Farmers are often concerned about plant disease since it can greatly affect crop productivity and quality. Expert manual inspection is required in traditional techniques of identifying plant diseases, which can be time- and money-consuming. Deep learning algorithms have made automated plant disease detection systems more practical. Convolutional neural networks (CNNs) are used in our proposed deep learning- based technique for the diagnosis of plant diseases. The suggested system uses plant photos as input to determine the presence (...) of illnesses in the plants and then provides treatment recommendations. The CNN can learn the visual patterns connected to various diseases because it is trained using a sizable dataset of photos of healthy and diseased plants. This system also includes IOT so that after the result is acquired, it will communicate via IOT. Several publicly accessible datasets of plant photos were used in our studies to gauge the performance of the suggested approach. The experimental results show that the suggested approach detects several plant diseases, such as tomato bacterial spot, potato early blight, and apple black rot, with high accuracy. Overall, our approach shows the promise of deep learning methods for automatically detecting plant diseases, which can assist farmers in identifying and treating plant diseases more quickly and successfully. (shrink)

Abstract: People all over the world have been drinking tea for thousands of centuries, and for good reason. Many types of teas can help you stay healthy by boosting your immune system, reducing inflammation, and even preventing cancer and heart disease. There is sufficient material to show that regularly consuming tea can improve your health over the long term. A deep learning model that categorizes tea disorders has been completed. When focusing on the tea, we must also focus (...) on and take into account the diseases that harm it. The aim of this paper is to diagnose tea diseases based on a dataset containing 8000 picture and 8 classes .The model gave the final accuracy test with an F1-score of 99.41% in an estimated time 3 second, as with these results it proved its effectiveness and good ability in the classification test. (shrink)

The early detection of plant diseases is essential to preventing crop losses in terms of quantity and productivity. Farmers typically identify illnesses using their prior knowledge or by spending a great deal of time, effort, and experience doing so. Exceedingly challenging to manually monitor plant diseases. In the development of automatic pathogens diagnosis machines, paddy disease detection is crucial . To identify previously known bacterial leaf blight, brown spot, leaf blast, leaf smut, and other narrow illnesses in (...) prior knowledge, we presented a VGG 16 model in this work. Images of various categories diseases are stored in a special data repository, with 80 percent of the data used for training and 20 percent for sample testing. ResNet, Inception v2 and Mobile Net were all used in the model-building process to add benefits. The results of the interpretation show that there is high level picture classification accuracy and low mistake rate. As a result, our model excels at identifying paddy diseases and may be used in practical applications in daily life. (shrink)

By Agriculture plays a major role in developing countries like India, however the food security still remains a vital issue. Most of the crops get wasted due to lack of storage facility, transportation, and plant diseases. More than 15% of the crops get wasted in India due to diseases and hence it has become one of the major concern to be resolved. There is a need of automatic system that can identify these diseases and help farmers to take appropriate steps (...) to get rid of crop loss. Farmers have followed the conventional method of identifying the plant disease with their naked eyes, and it not possible for all the farmers to identify these diseases the same way. With the advance in Artificial Intelligence, there is a need to incorporate the facilities of the computer vision in the field of agriculture. Deep Learning rich libraries and user as well as developer friendly environment to work with, all these qualities make Deep Learning as the favorable method to get started with this problem. In this paper we have used Deep Learning because of the advantages it offers to work with images especially in image classification to get improvised results. The methodology includes taking leaves of infected crops and label them as per the disease pattern. The images of infected leaves are processed pixel based operations are applied to improve the information from the image. As a next step feature extraction is done followed by image segmentation and at the last classification of crop diseases based on the patterns extracted from the diseased leaves. The CNN (Convolutional Neural Network) is used for the classification of diseases, for the demonstration purpose the public dataset is used consisting of around 87 K images (RGB type images) including healthy as well as diseased leaves. (shrink)

The Internet of Things (IoT) connects schemes, programs, data management, and operations, and as they continuously assist in the corporation, they may be a fresh entryway for cyber-attacks. Presently, illegal downloading and virus attacks pose significant threats to IoT security. These risks may acquire confidential material, causing reputational and financial harm. In this paper hybrid optimization mechanism and deep learning,a frame is used to detect the attack prevention in IoT. To develop a cybersecurity warning system in a huge (...) data set, the cybersecurity warning systems index system is first constructed, then the index factors are picked and measured, and finally, the situation evaluation is done.Numerous bio-inspired techniques were used to enhance the productivity of an IDS by lowering the data dimensionality and deleting unnecessary and noisy input. The Grey Wolf Optimization algorithm (GWO) is a developed bio-inspired algorithm that improves the efficacy of the IDS in detecting both regular and abnormal congestion in the network. The smart initialization step integrates the different pre-processing strategies to make sure that informative features are incorporated in the early development stages, has been improved. Researchers pick multi-source material in a big data environment for the identification and verification of index components and present a parallel reduction approach based on the classification significance matrix to decrease data underlying data characteristics. For the simulation of this situation, grey wolf optimization and whale optimization were combined to detect the attack prevention and the deep learning approach was presented. Utilizing system software plagiarism, the TensorFlow deep neural network is intended to classify stolen software. To reduce the noise from the signal and to zoom the significance of each word in the perspective of open-source plagiarism, the tokenization and weighting feature approaches are utilized. Malware specimens have been collected from the Mailing database for testing purposes. The experimental findings show that the suggested technique for measuring cyber security hazards in IoT has superior classification results to existing methods. Hence to detect the attack prevention in IoT process Whale with Grey wolf optimization (WGWO) and deep convolution network is used. (shrink)

Abstract: Context: grapevine leaves are an important agricultural product that is used in many Middle Eastern dishes. The species from which the grapevine leaf originates can differ in terms of both taste and price. Method: In this study, we build a deep learning model to tackle the problem of grape leaf classification. 500 images were used (100 for each species) that were then increased to 10,000 using data augmentation methods. Convolutional Neural Network (CNN) algorithms were applied (...) to build this model specifically using the pre-trained model on top of the VGG16 architecture. Then, dense layers were added to classify the output of the Convolutional layers and classify outputs to the five classes (species) the leaf belonged to. Results: It was found that feature extraction without fine-tuning the convolutional layers yielded poor results, about 86% accuracy, while training the whole network along with some data preprocessing gave the best results, about 99.45% accuracy on the testing dataset. Conclusions: The proposed CNN model is an effective one for the problem of classification of grape leaf species. (shrink)

