathy diagnosis in clinical scenarios. The diagnosis accuracy and generalization ability of DeepMT-ND are experimentally verified to be effective over both synthetic and real-world databases.

Artificial intelligence-based prediction of strengths of slag-ash-based geopolymer concrete using deep neural networks

The construction and building industry, one of the greatest emitters of greenhouse gases, is under tremendous pressure because of the growing concern about global climate change and its detrimental effects on societies. Given the environmental problems connected to cement production, geopolymer concrete has become a viable alternative. In addition, if the concrete strength results failed to meet the specified strength after being cast, modifications are impossible. Thus, it is particularly desirable to predict strength prior to casting concrete. This study presents the first effort in applying deep neural networks (DNN) of AI techniques to predict the mechanical strengths (GGBFS) of geopolymer concrete (GPC) produced from corncob ash and ground granulated blast furnace slag. The mixes were activated with 12–16 M of alkali solutions at ambiently cured conditions for 7–90 days. Following that, back propagation learning algorithms were created for forecasting the concrete strengths based on concrete mix proportions. The mechanical strengths estimated by the DNN were verified by laboratory testing results. Results revealed that GGBFS, mix grade, curing days, and alkali precursor are variables that govern the mechanical strengths of the GGBFS-CCA-GPC. Forecasting the mechanical properties of GPC produced using DNN shows that the relationship between the input and output arguments could be most accurately predicted by a 10–20–20–20-1 network topology, evident by approximately 99% correlation coefficient between the actual and predictive values for compressive and flexural strengths. However, the 10–17–17–17-1 network architecture showed the best DNN for predicting split tensile strength, with a 97% correlation coefficient between the actual and projected values. This study demonstrated that the DNN techniques are efficient in predicting the mechanical strengths of GPC based on the mix proportions. Application of these techniques will greatly advance concrete quality assurance.

Model construction and optimization for raising the concentration of industrial bioethanol production by using a data-driven ANN model

This work aims to model and predict the bioethanol produced from a conventional fermentation process. An industrial dataset was obtained from a sugar mill in India. These datasets comprised a total of 1300 experimental values acquired from a total 100 days of production of the sugar mill in the year 2021. Using this data, a framework based on deep learning Artificial Neural Network (ANN) technique model was developed and validated with the test data. Specifically, the normalized dataset was passed through an ANN model consisting of one input layer, two hidden layers and one output layer, and the percent model error calculated by using average absolute deviation metric and was found to be 4.48 and 1.99% for training and testing data, respectively. The optimization of the process variables was performed for the first time using a data synthesis technique in which the normalized dataset was first synthesized and then passed through an ANN model to get an optimized input variable set for an increase in bioethanol concentration (BEC) in the final product by 1°GL. The operating parameters which significantly influenced the BEC are concentration of cell in pre-fermentation (PFM), water fermentation pH (WFMpH), and water fermentation hardness (WFH). An increase of 8.45% in BEC was obtained for Sugar mill, India.

A study on nuclear binding energy and beta decay energy using deep neural networks

The Artificial Neural Network (ANN) is one of the innovative methods for predicting the structural and dynamic properties of atomic nuclei. In this work, we employed machine learning based on the Deep Neural Network (DNN) technique to find the ground state binding energy and beta decay energy of various nuclei. 3560 experimental nuclear data sets have been used for training and validating the DNN model. Neutron number (N), proton number (Z), the pairing term (δ), a magic number greater than or equal to N and Z, asymmetry factor (a) and promiscuity factor (P) are used as input parameters. The Rectified Linear Unit (ReLU) function and Mean Absolute Error (MAE) are used as activation and Loss functions for the training process. We perform the predictions of ground state binding energy per nucleon and beta decay energies in Zinc and Promethium isotopes using the DNN model, which is in good agreement with experimental results.

Real-Time Tracking of Laryngeal Motion via the Surface Depth-Sensing Technique for Radiotherapy in Laryngeal Cancer Patients.

