As computer networks expand swiftly, it is essential to guarantee that the information system upholds confidentiality, reliability, and accessibility. Intrusion detection systems (IDSs) are essential instruments for overseeing and safeguarding networks. Current intrusion detection systems have two significant issues: the detection rate for zero-day attacks is low, and the prevalence of false-positive alarms is considerable. We need detection systems that can learn to correctly identify intrusions if we are going to solve these problems. Numerous researchers have proposed hybrid solutions grounded in machine learning techniques. These strategies exploit the vulnerabilities of detection techniques to capitalize on their shortcomings. This study evaluates proposed model and shows that it is generating the optimized result with 98.97

Tuberculosis represents a major infectious pathology attributed to the Mycobacterium tuberculosis bacterium (MTB), and delayed detection can lead to serious consequences or death. Existing methods suffer from limited accuracy, poor bacilli segmentation and inadequate feature extraction. To address these issues, a Siberian Tiger Parrot Optimization-powered Deep High-Order Attention Model (STPO_DHA-Net) is proposed for tuberculosis severity-levels from sputum images. STPO is the hybridization of Siberian Tiger Optimization (STO) and Parrot Optimizer (PO), used to tune the parameters of DHA-Net. Initially, a high-boost filter enhances fine structural information. Then, the Shape-aware Loss-based Additive Manufacturing SegNet (AM-SegNet) precisely segments the bacilli. Then, discriminative features are derived using Local Adaptive Regression Kernels (LARK), complemented by Haralick texture descriptors, including entropy, uniformity, contrast, homogeneity and correlation. DHA-Net identifies severity level by focusing on subtle microbial-level patterns. STPO_DHA-Net attained 91.88% accuracy, 92.33% precision, 92.64% True Negative Rate (TNR), 91.65% F1-score and 90.67% True Positive Rate (TPR).

Cerebral tumor segmentation is deemed to be highly critical challenge in therapeutic imaging diagnostics. Early detection of intracranial neoplasms means that it can be easily treated and the patient will most likely survive. The diagnosis of the cancer tumors is an exhaustive and time-intensive activity of isolation of the cerebral tumors in the massive set of magnetic resonance scans produced during the clinical process. We will need some sort of computerized automated system to separate the brain tumors and normal brain images. Automatic segmentation using deep learning offers state-of-the-art solutions compared to traditional methods. Advanced neural network approaches are useful for interpreting large-scale MRI datasets effectively and objectively. There is an abundance of literature describing the best practices in the extraction of neoplastic zones within magnetic resonance imaging-based cerebral images. Many recent studies report tumor classification accuracies above 95

Convolutional neural network (CNN) models used to diagnose brain tumors are very sensitive to hyperparameter tuning, which influences the accuracy and stability of these models. The hyperparameters affect the complexity of a derived model, spatial resolution, convergence rate, feature extraction, and non-linear on the neural network. So as to maximize essential parameters such as filter size and number, batch size, the amount of layers, stride padding, learning rate, activation functions, and pooling modes; we shall introduce a carefully configured CNN hyperparameter model. We also use two open brain tumor sets of MRI in our work. The first set of 7,025 brain scans of human brains falls into 4 categories: meningioma, pituitary, glioma and No Tumor. The second variables lie in a set of 255 pictures labeled as either yes or no. Our approach generates high-quality results as they show average F1-score and precision, recall of 96, 94.25% and percent accuracy on dataset 1 and an average F1-score and precision, recall of 88 and 87.5 percent accuracy on dataset. Efforts are made to carry out an extensive comparison with the existing methods so that the findings can be proven to be correct and the findings of that comparison states that our approach records a consistently high score as compared to the traditional approaches. Our model is also trained after tuning these critical hyperparameters that improves its performance and strengthen its generalization ability. This improved CNN model will give the medical professionals with a more precise and efficient tool to diagnose brain cancers, thus simplifying their decision-making process.