Abstract: In this study, we explore the challenge of identifying and preventing diseases in apple trees, which is a popular activity but can be difficult due to the susceptibility of these trees to various diseases. To address this challenge, we propose the use of Convolutional Neural Networks, which have proven effective in automatically detecting plant diseases. To validate our approach, we use images of apple leaves, including Apple Rot Leaves, Leaf Blotch, Healthy Leaves, and Scab Leaves collected from Kaggle (...) which is part from the Plant Village dataset. We generate a comprehensive training dataset using techniques such as image filtering, compression, and generation. Our model achieves impressive accuracy scores for all classes, with an overall accuracy of 99.93% on a dataset of 10,000 labeled images . (shrink)

With the exponential growth of the Internet of Things (IoT), ensuring the security of IoT devices and networks has become a significant challenge. Anomaly detection techniques play a pivotal role in identifying unusual behaviors that may indicate cyber threats. Traditional anomaly detection systems often struggle with scalability, privacy concerns, and the need for continuous model improvement. In this paper, we propose a collaborative anomaly detection framework for IoT systems based on Federated Deep Learning (FDL). This framework allows IoT (...) devices to collaboratively train deep learning models for anomaly detection without sharing raw data, thereby addressing privacy issues. The federated approach also improves scalability by reducing the communication overhead typically associated with centralized learning systems. Experimental results demonstrate the effectiveness of the proposed framework in detecting anomalies while preserving privacy and ensuring scalability across diverse IoT devices. (shrink)

Crop diseases pose several challenges in the agricultural industry. Plant diseases can have a devastating impact on both yield and quality loss. This project presents a deep-learning based sugarcane disease classification and alert system to facilitate machine detection of disease, as well as actions to take in response to diagnosis. A dataset of images of sugarcane leaves, was modified through advanced pre-processing techniques such as cropping, rotating, image enhancement, detections edges, and adjusting for wavy images. The techniques used (...) for data augmentation helped address class imbalance as well as overfitting. The classification model implemented for this project is the EfficientNet architectures to produce better feature extraction and predictions. The classification model was trained on the pre-processed and augmented dataset and reached 97 percent accuracy on the test data after 150 epochs. The explainability of the classifying model was provided through LIME (Local Interpretable Model-Agnostic Explanations), which allowed users to see the explanation for classification for an individual image. The smart classifying system was developed as a web application (UI) and allows users to input their image to be diagnosed. Once diagnosed, the model classifies the disease and detail actions, including: recommended treatments and precautionary actions. This system shows the power of artificial intelligence to enhance management of diseases in agriculture, while providing a scalable user-friendly environment to help faciliate on farm decision-making regarding crop health and yield. (shrink)

Crop diseases pose several challenges in the agricultural industry. Plant diseases can have a devastating impact on both yield and quality loss. This project presents a deep-learning based sugarcane disease classification and alert system to facilitate machine detection of disease, as well as actions to take in response to diagnosis. A dataset of images of sugarcane leaves, was modified through advanced pre-processing techniques such as cropping, rotating, image enhancement, detections edges, and adjusting for wavy images. The techniques used (...) for data augmentation helped address class imbalance as well as overfitting. The classification model implemented for this project is the EfficientNet architectures to produce better feature extraction and predictions. The classification model was trained on the pre-processed and augmented dataset and reached 97 percent accuracy on the test data after 150 epochs. The explainability of the classifying model was provided through LIME (Local Interpretable Model-Agnostic Explanations), which allowed users to see the explanation for classification for an individual image. The smart classifying system was developed as a web application (UI) and allows users to input their image to be diagnosed. Once diagnosed, the model classifies the disease and detail actions, including: recommended treatments and precautionary actions. This system shows the power of artificial intelligence to enhance management of diseases in agriculture, while providing a scalable user-friendly environment to help faciliate on farm decision-making regarding crop health and yield. (shrink)

As the Internet of Things (IoT) continues to grow, the security challenges in these networks increase in complexity and scale. Traditional intrusion detection methods often struggle to handle the dynamic and diverse nature of IoT environments. Federated Learning (FL) has emerged as a promising solution to enhance privacy and scalability in IoT security, allowing multiple devices to collaboratively learn from data without sharing it. However, due to the heterogeneity of IoT devices and the limited data available on individual devices, (...) Federated Transfer Learning (FTL) is gaining traction as a means to improve model performance across diverse devices. This paper explores the integration of Federated Learning and Transfer Learning approaches to enhance IoT security, particularly in intrusion detection systems (IDS). By leveraging pre-trained models and transferring knowledge across devices, FTL helps overcome the challenges posed by limited data and diverse network environments. We propose an FTL-based framework for intrusion detection, focusing on its ability to generalize across various IoT devices and network conditions. Experimental results demonstrate the effectiveness of this approach, providing enhanced detection accuracy, faster convergence, and reduced communication overhead compared to traditional methods. (shrink)

Abstract: A corn crop typically refers to a large-scale cultivation of corn (also known as maize) for commercial purposes such as food production, animal feed, and industrial uses. Corn is one of the most widely grown crops in the world, and it is a major staple food for many cultures. Corn crops are grown in various regions of the world with different climates, soil types, and farming practices. In the United States, for example, the Midwest is known as the "Corn (...) Belt" due to its extensive corn production. Corn crops can be grown using a variety of methods, including conventional tillage, no-till farming, and the use of genetically modified crops. Due to its importance, there is a need for discovering the corn diseases and treating them. The aim of this study is to propose a deep learning model for the classification of corn leaf diseases based on Convolutional Neural Network. A model for classifying maize diseases was developed using a dataset for classifying Corn diseases that contains 4 classes of disease. The dataset was collected from Keggel website. The proposed model was trained, validated, and tested. The F1-score (99.83%), Recall (99.83%), precision (99.83%), Accuracy (99.83%). (shrink)

FarmSmart is an integrated digital agriculture platform that seeks to empower farmers with data-driven, intelligent decision-making. It brings together six must-have modules such as crop recommendation, fertilizer recommendation, crop disease forecasting, farmer-to-farmer marketplace, live commodity price tracking, and multilingual conversational chatbot into one integrated, easy-to-use solution, specifically designed for rural environments. With the combined strength of machine learning, computer vision, natural language processing, and realtime APIs of government data, FarmSmart is a holistic end-to-end solution for enabling farmers right through (...) the farming process. It includes suggesting the optimal crop in terms of environment and soil, suggesting the application of certain fertilizers, detection of crop disease using leaf image analysis through deep learning algorithms (ResNet-9), and enabling informed decisions regarding the time and place of sale of crops through live mandi price comparison. FarmSmart seeks to maximize productivity and profitability. In addition, the integrated e-commerce platform supports direct farmer-to-farmer transactions for farm produce, equipment, and produce sales and purchases, essentially doing away with intermediary functions. The platform has a multilingual AI chatbot, created by GROQ, that provides conversational assistance and voice assistance, thus improving usability for farmers from different linguistic backgrounds. Lastly, FarmSmart bridges the critical information gap in rural farming communities, increasing precision agriculture, reducing wastage of resources, and improving livelihoods through the integration of modern technology into traditional farming practices. (shrink)

