Browse Items (14421 total)
Sort by:
-
Deep fake detection using cascaded deep sparse auto-encoder for effective feature selection
In the recent research era, artificial intelligence techniques have been used for computer vision, big data analysis, and detection systems. The development of these advanced technologies has also increased security and privacy issues. One kind of this issue is Deepfakes which is the combined word of deep learning and fake. DeepFake refers to the formation of a fake image or video using artificial intelligence approaches which are created for political abuse, fake data transfer, and pornography. This paper has developed a Deepfake detection method by examining the computer vision features of the digital content. The computer vision features based on the frame change are extracted using a proposed deep learning model called the Cascaded Deep Sparse Auto Encoder (CDSAE) trained by temporal CNN. The detection process is performed using a Deep Neural Network (DNN) to classify the deep fake image/video from the real image/video. The proposed model is implemented using Face2Face, FaceSwap, and DFDC datasets which have secured an improved detection rate when compared to the traditional deep fake detection approaches. 2022. Balasubramanian et al. -
Deep Insights into 3D Face Reconstruction from Blurred 2D Inputs: A Comprehensive Framework
This framework outlines a multi-stage methodology for 3D face reconstruction driven by advancements in deep learning. The process involves image preprocessing with deblurring techniques and subsequent feature extraction using CNNs alongside traditional methods. Deep learning adapts to diverse image challenges, ensuring accuracy in 3D reconstructions. In medical imaging, the proficiency of 3D CNNs and GANs shines in extracting structures from MRI and CT scans. Post-processing steps encompass mesh smoothing and texture mapping for enhanced visual quality. Evaluation metrics (MAE, RMSE, IoU) guarantee the precision of depth estimations. Applications of deep learning span across CNNs, 3DMM, GANs, and networks for landmark detection and dense correspondence. Challenges include optimizing eye reconstruction, expanding applications, and addressing concerns related to data quality, privacy, and hardware requirements. 2024 IEEE. -
Deep Learning Advancements in E-commerce Supply Chain Management in Forecasting and Optimization Strategies
In this study, the influence of deep learning technologies on the optimization of supply chain management in the context of the e-commerce industry is examined. Using a dataset of historical data of sales, inventories, market fluctuations, and customer and supplier details, I investigate the efficiency of different deep learning models to predict demand and facilitate the optimal balance of inventories. Long Short-Term Memory (LSTM), Recurrent Neural Networks (RNN), Convolutional Neural Networks (CNN), and a model proposed by the authors are defined and applied, considering their accuracy, precision, recall, and F-1 score. The results show that the proposed model outperforms traditional products, achieving 97.5% of accuracy. In the context of the comparative analysis, the specific features of CNN, LSTM, and RNN are revealed, helping to understand the benefits and drawbacks of each recommendation. As a result, the proposed model proves that deep learning technologies have the power to change the approach to predictive analytics and supply chain management, allowing practitioners to focus on strengths and overcome the weaknesses of their structures. The impact of data preprocessing and hyperparameters is also considered along with the necessity to choose the most appropriate model evaluation technique. In the future, it is possible to implement other complex deep learning models, integrate additional data, and address the problem of data scaling and heterogeneity. In the era of modern technologies, e-commerce organizations should take these findings into consideration to discover the potential of deep learning, improve supply chain performance, reduce costs, and attract clients. This research contributes to the topic of using deep learning technologies in supply chain management, promoting innovation, and changes that may affect the industry drastically. 2024 IEEE. -
Deep Learning Algorithms Comparison forMultiple Biological Sequences Alignment
In this paper, deep learning algorithms are compared for aligning multiple biological molecular sequences such as DNA, RNA, and protein. Efficient algorithms are necessary for sequence alignment to identify significant insights, but there is a trade-off between time and accuracy. This study compares deep learning algorithms for multiple sequence alignment with better accuracy, using a new similarity measure to choose the best resemblance sequences in a set. Using a benchmark dataset, the algorithms compared include CNN, VAE, MLPNN, DBNs, Deep Boltzmann Machine, and GAN. 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. -