Radiotherapy (RT) is an important modality for laryngeal cancer treatment to preserve laryngeal function. During beam delivery, laryngeal motion remains uncontrollable and may compromise tumor-targeting efficacy. We aimed to examine real-time laryngeal motion by developing a surface depth-sensing technique with preliminary testing during RT-based treatment of patients with laryngeal cancer. A surface depth-sensing (SDS) camera was set up and integrated into RT simulation procedures. By recording the natural swallowing of patients, SDS calculation was performed using the Pose Estimation Model and deep neural network technique. Seven male patients with laryngeal cancer were enrolled in this prospective study. The calculated motion distances of the laryngeal prominence (mean ± standard deviation) were 1.6 ± 0.8 mm, 21.4 ± 5.1 mm, 6.4 ± 3.3 mm, and 22.7 ± 4.9 mm in the left-right, cranio-caudal, and anterior-posterior directions and for the spatial displacement, respectively. The calculated differences in the 3D margins for generating the planning tumor volume by senior physicians with and without SDS data were -0.7 ± 1.0 mm (-18%), 11.3 ± 6.8 mm (235%), and 1.8 ± 2.6 mm (45%) in the left-right, cranio-caudal, and anterior-posterior directions, respectively. The SDS technique developed for detecting laryngeal motion during swallowing may be a practical guide for individualized RT design in the treatment of laryngeal cancer.

DLIRIR : Deep learning based improved Reverse Image Retrieval

Deep feature search has transpired as a sub-problem of large-scale Reverse Image Retrieval (RIR) algorithms in computer vision. In light of recent progress, transforming deep features into binary codes and employing various search strategies to delineate similar images has gained popularity. However, these binary codes cannot accurately define the similarity between two images, which is essential for image retrieval. Moreover, when retrieving k closest images, the existing RIR approaches show a drop in precision value as k increases. In this paper, we put forward a method, Deep Learning based Improved Reverse Image Retrieval (DLIRIR), which solves the RIR problem by utilizing a deep neural network, clustering methodology, and dimensionality reduction technique. Motivated by the need to prevent information loss in converting deep features to binary codes and reduce search time, we cluster image features and select some features from each cluster called Representative Points (RPs). These RPs are then compared to features of an image (query) to retrieve images similar to it from the database. The Euclidean distances of the query feature and the RPs indicate similarity between the query and the cluster of images represented by that RP. We reduce the number of comparisons by selecting as few RPs as possible. We evaluate the performance of DLIRIR by comparing it with other state-of-the-art algorithms in the RIR domain. Our method achieves a precision greater than 99% in benchmark datasets Caltech 256, Corel 10k, and Adaptiope, while a precision between 80%–85% on large size datasets Cifar 100, Tiny ImageNet, and VGGFace2.

Quantifying the Potential Contribution of Submerged Aquatic Vegetation to Coastal Carbon Capture in a Delta System from Field and Landsat 8/9-Operational Land Imager (OLI) Data with Deep Convolutional Neural Network

Submerged aquatic vegetation (SAV) are highly efficient at carbon sequestration and, despite their relatively small distribution globally, are recognized as a potentially valuable component of climate change mitigation. However, SAV mapping in tidal marshes presents a challenge due to optically complex constituents in the water. The emergence and advancement of deep learning-based techniques in the field of habitat mapping with remote sensing imagery provides an opportunity to address this challenge. In this study, an analytical framework was developed to quantify the carbon sequestration of SAV habitats in the Atchafalaya River Delta Estuary from field and remote sensing observations using deep convolutional neural network (DCNN) techniques. A U-Net-based model, Wetland-SAV Network, was trained to identify the SAV percent cover (high, medium, and low) as well as other estuarine habitat types from Landsat 8/9-OLI data. The areal extent of SAV was up to 8% of the total area (47,000 ha). The habitat areas and habitat-specific carbon fluxes were then used to quantify the net greenhouse gas (GHG) flux of the study area for with/without SAV scenarios in a carbon balance model. The total net GHG flux was in the range of −0.13 ± 0.06 to −0.86 ± 0.37 × 105 tonne CO2e y−1 and increased up to 40% (−0.23 ± 0.10 to −0.90 ± 0.39 × 105 tonne CO2e y−1) when SAV was accounted for within the calculation. At the hectare scale, the inclusion of SAV resulted in an increase of ~60% for the net GHG sink in shallow areas adjacent to the emergent marsh where SAV was abundant. This is the first attempt at remotely mapping SAV in coastal Louisiana as well as a first quantification of net GHG flux at the scale of hectares to thousands of hectares, accounting for SAV within these sub-tropical coastal delta marshes. Remote sensing and deep learning models have high potential for mapping and monitoring SAV in turbid sub-tropical coastal deltas as a component of the increasing accuracy of net GHG flux estimates at small (hectare) and large (coastal basin) scales.