Abstract: Fish are important for both nutritional and economic reasons. They are a good source of protein, vitamins, and minerals and play a significant role in human diets, especially in coastal and island communities. In addition, fishing and fish farming are major industries that provide employment and income for millions of people worldwide. Moreover, fish play a critical role in marine ecosystems, serving as prey for larger predators and helping to maintain the balance of aquatic food chains. Overall, fish play (...) a vital role in supporting human well-being and the health of our planet. . Fishes have a lots of types such that: Red Mullet, Sea Bass, Striped Red Mullet and Shrimp. Each of them has its own shape and characteristics that differ from other types. We proposed a system that recognize nine types of fishes using deep learning. We trained the model with a dataset that contain 9000 images that were slit into 6300 images for training, 1350 for validation and 1350 for testing. The proposed model achieved accuracy (99.68%), precision (99.69%), recall (99.68%), and f1-score (99.68%). This indicates that our proposed model can effectively predicate and classify different types of fish with very high accuracy. (shrink)

The early and accurate detection of plant diseases plays a vital role in minimizing crop damage and enhancing agricultural productivity. Traditional methods for identifying plant diseases—such as manual observation and expert consultations—are often time-consuming, costly, and reliant on the availability of skilled personnel. To overcome these limitations, this study presents an automated system for plant disease detection using advanced machine learning techniques. The proposed framework utilizes convolutional neural networks (CNNs), specifically pretrained models like ResNet and SqueezeNet, to analyze images (...) of plant leaves and classify them as healthy or diseased. The system begins with the collection and preprocessing of a diverse dataset comprising both healthy and infected leaf samples. Image augmentation techniques such as rotation, scaling, and flipping are applied to increase dataset variability and enhance model robustness. Subsequently, feature extraction is conducted through CNN-based models that capture color, shape, and texture-based characteristics from the leaf images. Transfer learning is employed to finetune the models on the specific dataset, leveraging previously learned features for improved accuracy. The performance of the trained models is evaluated using key metrics including accuracy, precision, recall, and F1- score. Experimental results indicate a high degree of classification accuracy and generalization capability on unseen data. The final model is deployed through a user-friendly interface that can be integrated into mobile or web-based applications, allowing farmers to upload leaf images and receive instant disease diagnosis and possible treatment recommendations. The proposed system holds significant promise for real-world applications, particularly in regions with limited access to agricultural experts. Moreover, its scalability and adaptability pave the way for future extensions to other crops and disease types. This research underscores the transformative potential of machine learning in precision agriculture, offering a cost-effective, scalable, and efficient solution for early plant disease detection. (shrink)

Bone Fractures are a prevalent clinical condition that necessitates and timely diagnosis. Conventional manual identification of fractures from X-ray images is slow and error-prone. In response to these issues, this paper introduces a deep learning approach for automated detection and classification of bone fractures. A Convolutional Neural Network (CNN) model was implemented and trained on an X-ray image dataset, using data augmentation methods to improve generalization and avoid overfitting. The suggested model was tested with various experimental configurations, such (...) as cross-validation and showed better performance than current state-of-the-art techniques. Adam optimizer helped achieve better classification accuracy than traditional methods. Additional improvements, including sophisticated augmentation methods, transfer learning and ensemble models, may enhance predication accuracy. This research underscores the potential of deep learning in medical imaging for precise and computerized fracture identification, lowering diagnosis time and enhancing patient outcomes. (shrink)

Speech Emotion Recognition (SER) is an essential component in human-computer interaction, enabling systems to understand and respond to human emotions. Traditional emotion recognition methods often rely on handcrafted features, which can be limited in capturing the full complexity of emotional cues. In contrast, deep learning approaches, particularly convolutional neural networks (CNNs), recurrent neural networks (RNNs), and long short-term memory (LSTM) networks, offer more robust solutions by automatically learning hierarchical features from raw audio data. This paper reviews recent (...) advancements in deep learning-based speech emotion recognition, discusses the various architectures used, and evaluates the challenges in real-world applications. We focus on the application of deep learning models to enhance the accuracy and robustness of SER, particularly in noisy environments. The study also discusses future directions for research, including multimodal emotion recognition and transfer learning to address challenges such as small datasets and cross-domain applications. (shrink)

Abstract: Dates are the fruit of date palm trees, and it is one of the fruits famous for its high nutritional value. It is a summer fruit spread in the Arab world. In the past, the Arabs relied on it in their daily lives. Dates take an oval shape and vary in size from 20 to 60 mm in length and 8 to 30 mm in diameter. The ripe fruit consists of a hard core surrounded by a papery cover called (...) the tartar that separates the core from the fleshy part that is eaten. Historians disagreed about the place of its origin, so some historians expressed their belief that it had originated around the Arabian Gulf, and some of them say that the oldest known information about palm trees was in Babylon 4 thousand years ago BC. It is known about the ancient Egyptians using dates in wine. Dates contain a high nutritional value and are considered a basic food for man since ancient times. The fruits of dates are considered the highest fruits in containing sugars. These components vary according to the nature of the fruit, whether it is wet, semi-dry or dry, as well as according to the environmental conditions surrounding the trees. The components of the fruits also differ in different varieties and increase the percentage of sugars in dates is 70-78% of the fruit’s components. These sugars are characterized by their rapid absorption and transfer into the blood directly, digestion and burning. 10 dates (about 100 grams) per day provide a person with all his daily needs of magnesium, manganese, copper, sulfur and half of his needs of calcium and potassium. In this paper we presented a system that recognize the nine types of dates based on deep learning using python on Colab, and classifying using a dataset contain 1350 images. Our trained model achieved an accuracy of 99.44% on a proven test set. (shrink)