Deep learning algorithms for intrusion detection systems in internet of things using CIC-IDS 2017 dataset
Due to technological advancements in recent years, the availability and usage of smart electronic gadgets have drastically increased. Adoption of these smart devices for a variety of applications in our day-to-day life has become a new normal. As these devices collect and store data, which is of prime importance, securing is a mandatory requirement by being vigilant against intruders. Many traditional techniques are prevailing for the same, but they may not be a good solution for the devices with resource constraints. The impact of artificial intelligence is not negligible in this concern. This study is an attempt to understand and analyze the performance of deep learning algorithms in intrusion detection. A comparative analysis of the performance of deep neural network, convolutional neural network, and long short-term memory using the CIC-IDS 2017 dataset. 2023 Institute of Advanced Engineering and Science. All rights reserved. -
Deep Learning Analysis of Satellite Images for UN SDG Monitoring in Mauritius' Black River District
This paper proposes an integrated strategy to analyse the progress of selected United Nations Sustainable Development Goals (SDGs 1, 2, and 13) using Earth Observation (EO) data and deep learning (DL) based classification. The research focuses on Mauritius's Black River district, which is facing growing urbanization, agricultural land demand, forest conservation needs, and land degradation. These challenges are closely related to the reduction of poverty (SDG 1) through settlement monitoring, food security (SDG 2) through green farmland analysis, and climate action (SDG 13) through forest cover and bare land tracking. High-resolution satellite images from the Satellogic constellation were pre-processed, classified, and mapped to the SDGs' key land cover categories. A convolutional neural network (CNN) model was trained to distinguish city structures, agriculture farmland, forest land, and barren land, with up to 99% overall precision. DL-based image analysis has the ability to monitor the UN SDGs in specific regions and provide actionable information for the sustainable development plans of small island governments. 2025 IEEE. -
Deep Learning Approaches for Detection and Classification of Microplastics in Water for Clean Water Management
Microplastic pollution is a growing environmental concern, threatening aquatic ecosystems and human health. This study presents a dual deep learning approach for microplastic detection and classification using two datasets. For water microplastics, YOLOv8 and YOLOv11 were employed for object detection. InceptionV3, VGG19, ResNet50, ResNet152, DenseNet121, EfficientNetB0, and a custom CNN were applied for classification, classifying three distinct microplastic types in non-aquatic environments. Experimental findings display high accuracy, and indicate the potential of AI-enabled solutions for environmental monitoring. This research contributes to SDG 6 Clean Water and Sanitation, promoting sustainable management of water. 2025 IEEE. -
Deep Learning Approaches for Environmental Monitoring in Smart Cities
It introduces a novel integrated environmental monitoring system capable of doing on-the-go measurements. In metropolitan settings, air pollution is one of the most serious environmental threats to human health. The widespread use of automobiles, emissions from manufacturing processes, and the use of fossil fuels for propulsion and power generation have all contributed to this issue. Air quality predictions in smart cities may now be made using deep learning methods, thanks to the widespread adoption of these tools and their continued rapid growth. Particulate Matter (PM) with a width of less than 2.5 m (PM2.5) is one of the most perilous kinds of air pollution. To anticipate the hourly gauge of PM2.5 focus in Delhi, India, we utilized verifiable information of poisons, meteorological information, and PM2.5 fixation in the adjoining stations to make a spatial-worldly element for our CNN-LSTM-based deep learning arrangement. According to our experiments, our 'hybrid CNN-LSTM multivariate' method outperforms all of the above conventional models and allows for more precise predictions. 2024 IEEE. -
Deep learning approaches to understanding psychological impacts on vulnerable populations