Breast cancer is the foremost cause of mortality among females. Early diagnosis of a disease is necessary to avoid breast cancer by reducing the death rate and offering a better life to the individuals. Therefore, this work proposes a Parallel Convolutional SpinalNet (PConv-SpinalNet) for the efficient detection of breast cancer using mammogram images. At first, the input image is pre-processed using the Gabor filter. The tumour segmentation is conducted using LadderNet. Then, the segmented tumour samples are augmented using Image manipulation, Image erasing, and Image mix techniques. After that, the essential features, like CNN features, Texton, Local Gabor binary patterns (LGBP), scale-invariant feature transform (SIFT), and Local Monotonic Pattern (LMP) with discrete cosine transform (DCT) are extracted in the feature extraction phase. Finally, the detection of breast cancer is performed using PConv-SpinalNet. PConv-SpinalNet is developed by an integration of Parallel Convolutional Neural Networks (PCNN) and SpinalNet. The evaluation results show that PConv-SpinalNet accomplished a superior range of accuracy as 88.5%, True Positive Rate (TPR) as 89.7%, True Negative Rate (TNR) as 90.7%, Positive Predictive Value (PPV) as 91.3%, and Negative Predictive Value (NPV) as 92.5%.

Accurate and efficient detection of furcation involvement in dental X-ray images is crucial for timely diagnosis and treatment planning of periodontal disease. Current methods often rely on manual assessment, which is time-consuming, subjective, and prone to inter-observer variability. This research proposes a novel approach to automated furcation involvement detection using a hybrid deep learning model that combines the strengths of Convolutional Neural Networks (CNNs) and Support Vector Machines (SVMs). The proposed methodology leverages the VGG16 architecture, a pre-trained CNN known for its robust feature extraction capabilities in image classification tasks. To enhance the model’s performance and address the limitations of relying solely on a CNN, SVM classifier is incorporated within a hybrid deep learning framework. Two fusion strategies are explored: early fusion, where the features extracted from the VGG16 fully connected layer are directly concatenated with the SVM; and late fusion, where the outputs of the VGG16 softmax layer and the SVM are combined. This comparative approach allows us to assess the effectiveness of each fusion technique in improving furcation detection accuracy. To assess model performance, 10-fold cross-validation is used, which reduces bias and offers a credible estimate of generalization performance. The models' diagnostic skills are assessed using accuracy, precision, recall, and F1-score. For furcation identification, the hybrid deep learning strategy outperformed the VGG16 model in the testing. Adding the SVM classifier earlier in the feature processing pipeline improves performance, as the early fusion technique consistently beats the late one. These results demonstrate the SVM’s capacity to learn complicated decision boundaries and enhance classification accuracy in this application. The best model, based on early fusion and the SVM classifier, has 91.24% accuracy, 90.9% precision, 93.04% recall, and 91.52% F1-score. This shows that our hybrid deep learning strategy can accurately and efficiently detect furcation involvement in dental X-ray images. This automated method potentially helps doctors diagnose and treat periodontal disease early, increasing patient outcomes.

To ensure crop quality and productivity, almond damage detection requires efficient and reliable diagnostic techniques. This research presents an automated almond damage detection method using segmentation and deep learning-based classification. UNet is used to identify almonds in the input image properly. Deep learning architectures like VGG16, VGG19, Xception, InceptionV3, Densenet201, and NASNetMobile are used to categorize damage after segmentation. Using the cross-entropy loss function, Dice coefficient, Jaccard index, and area-under-the-curve (AUC), the performance of the proposed system is evaluated. For robust training, the segmentation models were trained on a large dataset of almond images, including damaged and undamaged examples. Evaluation metrics showed segmentation and classification accuracy and reliability. UNet+Xception leads with 98.23% accuracy. Because of its precision, deep learning can automate almond crop damage detection. With a 95.67% Dice coefficient and 92.45% Jaccard index, UNet can accurately identify damage. Precision agriculture is using more AI, enabling crop health monitoring and management research. In order to improve almond yield management and sustainability, agricultural methods must employ advanced machine-learning algorithms.