The article conceptualizes the term Pedagogy of Fear as the master narrative of educational systems around the world. Pedagogy of Fear stunts the active and vital educational growth of the young person, making him/her passive and dependent upon external disciplinary sources. It is motivated by fear that prevents young students—as well as teachers—from dealing with the great existential questions that relate to the essence of human beings. One of the techniques of the Pedagogy of Fear is the internalization of the (...) view that without evaluation and assessment we cannot know a child’s level or “worth”—and therefore are unable to help him/her if he is “slow in learning.” In contrast, Philosophy for/with Children offers a space for addressing existential questions, some of which deal with urgent social issues. The willingness to make philosophy inquiry an alternative already from an early age seeks to allow the child to challenge him/herself with new and fresh questions. Philosophy for/with Children does not regard children as a “space of lack” (experience, knowledge, values, etc.) The new and fresh philosophical perspective of children demands the presence of a willingness to engage in dialogue and rejection of the fear of the innocent and deep questions of philosophy. Shaking free of the Pedagogy of Fear and restoring honor to children’s questions demands a fundamental conceptual change within education. The replacement of existential certainty as it is depicted by adults in the existing education system with an existential question is a heavy intellectual task that in most cases is viewed as subversive—primarily on the part of the adult. It demands a return to starting points and a willingness to allow children a free and safe educational space in which to ground preliminary and fertile questions about themselves, their lives, their environment, and, most of all, the changing world they discover with the form of originality that is right for them. (shrink)

Abstract: Sign language recognition is one of the most rapidly expanding fields of study today. Many new technologies have been developed in recent years in the fields of artificial intelligence the sign language-based communication is valuable to not only deaf and dumb community, but also beneficial for individuals suffering from Autism, downs Syndrome, Apraxia of Speech for correspondence. The biggest problem faced by people with hearing disabilities is the people's lack of understanding of their requirements. In this paper we try (...) to fill this gap. By trying to translate sign language using artificial intelligence algorithms, we focused in this paper using transfer learning technique based on deep learning by utilizing a MobileNet algorithm and compared it with the previous paper results[10a], where we get in the Mobilenet algorithm on the degree of Accuracy 93,48% but the VGG16 the accuracy was 100% For the same number of images (43500 in the dataset in size 64*64 pixel ) and the same data split training data into training dataset (70%) and validation dataset(15%) and testing dataset(15%) and 20 epoch . (shrink)

American Sign Language, or ASL as its acronym is commonly known, is a fascinating language, and many people outside of the Deaf community have begun to recognize its value and purpose. It is a visual language consisting of coordinated hand gestures, body movements, and facial expressions. Sign language is not a universal language; it varies by country and is heavily influenced by the native language and culture. The American Sign Language alphabet and the British Sign Language alphabet are completely contrary. (...) Fingerspelling is one-handed in ASL and two-handed in BSL, with the exception of the letter C. AI technologies, particularly deep learning, can play an important role in breaking down communication barriers between deaf or hearing-impaired people and other communities, significantly contributing to their social inclusion. Recent advances in sensing technologies and AI algorithms have paved the way for the development of a wide range of applications aimed at meeting the needs of the deaf and hearing-impaired communities. To that end, the purpose of this paper is By attempting to translate sign language using artificial intelligence algorithms, we focused in this paper on a transfer learning technique based on deep learning using a ResNet algorithm and compared it to previous papers in which we used VGG19[1]and mobileNet [2], where we get in the ResNet algorithm on the degree of Accuracy 93,48% e number of images (43500 in the dataset in size 64*64 pixel ) and the data split training data into training dataset (70%) and validation dataset(15%) and testing dataset(15%) and 20 epoch. (shrink)

Abstract: Accurate identification of marine species is critical for effective fishery management, biodiversity conservation, and the aquaculture industry. Traditional methods of fish identification rely on expert knowledge and manual labor, making them time- consuming, expensive, and error-prone. In this research, we explore a machine learning-based approach to automate the classification of nine fish species using image recognition techniques. The fish species under study include Black Sea Sprat, Gilt- Head Bream, Horse Mackerel, Red Sea Bream, Shrimp, Trout, Striped Red Mullet, (...) Sea Bass, and Red Mullet. We collected 1000 images per species, yielding a total of 9000 images, with 750 used for training and 250 reserved for testing. We utilized convolutional neural networks (CNNs) pre-trained on the ImageNet dataset and fine-tuned them to perform fish species classification. Preliminary experiments showed that CNNs, particularly transfer learning models like VGG16 and ResNet50, achieved accuracy rates as high as 98.6% on the test set. This research highlights the potential of deep learning in automating marine species identification, enabling faster, more reliable monitoring. (shrink)

Text classification is one of the most important task in natural language processing, In this research, we carried out several experimental research on three (3) of the most popular Text classification NLP classifier in Convolutional Neural Network (CNN), Multinomial Naive Bayes (MNB), and Support Vector Machine (SVN). In the presence of enough training data, Deep Learning CNN work best in all parameters for evaluation with 77% accuracy, followed by SVM with accuracy of 76%, and multinomial Bayes with least (...) performance of 69% accuracy. CNN has the best performance in the presence of large enough training dataset because of the presence of filter/ kernels which help to indentify patterns in text data regardless of their position in the sentence. We repeated the training again with just one-third of our data, at this point SVM comes with the best performance, the performance of CNN noticeably drops but still better than Multinomial Naive Bayes, the reason why SVM works best when we reduce the training data was because of its ability to look for a hyper-plane that creates a boundary between different classes of data so as to properly classify them, so we believed that getting the hyper-plane was more efficient when we reduce the dataset, hence reason for the good performance. Multinomial Naive Bayes have the least performance which we attributed to its assumption of independency between the features which sometimes does not hold true. We concluded that availability of data should be an important factor when choosing classifier for Natural Language Processing Text Classification task. CNN should be use in the presence of enough dataset, and SVM should be use when data is not enough. Multinomial Naive Bayes must not be trusted with state of the art NLP task due to its assumption of independency between the features . (shrink)

Guava is a widely cultivated tropical fruit crop valued for its nutritional richness and economic benefits. However, guava production is often hindered by the prevalence of various leaf diseases, such as anthracnose, rust, and bacterial leaf spot, which significantly affect both yield and fruit quality. Timely and accurate detection of these diseases is critical to prevent widespread damage and ensure better crop management. Traditional disease detection techniques primarily rely on manual observation and expertise, which are not only time-consuming (...) and labor-intensive but also prone to human error. These limitations highlight the need for an automated, efficient, and scalable solution for guava disease identification. This project proposes a deep learning-based approach for the automated detection of guava leaf diseases using Convolutional Neural Networks (CNNs). (shrink)