This chapter investigates the psychological effects on vulnerable groups, with a particular emphasis on the relationship between deep learning techniques and the impact of climate. Vulnerable groups confront particular problems, which might lead to negative psychological results. Investigating this complexity is critical to designing effective intervention techniques. Using sophisticated deep learning techniques, this study seeks to find subtle patterns and correlations in a variety of datasets, including psychological markers, socioeconomic characteristics, and climatic variables. The work employs a comprehensive technique that includes deep learning models, feature extraction, and interpretability analysis to untangle complicated relationships. Preliminary findings imply that deep learning approaches might uncover previously unknown links between climate change and psychological effects on vulnerable groups. This insight adds to a more comprehensive understanding of the difficulties. This understanding contributes to a more holistic grasp of the challenges faced by these groups. By including climate-related factors into the deep learning framework, this study hopes to close the gap between environmental impacts and psychological 2024, IGI Global. All rights reserved. -
Deep learning architectures for multimodal fusion
The advancement in technology during the recent years has provided deep learning technology as an emerging and powerful paradigm which can be used for processing and understanding complex data across various domains. Multimodal fusion is integrating the information that is collected from various sources or modalities, which requires a comprehensive understanding of data like autonomous driving, medical diagnosis, etc. In this chapter, we will explore the various advanced deep learning architectures that have been specially designed based on the multimodal fusion. The various challenges that are being faced in multimodal data, which include heterogeneity, noise, reliability of data, etc. Various deep learning architectures that are built to address the various challenges, like convolutional neural networks, recurrent neural networks, are reviewed, and the suitability of the fusion strategies is highlighted. The various techniques that are used for combining the information from disparate modalities, like early fusion, late fusion, and hybrid approaches, are also discussed with their pros and cons. Various real-time applications in the field of healthcare, multimedia, robotics, etc., are demonstrated based on the impact of the architectures. Finally, the potential of deep learning architecture based on the revolutionary multimodal fusion will be discussed. 2026 Elsevier Inc. All rights reserved. -
Deep Learning Based Age Estimation Model
To improve accuracy and resilience in demographic categorization, this research presents a novel use of Convolutional Neural Networks (CNNs) for age prediction. Deep learning is utilized to achieve this goal. Precise estimation of age has become essential in a variety of areas, including human-computer interaction, marketing, and healthcare. The ability of CNNs to handle the intricacies of facial features for accurate demographic forecasts is examined in this study. The research covers every step of the age prediction process, including dataset collection, prepossessing, model architecture, and assessment measures. The CNN is trained to automatically extract hierarchical characteristics from facial photos, which enables the model to recognize complex patterns related to age. The architecture's flexibility to different lighting conditions, facial expressions, and postures. In this research, we deal with deep learning-based perceived age estimation in still-face pictures. Our Convolution Neural Network models (CNNs) have been trained prior on Image Net for picture classification, as they use the VGG architecture. In addition, we analyze the effects of tailoring over Web photos having known age, considering a lack of apparent age-annotated annotated images. In addition, this work adds to the increasing library of studies on the use of deep learning methods for demographic data evaluation by showing the effectiveness of CNNs to predict age. The results show how, in practical situations, CNNs could significantly enhance the precision and dependability of age prediction systems. 2024 IEEE. -
Deep learning based classification of microplastic in edible food using optical microscopy images
Microplastics (MPs), a prevalent pollution in food, water, and ecosystems around the world, have become a serious environmental and health concern. The traditional detection and classification techniques are labor-intensive by nature and do not support extensive, large-scale monitoring. The main emphasis of this study is to generate a novel image dataset via a simple extraction method that will be useful for classification applications in high-consumption edible food by integrating with the deep-learning model. This study compares the efficacy of several Deep learning (DL) architectures, including MobileNetV2, ResNet101V2, ResNet50V2, InceptionV3, EfficientNetB0, and a baseline Convolutional Neural Network (CNN) in classification into three groups: threads, beads, and fragments. The best performance was recorded by MobileNetV2, ResNet101V2, and ResNet50 V2, all with 98 percent test accuracy and weighted F1-scores of 0.986 and 0.983, respectively, which is a strong and consistent MPs classification. The outcome indicates that the DL models, especially ResNet101V2 and MobileNetV2, outperform the baseline CNN in terms of classification accuracy (98%). The present study provides strong, scalable opportunities for Artificial Intelligence (AI) based solutions for the assessment and reduction of MPs contamination globally in edible food. The Author(s) 2026. -