This study uses a hybrid deep learning technique to classify asphalt, pavement, and unpaved roads. In real-world circumstances, image data noise can damage image categorization algorithms. This issue can be addressed by a deep neural network (DNN)-based classification system that uses advanced denoising algorithms to improve input images before categorization. We start by denoising noisy native images with autoencoder (AE) approaches. We use two autoencoders: Denoising Autoencoder(DAE) and Convolutional Denoising Autoencoder(CDAE). Proper categorization requires models that filter noise and increase visual clarity. The CDAE employs convolutional layers to maintain spatial hierarchies and local characteristics during denoising, whereas the DAE involves encoding and decoding to rebuild images. The rebuilt images are classified using a CNN after denoising. The CNN is a preferred DNN architecture for this job since it can gather and represent complex visual input. CNN identifies classification-boosting features using noise-free image training. Experiments show this hybrid model works. With 97.92% classification accuracy, the CDAE-CNN architecture could recognize road surface types and conditions under noisy environments. This performance proves the hybrid approach's durability despite training on noise-corrupted images. It improves image classification in noisy data. Denoising algorithms improve deep learning classifier accuracy and make them more relevant in real-world applications with low image quality. These hybrid DAE-CNN/CDAE-CNN models minimize noise and properly categorize road surfaces.

The abnormalities arises withinh the brain cells results in brain tumors (BT) and this emerges as a life-threatening diseases that results in increased death rate day by day. However, accurate segmentation is required to detect BT because large spatial and structural variability between BT makes automatic segmentation difficult. In order to improve the detection rate, the proposed method named the Adaptive Neuro-fuzzy Inference System-Fusion-Deep Belief Network (ANFIS-F-DBN) model is developed in this research. At first, the brain image acquired form the specified dataset is pre-processed using the Gaussian filter and then, the extraction of Regions of interest (ROI) is done. Then, the Thresholding Transformation is used for the image enhancement process. After that, the Deep Fuzzy Clustering (DFC) is employed to segment the brain tumor area and the image augmentation process is done by different steps, like sharpening, translation, zooming and padding. In addition, the extraction of features namely, statistical features and Entropy with Local Directional Pattern Variance (LDPv) are done in the feature extraction phase. At last, the BT detection is executed based on hybrid deep learning (DL) namely, ANFIS-F-DBN. Moreover, the metrics like accuracy, sensitivity and specificity are used to analyze the performance of the devised scheme and obtained a value of 90.00%, 90.60% and 91.90%, respectively.

Safety helmets must be worn at construction to avoid injuries. Manual inspection of safety helmets for every worker requires a lot of human effort which is a costlier and tedious task. Hence, visual inspection can save human effort and reduce the cost. It helps the manager to send safety reminders to the workers. Therefore, this paper describes a feature fusion method to detect safety helmets at construction sites without human intervention. Because of the complex sciences of a construction site, this paper proposed a vibrant deep-learning solution with a better accuracy rate. Here, safety helmets are identified with a feature fusion deep learning model. The features of images are extracted with a hand-crafted deep learning model and pre-trained models that are later fused into a matrix with the fusion technique. Finally, the classification was employed with a convolution neural network. The proposed model was compared with other state-of-the-art deep learning models, and it outperforms with 98.90% correct classification accuracy.

More than half of the population is directly dependent on agriculture and it also contributes to the GDP of many countries. Rice crop is one of the main yields in the field of agriculture and in many countries rice is considered the main food. But there are many rice leaf diseases that affect the rice crop very badly. Which will cause low-quality crops and also affects the growth of the crop and it's very difficult for the farmers to detect rice crop diseases at right time without the help of an expert. So automated system is a very helpful way to detect rice leaf disease. So that farmers can take precautions on time can save their crops from damage. In our proposed work a dataset of 5368 images which has been collected from the Kaggal and Plant Village and the SqueezeNet pre-trained model is applied for feature extraction and the neural network classifier used to classify the rice leaf disease. In which 96.5% accuracy was achieved on 10 cross-folds and98.3 % accuracy on 20 cross folds these results are compared with other classifiers like SVM, Random forest, KNN, Naive Bayes, and Adaboost. The neural network achieved the highest accuracy of the above algorithms.