The increasing demand for portable, cost-effective, and energy-efficient computing solutions has led to the development of compact mini computers. In this paper, we’ve designed and built a compact mini computer using the BusyBirds chip. The system includes a built-in power bank for better portability, an HDMI port for connecting to high-definition displays, and Bluetooth 5.0 for reliable wireless communication. It’s developed as a simple, affordable computing option for students, developers, and professionals who need a small, efficient system for different tasks. (...) The BusyBirds Chip is used as the main processing unit, offering better performance while keeping power usage low. One of the standout features of this mini computer is that, unlike regular desktops or laptops that rely on being plugged in all the time, this one can run on battery power. That makes it a handy option for working in places where there’s no easy access to electricity. The in-built power bank helps keep the system running smoothly without interruptions, which makes it especially useful for learning activities and IoT projects. Modern systems need strong connectivity options, and this setup includes both HDMI and Bluetooth 5.0 to improve usability. The HDMI output lets you connect to highquality displays, while Bluetooth 5.0 makes it simple to pair devices like a mouse, keyboard, or speakers. After testing its performance, results showed that the mini computer works as expected offering a good mix of portability, efficiency, and low cost. (shrink)

Deep learning algorithms are rapidly changing the way in which audiovisual media can be produced. Synthetic audiovisual media generated with deep learning—often subsumed colloquially under the label “deepfakes”—have a number of impressive characteristics; they are increasingly trivial to produce, and can be indistinguishable from real sounds and images recorded with a sensor. Much attention has been dedicated to ethical concerns raised by this technological development. Here, I focus instead on a set of issues related to the (...) notion of synthetic audiovisual media, its place within a broader taxonomy of audiovisual media, and how deep learning techniques differ from more traditional approaches to media synthesis. After reviewing important etiological features of deep learning pipelines for media manipulation and generation, I argue that “deepfakes” and related synthetic media produced with such pipelines do not merely offer incremental improvements over previous methods, but challenge traditional taxonomical distinctions, and pave the way for genuinely novel kinds of audiovisual media. (shrink)

Text classification is one of the most important task in natural language processing, In this research, we carried out several experimental research on three (3) of the most popular Text classification NLP classifier in Convolutional Neural Network (CNN), Multinomial Naive Bayes (MNB), and Support Vector Machine (SVN). In the presence of enough training data, Deep Learning CNN work best in all parameters for evaluation with 77% accuracy, followed by SVM with accuracy of 76%, and multinomial Bayes with least (...) performance of 69% accuracy. CNN has the best performance in the presence of large enough training dataset because of the presence of filter/ kernels which help to indentify patterns in text data regardless of their position in the sentence. -/- We repeated the training again with just one-third of our data, at this point SVM comes with the best performance, the performance of CNN noticeably drops but still better than Multinomial Naive Bayes, the reason why SVM works best when we reduce the training data was because of its ability to look for a hyper-plane that creates a boundary between different classes of data so as to properly classify them, so we believed that getting the hyper-plane was more efficient when we reduce the dataset, hence reason for the good performance. Multinomial Naive Bayes have the least performance which we attributed to its assumption of independency between the features which sometimes does not hold true. We concluded that availability of data should be an important factor when choosing classifier for Natural Language Processing Text Classification task. CNN should be use in the presence of enough dataset, and SVM should be use when data is not enough. Multinomial Naive Bayes must not be trusted with state of the art NLP task due to its assumption of independency between the features. (shrink)

Current state-of-the-art out-of-distribution algorithm does not address the variation in dynamic and static behavior between malware variants from the same family as evidence in their poor performance against an out-of-distribution malware attack. We aims to address this limitation by: 1) exploitation of the in-dimensional embedding space between variants from the same malware family to account for all variations 2) exploitation of the inter-dimensional space between different malware family 3) building a deep learning-based model with a shallow neural network (...) with maximum of two connected layers to overcome overfitting from the scratch 4) building a Bayesian inference based computation algorithm that intertwine with connected network and is able to create new and adjust existing data points in response to an exposure to new out-of-distribution variants of existing or new malware family which determines the extent at which model weight is adjusted which in turn triggers update on the gradient. Preliminary result of our proposed framework gave an accuracy of 81\% in the successful classification of a novel out-of-distribution malware attack, something that could not be achieve by any of the state-of-the-art algorithms on novel malware classification. (shrink)

We claim that scientists working with deep learning (DL) models exhibit a form of pragmatic understanding that is not reducible to or dependent on explanation. This pragmatic understanding comprises a set of learned methodological principles that underlie DL model design-choices and secure their reliability. We illustrate this action-oriented pragmatic understanding with a case study of AlphaFold2, highlighting the interplay between background knowledge of a problem and methodological choices involving techniques for constraining how a model learns from data. Building (...) successful models requires pragmatic understanding to apply modelling strategies that encourage the model to learn data patterns that will facilitate reliable generalisation. (shrink)

Current state-of-the-art out-of-distribution algorithm does not address the variation in dynamic and static behavior between malware variants from the same family as evidence in their poor performance against an out-of-distribution malware attack. We aims to address this limitation by: 1) exploitation of the in-dimensional embedding space between variants from the same malware family to account for all variations 2) exploitation of the inter-dimensional space between different malware family 3) building a deep learning-based model with a shallow neural network (...) with maximum of two connected layers to overcome overfitting from the scratch 4) building a Bayesian inference based computation algorithm that intertwine with connected network and is able to create new and adjust existing data points in response to an exposure to new out-of-distribution variants of existing or new malware family which determines the extent at which model weight is adjusted which in turn triggers update on the gradient. Preliminary result of our proposed framework gave an accuracy of 81\% in the successful classification of a novel out-of-distribution malware attack, something that could not be achieve by any of the state-of-the-art algorithms on novel malware classification. (shrink)

Internet of Things (IoT) devices have transformed various industries, enabling advanced functionalities across domains such as healthcare, smart cities, and industrial automation. However, the increasing number of connected devices has raised significant concerns regarding their security. IoT networks are highly vulnerable to a wide range of cyber threats, making Intrusion Detection Systems (IDS) critical for identifying and mitigating malicious activities. This paper proposes a hybrid approach for intrusion detection in IoT networks by combining Machine Learning (ML) techniques with Signature-Based (...) Methods. The hybrid model leverages the strengths of both methodologies to achieve high detection accuracy, reduced false positives, and the ability to identify both known and unknown threats. We explore the integration of ML classifiers such as Random Forest, Support Vector Machines, and k-Nearest Neighbors with traditional signature-based techniques to create a robust IDS solution. The effectiveness of the proposed approach is evaluated using a publicly available IoT dataset, demonstrating its capability to detect a wide variety of attacks with high precision and recall. (shrink)