Deep Learning Based Face Recognized Attendance Management System using Convolutional Neural Network
In today's digital age, manual attendance tracking is plagued by inefficiency and the potential for inaccuracies, often leading to proxy attendance. The main aim of this research work is to manage and monitor the student's attendance by using face recognition technology. This proposed model is mainly categorized four major modules. First module is database creation. Second module is face detection. Then third module is face recognition and final module is automatic attendance updating process. Student images are compiled to create a comprehensive database, ensuring inclusivity across the class roster. The system utilizes the face recognition library, which relies on deep learning based algorithms for face detection and recognition during testing. This face recognition part Convolutional Neural Network algorithm is used. The system matches detected faces with the known database and marks attendance, ensuring a streamlined and accurate attendance tracking process. This innovative approach has the potential to revolutionize attendance management in educational settings, offering a contactless and efficient solution while mitigating proxy attendance concerns. The proposed model is to compare the accuracy level of face recognition. 2023 IEEE. -
Deep learning based federated learning scheme for decentralized blockchain
Blockchain has the characteristics of immutability and decentralization, and its combination with federated learning has become a hot topic in the field of artificial intelligence. At present, decentralized, federated learning has the problem of performance degradation caused by non-independent and identical training data distribution. To solve this problem, a calculation method for model similarity is proposed, and then a decentralized, federated learning strategy based on the similarity of the model is designed and tested using five federated learning tasks: CNN model training fashion-mnist dataset, alexnet model training cifar10 dataset, TextRnn model training thusnews dataset, Resnet18 model training SVHN dataset and LSTM model training sentiment140 dataset. The experimental results show that the designed strategy performs decentralized, federated learning under the nonindependent and identically distributed data of five tasks, and the accuracy rates are increased by 2.51, 5.16, 17.58, 2.46 and 5.23 percentage points, respectively. 2024 selection and editorial matter, Arvind Dagur, Karan Singh, Pawan Singh Mehra & Dhirendra Kumar Shukla; individual chapters, the contributors. -
Deep learning based federated learning scheme for decentralized blockchain
Blockchain has the characteristics of immutability and decentralization, and its combination with federated learning has become a hot topic in the field of artificial intelligence. At present, decentralized, federated learning has the problem of performance degradation caused by non-independent and identical training data distribution. To solve this problem, a calculation method for model similarity is proposed, and then a decentralized, federated learning strategy based on the similarity of the model is designed and tested using five federated learning tasks: CNN model training fashion-mnist dataset, alexnet model training cifar10 dataset, TextRnn model training thusnews dataset, Resnet18 model training SVHN dataset and LSTM model training sentiment140 dataset. The experimental results show that the designed strategy performs decentralized, federated learning under the nonindependent and identically distributed data of five tasks, and the accuracy rates are increased by 2.51, 5.16, 17.58, 2.46 and 5.23 percentage points, respectively. 2024 The Author(s). -
Deep learning based model for computing percentage of fake in user reviews using topic modelling techniques
Sentiment analysis plays a vital role in real time environment for knowing the history of a product or any other specific entity. Due to large number of users in the www, chances are there that many fake users may upload the fake reviews to damage the business for the sake of money. Identifying the fake reviews or percentage of fake content in the review is yet a challenging task. In this paper, an attempt has been made to find the percentage of fake in the review data. Two methodologies are combined to address this issue. Concept of spelling checking, topic modelling and deep learning for context extraction is extensively used to build the effective model. Proposed technique is exhaustively checked for efficiency with many trails of experiments. Also, the training and testing samples were shuffled for experimentation. The results of the models show its goodness. The details of the results can be found at experiments section. 2024 The Author(s) -