In today's life, agriculture holds considerable importance in human life and the economy of a nation. Agriculture, including tomato farming, plays a vital role as one of the most extensively consumed vegetables worldwide. However, tomato crops are very prone to diseases, leading to reduced production and economic down in agricultural fields. To solve these issues, an effective method is proposed named Skill-Honey Badger Optimisation Algorithm-enabled deep convolutional neural network (CNN) (SHBOA_DeepCNN) for detecting leaf disease in tomato plants. In this method, the input is primarily preprocessed by utilising Savitzky-Golay (SG) filtering. Then, segmentation is performed by utilising Dense-Res-Inception Net (DRINet), which is trained by using devised SHBOA. The proposed SHBOA is designed by incorporating the Skill Optimisation Algorithm (SOA) and Honey Badger Algorithm (HBA). Subsequently, image augmentation is performed on segmented images by using two augmentation techniques, namely, colour augmentation and position augmentation. At last, multiclass leaf disease detection is performed using DeepCNN, which is trained by devised SHBOA. The experimental analysis of the devised SHBOA_DeepCNN method showed a high accuracy of 91.91% and a true positive rate (TPR) of 90.24%. Moreover, it achieved a minimum false positive rate (FPR) of 7.38%. The code of the article is available at "".

Strawberries are one of the most demanding horticulture crops due to their flavor and nutritional value. Therefore, the demand for strawberries growing yearly and production of the same is very low. The prime factor of low production is leaf disease because the crop is susceptible to various types of leaf diseases. This causes huge financial losses every year to the farmers. Therefore, adequate action is needed to save strawberry crops from diseases and effectively manage their spread. Farmers were completely dependent on expert skilled personnel to identify crop diseases that may be the reason for the loss of crop yield. Therefore, this research proposed a novel deep-learning method to recognize plant leaf disease to optimize losses. However, with the invention of this method, it is easy to recognize different disease patterns within the plant leaf. The research proposed a feature fusion-based deep learning model named FFIR (Feature Fusion with Inception-RestNet) for recognizing leaf diseases. The features were extracted with a composition of the Inception-RestNet model. Later, the extracted features are fused in a fusion matrix using canonical correlation analysis (CCA). Afterward, the classification was performed with a Convolution Neural network. The proposed model outperformed with an accuracy of 99.34%.

As machine learning evolves, many individuals and businesses utilize numerous algorithms to evaluate massive datasets and create actionable insights that aid in predicting behavior. And this type of technology is increasingly employed in the medical industry to forecast the early stages of certain severe diseases, such as cervical cancer. There has been a significant amount of research conducted on cervical cancer in recent years. Studies have focused on various topics such as risk factors for cervical cancer, early detection and screening, and the effectiveness of different treatment options. In this study, we conduct an in-depth comparison of the various machine learning methods, discussing their relative merits and shortcomings in terms of accuracy and overall performance. Staking which is an ensemble machine learning approach emerges as the best approach for cervical cancer classification.

This study offers a software determination approach to classify and categorize Groundnut leaf diseases automatically. Determining Groundnut leaf disease detection and classification using photo classification has recently garnered much attention. Accurate classification results can be obtained when effective feature extraction and learning methods are used. The handcrafted features used in this work are mainly created intuitively, in contrast to the deep feature's complicated interpretation requirements and significant training sample needs. A new feature fusion approach called multi-layer visual feature fusion (MLVSF) was created to better capture the diseases seen in Groundnut photos. The MLVSF model can improve the discriminatory strength of features used in Groundnut leaf image identification by combining deep and handcrafted elements from local binary pattern variants, bag-of-visual words, and convolutional neural networks(CNN). With its ability to improve CNN and achieve more precise classification, MLVSF outperforms other modern methods when evaluated on Groundnut image datasets and generates 98.9% classification accuracy.