Abstract: Tomatoes are among the most popular vegetables in the world due to their frequent use in many dishes, which fall into many varieties in common and traditional foods, and due to their rich ingredients such as vitamins and minerals, so they are frequently used on a daily basis, When we focus our attention on this vegetable, we must also focus and take into consideration the diseases that affect this vegetable, a deep learning model that classifies tomato diseases (...) has been proposed. The aim of this paper is to diagnose tomato diseases based on a dataset containing 11000 picture and 11 classes. The model gave the final accuracy test with Accuracy (99.87%), F1-score (99.87%), Recall (99.87%), and Precision (99.88%) in an estimated time 1.36 second, as with these results it proved its effectiveness and good ability in the classification of the testing data. (shrink)

Near the end of December 2019, the globe was hit with a major crisis, which is nothing but the coronavirusbased pandemic. The authorities at the train station should also keep in mind the need to limit the spread of the covid virus in the event of a global pandemic. When it comes to controlling the COVID-19 epidemic, public transportation facilities like train stations play a pivotal role because of the proximity of so many people who may be exposed to the (...) virus. Using common place CCTV cameras and deep learning with simple online and real-time (DeepSORT) methods, this study develops social distance monitoring using a YOLOv4 identification of a Surveillance Object Model. Based on experiments conducted with a minicomputer equipped with an Intel 11th Gen Intel(R) Core(TM) i3-1115G4 at 3.00GHz, 2995 Mhz, two Core(s), four Logical processor, four gigabytes of random-access memory (RAM), this paper makes use of CCTV surveillance, which was put into practice at the Guindy railway station, Chennai, Tamilnadu in India in order to detect the violation of social distancing. (shrink)

Training deep neural networks (DNNs) has become a cornerstone of modern artificial intelligence, powering advancements in computer vision, natural language processing, and autonomous systems. However, the computational demands of DNN training are immense, necessitating the use of high-performance hardware like Graphics Processing Units (GPUs). Cloud computing has revolutionized this domain by offering scalable, on-demand GPU resources that dramatically accelerate model development and deployment. This paper explores the methodology, benefits, and challenges of training DNNs at scale using cloud-based GPUs. We (...) examine the technical underpinnings that make cloud GPUs suitable for deep learning, including parallel computation, memory optimization, and integration with popular machine learning frameworks such as TensorFlow and PyTorch. A critical review of cloud platforms—such as AWS EC2 P-series, Google Cloud TPUs, and Azure NVseries—highlights the unique capabilities and pricing structures that influence deployment strategies. We also discuss training optimization techniques including mixed-precision training, data parallelism, and container orchestration with Kubernetes. Key findings emphasize significant improvements in training time, model accuracy, and development agility when leveraging cloud GPU infrastructure. Nonetheless, challenges such as cost management, data transfer bottlenecks, and dependency on proprietary ecosystems persist. This study also presents use cases in domains like autonomous driving and large language model (LLM) development, demonstrating the transformative impact of cloud GPUs. The paper concludes with a proposed workflow for cloud-based DNN training and outlines future research directions, including green AI, federated learning with GPUs, and intelligent resource allocation. By following best practices and leveraging cloud-native tools, organization. (shrink)

Real-time monitoring systems use surveillance videos to automatically detect motorcycle helmet requirements and triple-riding violations, which protect road safety. Deep learning methods currently show practical worth for addressing surveillance system constraints because they have developed superior capabilities in object detection and classification. The models deliver poor results repeatedly because they are limited by low-resolution video, together with adverse weather conditions, as well as problems from occlusions and deficient illumination conditions. The issue of recognizing multiple individuals riding together on (...) one motorcycle requires better solution methods because it remains unresolved. YOLOv8 operates in real-time to detect both helmet violations and multiple riders on vehicles using CCTV video and webcam feeds. The detection system reaches a 95% accuracy level, thereby establishing itself as an effective monitoring method in automated traffic control. A YOLOv8 detection model serves for detecting triple riders, along with YOLOv8-based detection of helmet violations. The model has tailored detection skills since its training with helmet-specific data enables it to monitor properly non-helmeted riders. The system supplies capabilities for processing video and image data, which helps it create visual violation marks for practical realworld applications. Quality evaluation of the system happens through precision and recall metrics, which measure its reliable operation functions. The method utilizes publicly available data and self-collected information to solve illumination conditions and issues that affect detection systems. The evolution of traffic rule enforcement systems has improved thanks to the resolution of implementation problems. The research application proves how deep learning transforms surveillance systems, which advances road safety through system improvements. Better traffic management possibilities, combined with improved efficiency and innovation, enable wider expansion of safety programs in urban areas. (shrink)

The identification of disease on the plant is a very important key to prevent a heavy loss of yield and the quantity of agricultural product. The symptoms can be observed on the parts of the plants such as leaf, stems, lesions and fruits. The leaf shows the symptoms by changing color, showing the spots on it. This identification of the disease is done by manual observation and pathogen detection which can consume more time and may prove costly. The (...) aim of the project is to identify and classify the disease accurately from the leaf images. The steps required in the process are Preprocessing, Training and Identification. For identification of disease features of leaf such as major axis, minor axis etc. are extracted from leaf and given to classifier for classification. In our project we have used cassava leaf for identifying its disease. We have used algorithms namely SqueezeNet as existing and ResNet-50 as proposed system in terms of accuracy. From the results its proved that ResNet-50 works better than SqueezeNet. The tool we have used for the detection of leaf disease is MATLAB. (shrink)

The ever-evolving ways attacker continues to improve their phishing techniques to bypass existing state-of-the-art phishing detection methods pose a mountain of challenges to researchers in both industry and academia research due to the inability of current approaches to detect complex phishing attack. Thus, current anti-phishing methods remain vulnerable to complex phishing because of the increasingly sophistication tactics adopted by attacker coupled with the rate at which new tactics are being developed to evade detection. In this research, we proposed an adaptable (...) framework that combines Deep learning and Randon Forest to read images, synthesize speech from deep-fake videos, and natural language processing at various predictions layered to significantly increase the performance of machine learning models for phishing attack detection. (shrink)