Deep learning based modeling of groundwater storage change
The understanding of water resource changes and a proper projection of their future availability are necessary elements of sustainable water planning. Monitoring GWS change and future water resource availability are crucial, especially under changing climatic conditions. Traditional methods for in situ groundwater well measurement are a significant challenge due to data unavailability. The present investigation utilized the Long Short Term Memory (LSTM) networks to monitor and forecast Terrestrial Water Storage Change (TWSC) and Ground Water Storage Change (GWSC) based on Gravity Recovery and Climate Experiment (GRACE) datasets from 20032025 for five basins of Saudi Arabia. An attempt has been made to assess the effects of rainfall, water used, and net budget modeling of groundwater. Analysis of GRACE-derived TWSC and GWSC estimates indicates that all five basins show depletion of water from 20032020 with a rate ranging from -5.88 1.2 mm/year to -14.12 1.2 mm/year and -3.5 1.5 to -10.7 1.5, respectively. Forecasting based on the developed LSTM model indicates that the investigated basins are likely to experience serious water depletion at rates ranging from -7.78 1.2 to -15.6 1.2 for TWSC and -4.97 1.5 to -12.21 1.5 for GWSC from 20202025. An interesting observation was a minor increase in rainfall during the study period for three basins. 2022 Tech Science Press. All rights reserved. -
Deep Learning Based Multi Constraint Hybrid Optimization Algorithm for Transshipment-Based Inventory Routing with Dynamic Demands
The Inventory-Routing Problem (IRP) is considered a major issue in supply chain management as it comprises two areas: vehicle routing and inventory control. The existing techniqueswere unable to incorporate location details for enhancing the decision-making and it failed to consider the uncertainty of the demands. Hence to solve this issue, a Snake Artificial Ecosystem Optimization (SAEO) algorithm is proposed in this paper. The SAEO algorithm is developed to address the transshipment IRP with dynamic demands by combining the AEO model and SO to enhance the optimizer's performance. Further, a penalty strategy is proposed, where Deep Quantum Neural Network (DQNN) is employed for calculating the penalty for verifying the feasibility of the solution generated in case of violations in model constraints. In addition, the efficiency of the proposed SAEO-DQNN technique is examined by considering metrics, like transportation cost, transshipment cost, and total cost, and it achieved improved values of 0.391, 0.518, and 1.012 when compared to existing techniques such as Genetic Algorithm with Deep Reinforcement Learning (GA + Deep RL) and Kernel Search Multi-vehicle IRP (KSMIRP). The Author(s) 2024. -
Deep Learning Based Multi Constraint Hybrid Optimization Algorithm for Transshipment-Based Inventory Routing with Dynamic Demands
The Inventory-Routing Problem (IRP) is considered a major issue in supply chain management as it comprises two areas: vehicle routing and inventory control. The existing techniqueswere unable to incorporate location details for enhancing the decision-making and it failed to consider the uncertainty of the demands. Hence to solve this issue, a Snake Artificial Ecosystem Optimization (SAEO) algorithm is proposed in this paper. The SAEO algorithm is developed to address the transshipment IRP with dynamic demands by combining the AEO model and SO to enhance the optimizer's performance. Further, a penalty strategy is proposed, where Deep Quantum Neural Network (DQNN) is employed for calculating the penalty for verifying the feasibility of the solution generated in case of violations in model constraints. In addition, the efficiency of the proposed SAEO-DQNN technique is examined by considering metrics, like transportation cost, transshipment cost, and total cost, and it achieved improved values of 0.391, 0.518, and 1.012 when compared to existing techniques such as Genetic Algorithm with Deep Reinforcement Learning (GA + Deep RL) and Kernel Search Multi-vehicle IRP (KSMIRP). The Author(s) 2024. -
Deep Learning Based Performance Prediction of Sustainable Microwave Absorbers
This paper proposed a convolutional neural network (CNN) based deep learning (DL) approach to predict performance of sustainable microwave absorbers. This study explores the transformative potential of DL in predicting and optimizing microwave absorber performance, offering a datadriven alternative to traditional approaches. The absorber is a composite of tea and carbon powder considered as waste mixed in various composition percentages. The measured S21 data is used for training the proposed DL model. The prediction of absorber's S21 performance shows an accuracy of above 98 %. 2025 IEEE.