In the world, breast cancer kills more women than any other disease. An accurate forecast model was created, and the highest risk was identified by a data mining-based classification effort that used many approaches. The development of automated breast cancer diagnostic tools has been enabled by advances in data mining and machine learning. It is possible to tell benign breast lesions from malignant ones during a breast cancer diagnosis. More than that, people who have had tumors surgically removed can be given prognoses that tell them when the cancer will return. Consequently, the issues surrounding categorization include these two concerns. There are a lot of data mining techniques used to categorize patients as either "cancerous" or "non-cancerous" in this field. Breast cancer diagnostic concerns are thus encapsulated in the domain of hotly debated classification issues. This work used deep learning to diagnose breast cancer via federated learning while protecting user and hospital data. This study uses YOLO and RestNet-50 DL models. YOLO trains RestNet using client data. The server receives client-trained models for global model integration. Integration gives the client access to the global model for performance evaluation. The suggested model outperforms another state-of-the-art deep learning model in the identical scenario with 98.73% accuracy when all clients receive data equally. The federated research environment has hyper-performance elements like many clients and communication cycles.

Violence against humans in society is one major issue and cases are increasing rapidly. This creates an imbalance in society. Violence takes place suddenly in lonely places and it is difficult to handle it as there is a lack of information exchange. Although, surveillance cameras are installed at various places and the videos from these cameras can be utilized to detect violence. Several centralized deep learning(DL) and machine learning(ML) algorithms are used to address the problem, however, these methods are not suitable for protecting sensitive information. This study presented a deep learning method that uses federated learning to identify violent behaviors in CCTV video while protecting the privacy of individual users. This research uses the power of two deep learning models YOLO and RestNet-50. RestNe is trained with the data fetched with YOLO from CCTV footage available on the client side. Afterward, the trained models from clients are transferred to the server to integrate as a global model. After integration, the global model will be shared with the client for performance evaluation. It has been found that the suggested model beats another state-of-the-art deep learning model in the same situation, with an accuracy rate of 98.73%, when the data is distributed uniformly to all clients in the scenario. In research, the federated environment is set up with various hyper-performance parameters such as varying clients and communication rounds.

Fruit ripeness is a crucial factor in agriculture as it dictates the quality of the fruit. Manually determining the maturity of fruit has various drawbacks, including the need for a lot of labor, time, and potential for inconsistency. One of the most important economic sectors in the world is agriculture. Still, there are instances when the manual method of judging fruit maturity is utilized. Fruit ripeness may be automatically classified with the emerging technologies of artificial intelligence. A hybrid deep-learning approach is suggested in this study for fruit ripeness identification. In order to determine when bananas are ripe, this study uses deep learning models called Xception and Inception that have already been trained. First, the input photographs were processed using both models to extract primary characteristics. Then, the pictures were categorized according to the degree of ripeness of the bananas. Incorporating Co-Lab and other deep learning software, the models were put into action. The Xception pre-trained network achieved a classification accuracy of 98.99%, significantly outperforming state-of-the- art models in a performance evaluation.

A systemic inflammatory illness affecting the musculoskeletal system, rheumatoid arthritis (RA) is characterized by chronic, systemic symptoms. Injuries to joints and muscles from wear and tear are expected outcomes of rheumatoid arthritis (RA)., a degenerative illness that worsens with time. Arthritis rheumatoid generally degenerates muscles and joints while damaging bone and cartilage. To categorize medical conditions according to RA, this study employs machine learning approaches. Real-time RA data from the Sakthi Rheumatology Center, which keeps track of one thousand patient profiles has been used in this study. Multiple numerical features of the dataset are used for the classification task RA. The machine learning approaches used in this experiment were support vector machine (SVM), logistic regression, SDG, kNN, and ada-boosting. Ada-boost outperforms other classifiers regarding prediction accuracy rate and generating 98.7% classification accuracy. To classify the data, 20-fold cross-validation is used, and its efficacy is evaluated with metrics like precision, recall, sensitivity, and accuracy. The values of these measures were compared to a range of classifiers and baseline techniques.