The COVID-19 outbreak has had a significant influence on the health of people all across the world, and preventing its further spread requires an early and correct diagnosis. Imaging using X-rays is often used to identify respiratory disorders like COVID-19, and approaches based on machine learning may be used to automate the diagnostic process. In this research, we present a deep learning approach for COVID-19 identification in X-ray pictures utilizing global thresholding. Our framework consists of two main (...) components: (1) global thresholding to preprocess X-ray images and extract features, and (2) a Convolutional Neural Network (CNN) for classifying the images as positive (indicating COVID-19) or negative (indicating no COVID-19). Global thresholding is used to convert X-ray images into binary images, which can highlight the features of interest, such as lung opacity. The Network is instructed to use a dataset consisting of X-ray pictures that have been labeled with the COVID-19 status of each image. The usefulness of our approach was shown by experimental findings obtained using a dataset of X-ray pictures that were made accessible to the public. We achieve high accuracy in COVID-19 detection, outperforming traditional machine learning techniques. Our framework is also able to correctly identify COVID-19 in images that were misclassified by human experts. (shrink)

Abstract: Lotteries have long been a source of fascination and intrigue, offering the tantalizing prospect of unexpected fortunes. In this research paper, we delve into the world of lottery predictions, employing cutting-edge AI techniques to unlock the secrets of lottery outcomes. Our dataset, obtained from Kaggle, comprises historical lottery draws, and our goal is to develop predictive models that can anticipate future winning numbers. This study explores the use of deep learning and time series analysis to achieve this (...) elusive feat. Through rigorous experimentation and data-driven approaches, we seek to determine the viability of AI in the realm of lottery predictions. Our findings reveal both the promise and limitations of AI in this context, shedding light on the complexities of lottery data and the potential need for quantum computing as a last resort. (shrink)

Given the growing dependence on digital systems and the escalation of financial fraud occurrences, it is imperative to implement efficient cyber security protocols and fraud detection methodologies. The threat's dynamic nature often challenges conventional methods, necessitating the adoption of more sophisticated strategies. Individuals depend on pre-established regulations or problem-solving processes, which possess constraints in identifying novel and intricate fraudulent trends. Conventional techniques need help handling noise data and the substantial expenses incurred by false positives and true positives. To tackle these (...) obstacles, the present study introduces Deep Fraud Net, a framework that utilizes deep learning to detect and classify instances of financial fraud and cyber security threats. The Deep Fraud Net system model entails the utilization of a deep neural network to acquire intricate patterns and characteristics from extensive datasets through training. The framework integrates noise reduction methods to enhance the precision of fraud detection and improve the quality of input data. The Deep Fraud Net method attains a precision of 98.85%, accuracy of 93.35%, sensitivity of 99.05%, specificity of 93.16%, false positive rate of 7.34%, and true positive rate of 89.58%. The findings suggest that Deep Fraud Net can effectively detect and categorize instances of fraudulent behavior with a reduced occurrence of misclassifications. The method exhibits potential implications for diverse domains that prioritize robust security and fraud detection, including but not limited to banking, e-commerce, and online transactions. (shrink)

This study explores sentiment analysis of Amazon reviews using NLP and deep learning techniques to understand customer opinions. The main objective is to develop an NLP model using TensorFlow to analyze sentiment in customer reviews. By classifying reviews as positive or negative, the study aims to provide insights into customer satisfaction. The process involves collecting Amazon review data, cleaning the text, and applying deep learning methods to train the model. Various NLP techniques, such as tokenization and (...) word embeddings, help the model understand language patterns. The performance of the model is evaluated to ensure accurate sentiment classification. Challenges like handling sarcasm and ambiguous reviews are considered. The study highlights the importance of deep learning in improving sentiment analysis accuracy. The findings can be useful for businesses to enhance customer experience based on feedback. Future work may explore more advanced techniques or combine text with other forms of data for a more detailed analysis. (shrink)

This paper explores the potential of a decision support system for aerospace design that integrates semantic vector search and transfer learning. The study investigates the application of transfer learning to adapt pre-trained neural networks for aerospace-specific datasets and examines how semantic vector search can transform complex design data into high-dimensional vector representations to reveal latent relationships. Various supervised and unsupervised learning approaches are evaluated to address different phases of the aerospace lifecycle. Preliminary validation on both (...) synthetic and real-world datasets indicates promising improvements in design analysis efficiency and predictive accuracy, laying a foundation for further development of AI-based decision support systems in aerospace engineering. (shrink)

Artificial intelligence (AI), deep learning, machine learning and neural networks represent extremely exciting and powerful machine learning-based techniques used to solve many real-world problems. Artificial intelligence is the branch of computer sciences that emphasizes the development of intelligent machines, thinking and working like humans. For example, recognition, problem-solving, learning, visual perception, decision-making and planning. Deep learning is a subset of machine learning in artificial intelligence that has networks capable of learning unsupervised (...) from data that is unstructured or unlabeled. Deep learning is a technique used to generate face detection and recognize it for real or fake by using profile images and determine the differences between them. In this study, we used deep learning techniques to generate models for Real and Fake face detection. The goal is determining a suitable way to detect real and fake faces. The model was designed and implemented, including both Dataset of images: Real and Fake faces detection through the use of Deep learning algorithms based on neural networks. We have trained dataset which consists of 9,000 images for total in 150 epochs, and got the ResNet50 model to be the best model of network architectures used with 100% training accuracy, 99.18% validation accuracy, training loss 0.0003, validation loss 0.0265, and testing accuracy 99%. (shrink)

A new trend in deep learning, represented by Mutual Information Neural Estimation (MINE) and Information Noise Contrast Estimation (InfoNCE), is emerging. In this trend, similarity functions and Estimated Mutual Information (EMI) are used as learning and objective functions. Coincidentally, EMI is essentially the same as Semantic Mutual Information (SeMI) proposed by the author 30 years ago. This paper first reviews the evolutionary histories of semantic information measures and learning functions. Then, it briefly introduces the author’s semantic (...) information G theory with the rate-fidelity function R(G) (G denotes SeMI, and R(G) extends R(D)) and its applications to multi-label learning, the maximum Mutual Information (MI) classification, and mixture models. Then it discusses how we should understand the relationship between SeMI and Shannon’s MI, two generalized entropies (fuzzy entropy and coverage entropy), Autoencoders, Gibbs distributions, and partition functions from the perspective of the R(G) function or the G theory. An important conclusion is that mixture models and Restricted Boltzmann Machines converge because SeMI is maximized, and Shannon’s MI is minimized, making information efficiency G/R close to 1. A potential opportunity is to simplify deep learning by using Gaussian channel mixture models for pre-training deep neural networks’ latent layers without considering gradients. It also discusses how the SeMI measure is used as the reward function (reflecting purposiveness) for reinforcement learning. The G theory helps interpret deep learning but is far from enough. Combining semantic information theory and deep learning will accelerate their development. (shrink)

Internet of Things (IoT) devices have transformed various industries, enabling advanced functionalities across domains such as healthcare, smart cities, and industrial automation. However, the increasing number of connected devices has raised significant concerns regarding their security. IoT networks are highly vulnerable to a wide range of cyber threats, making Intrusion Detection Systems (IDS) critical for identifying and mitigating malicious activities. This paper proposes a hybrid approach for intrusion detection in IoT networks by combining Machine Learning (ML) techniques with Signature-Based (...) Methods. The hybrid model leverages the strengths of both methodologies to achieve high detection accuracy, reduced false positives, and the ability to identify both known and unknown threats. We explore the integration of ML classifiers such as Random Forest, Support Vector Machines, and k-Nearest Neighbors with traditional signature-based techniques to create a robust IDS solution. The effectiveness of the proposed approach is evaluated using a publicly available IoT dataset, demonstrating its capability to detect a wide variety of attacks with high precision and recall. (shrink)

Abstract: Age and gender classification has been around for a long time, and efforts are still being made to improve the findings. This has been the case since the inception of social media platforms. Visible understanding has become more important in the computer vision society with the emergence of AI increase in performance and help train a model to achieve age and gender classification. Although these networks built for the mobile platform are not always as accurate as the larger, more (...) resource- intensive networks we've come to know and love, they stand out when it comes to the accuracy trade-off. Despite the importance of these attributes in our daily lives, the ability to estimate them efficiently from face images is still far from meeting the requirements of commercial applications. (shrink)

This study explores the application of deep learning and image processing techniques for the detection of skin cancer. Leveraging convolutional neural networks (CNNs) and advanced image processing algorithms, the proposed system aims to accurately identify and classify skin lesions. The model is trained on a diverse dataset, encompassing various skin conditions, to enhance its diagnostic capabilities. Results demonstrate the potential for automated and reliable skin cancer detection, offering a promising approach for early diagnosis and improved patient outcomes. The (...) deep learning model is trained on a comprehensive dataset, including various types of skin lesions and conditions, to ensure robust performance across a spectrum of cases. Image preprocessing techniques are employed to enhance feature extraction and improve the model's ability to discern subtle patterns indicative of skin cancer. The study further investigates the interpretability of the deep learning model, employing techniques to visualize and understand the decision-making process. This transparency aids in building trust in the system's predictions and facilitates collaboration between AI and medical practitioners. As the landscape of healthcare continues to evolve, the combination of deep learning and image processing offers a scalable and efficient solution for skin cancer detection, fostering advancements in early intervention and personalized patient care. (shrink)

Coronavirus disease has a crisis with high spread throughout the world during the COVID19 pandemic period. This disease can be easily spread to a group of people and increase the spread. Since it is a worldly disease and not plenty of vaccines available, social distancing is the only best approach to defend against the pandemic situation. All the affected countries' governments declared locked-down to implement social distancing. This social separation and persons not being in a mass group can slow down (...) the spread of COVID19. It reduces the physical contact between infected persons and normal healthy persons. Almost every health organization tells that to follow social distancing people should maintain at least 6 feet of distance from each other. This research proposes a deep learning approach for social distancing which is developed for tracking and detecting people who are in indoor as well as outdoor scenarios using YOLO V3 video analytic technique. This approach focuses to inspect whether the people are maintaining social distancing in many areas, using surveillance video with measuring the distance in real-time performance. Most of the early studies of detecting social distance monitoring were based on GPS for tracking the movements of people where the signals could be lost. On the other hand, some countries use drones to detect large gatherings of people who cannot have a clear view at night times [10]. In the future, the proposed system can be used fully for detecting threats in the public crowded or it can detect any person affected by critical situations (ie fainting, Cordia arrest) or planting the crops in the forms evenly with a uniform measurement. This proposal could be used in many fields like crowd analysis, autonomous vehicles, and human action recognition and could help the government authorities to redesign the public place layout and take precautionary action in the risk zones. This system analyses the social distancing of people by calculating the distance between people to slow downing the spread of the COVID 19 virus. (shrink)

The growing adoption of deep learning models in critical domains, such as healthcare, finance, and autonomous systems, has highlighted the need for interpretability and transparency. Explainable AI (XAI) aims to provide insights into the decision-making processes of complex models, improving their trustworthiness and enabling accountability. However, one of the key challenges is balancing the trade-off between model transparency and performance. While explainability can sometimes compromise the predictive power of models, deep learning, with its inherent complexity, exacerbates (...) this issue. This paper explores the current landscape of XAI in deep learning, examining methods such as LIME, SHAP, and attention mechanisms. We also investigate the impact of explainability on model accuracy, fairness, and user trust, and propose strategies to achieve a balance between transparency and performance. Finally, we present case studies where XAI has been successfully integrated into deep learning applications, demonstrating its potential in various real-world scenarios. (shrink)

Scalability has emerged as a defining challenge and opportunity in deploying deep learning models. As datasets grow and models become more complex, traditional computing infrastructure often falls short in providing the necessary resources for training and inference. Cloud infrastructure offers a compelling solution, enabling the deployment of deep learning models at scale through dynamic resource provisioning, GPU and TPU acceleration, and integrated services. This paper investigates the role of cloud platforms—such as AWS, Microsoft Azure, and Google (...) Cloud—in making scalable AI a practical reality. We examine the infrastructure elements that support deep learning workflows, including containerization, orchestration, distributed computing, and model monitoring. The research includes a comprehensive literature review, methodological framework, and analysis of existing deployment architectures and tools like TensorFlow Serving, TorchServe, and Kubernetes-based solutions. Key findings reveal that cloud-native services significantly reduce deployment complexity, offer robust failover and scaling capabilities, and support DevOps and MLOps pipelines. A typical cloud-based workflow is discussed, highlighting stages such as data ingestion, model training, containerization, orchestration, and monitoring. The paper further explores the advantages of cloud deployment, including elasticity, cost-effectiveness, and global accessibility, while also identifying disadvantages such as vendor lock-in, data privacy risks, and latency in edge applications. Future directions include exploring serverless AI, green AI, and federated learning to improve sustainability, cost control, and data governance. (shrink)