Search

Traffic congestion has become a major problem in this rapidly growing world. Everyone operating a vehicle, as well as the traffic police in charge of managing the traffic, finds it difficult to become stuck in heavy traffic. For this a set, predetermined timing for traffic flow for each direction at the junction is utilized by traditional traffic light controllers. However, the concept of a fixed time traffic signal controller does not work well in places with uneven traffic. A dynamic traffic control system is therefore required, which regulates the traffic signals in accordance with the volume of traffic. This paper proposes a model that uses reinforcement learning (RL) along with deep neural networks (DNN) to manage discretions (signal status) for an environment with the help of Simulation of Urban MObility (SUMO). A simulation of real-world environment consisting a network of Four-way crossroad junction that contains 4 arriving lanes and 4 exiting lanes is used to train the agent. The main objective of this research study is to construct a model that can independently determine the best course of action and aims to provide better traffic management that will decrease the average waiting time, cause lower congestion, and provide a smooth flow of traffic. [ABSTRACT FROM AUTHOR]

The Consumer Price Index (CPI) is an indicator of inflation and is tracked by many government and economic agencies to make decisions of major importance. Its prediction is a valuable input into government policies such as taxation, and it greatly impacts the cost of borrowing money. The CPI has been traditionally predicted with statistical methods such as the Autoregressive Integrated Moving Average (ARIMA) model. In this paper, we forecast the Saudi Arabian Consumer Price Index with six machine learning (ML) methods, using the Orange 3 data mining and analytics tool, and based on the published historical January 2013 to November 2020 CPI data. We compare the performances of Decision Tree (Tree), k-Nearest Neighbors (kNN), Linear Regression (LR), Neural Networks (NN), Random Forest (RF), and Support Vector Machine (SVM), all applied to the 2013-2020 Saudi Arabian CPI dataset. Multiple experiments were conducted to vary the training and testing sets, optimize the machine learning parameters, and improve the MSE and R² metrics. The predicted CPI values of these ML methods were also compared to the 2021-2024 International Monetary Fund (IMF) CPI forecast and the actual 2021-2024 CPIs (post mortem). The results indicate that the multilayer perceptron neural network model outperforms the other ML models, is nearest to the actual CPI, and may be used to forecast the CPI for up to 3 years from the latest CPI data in the training dataset. The kNN model follows the neural network model in second place. The best fitting Excel trend line underperformed all ML methods in forecasting the Saudi Arabian CPI. [ABSTRACT FROM AUTHOR]

Multimedia data mining plays a crucial role in various fields, such as image and video analysis, natural language processing, and recommendation systems. Multimedia data refers to any form of data that involves multiple modes of communication, such as text, images, audio, and video. To effectively mine valuable insights from multimedia data, a new framework is proposed in this paper that employs a transformer-based intelligent deep neural network (DNN) model architecture. The framework includes an extensive data preprocessing step that involves obtaining multimedia data from internet searches and removing duplicates to ensure that each image is unique. The proposed transformer-based intelligent DNN model architecture processes the multimedia data in a hierarchical manner and utilizes shifted windows to achieve high accuracy in image classification task. The exploited dataset details are provided in the experimental evaluation section. Experimental results show that the proposed framework outperforms existing multimedia data mining methods in terms of accuracy and efficiency. This framework provides valuable insights that can be used in various applications, including content-based image retrieval, sentiment analysis, and automated captioning. [ABSTRACT FROM AUTHOR]

This study proposes an intelligent system for automated diseases detect and categorization in banana fruit, as well as an integrated grading system. To accomplish accurate illness identification and grading, the suggested system combines computer vision methods, machine learning algorithms, and deep learning models. The system extracts key information from banana fruit images using image processing techniques, which are subsequently input into a trained classification model. The categorization model uses cutting-edge algorithms to categorize the banana fruit into several illness groups. Furthermore, the sophisticated grading system evaluates the severity and quality of the diseased fruit based on a variety of characteristics such as size, color, and texture. The experimental findings reveal that the proposed method is successful, with high accuracy rate of 99.8% in illness diagnosis and accurate banana grading. This automated technology provides a time-efficient and cost-effective approach for disease control in banana plantations, allowing producers and agricultural stakeholders to make more informed decisions. [ABSTRACT FROM AUTHOR]

Real-time monitoring of output electrical parameters of the transmitted signals in a capacitive resistivity underground imaging system is necessary because these are significant in the calculation of underground resistivity, however, machine learning has not yet been applied in this application to improve the accuracy of measurement. This study aims to develop and select the best prediction models that can be implemented for a digital measuring unit suitable for capacitive resistivity underground imaging. Three deep neural network models namely Elman recurrent neural network (ERNN), long short-term memory (LSTM), and gated recurrent unit (GRU) were explored to build prediction models for the current and voltage of the transmitter circuit. The prediction models' performance was assessed using mean squared error (MSE), which is reduced to its absolute lowest value. The result shows that the best-trained models for current and voltage prediction are the ERNN models with configurations of 900-600-500 hidden neurons network with training MSE of 9.82 X 10-9 and the configured 1300-1000-900 hidden neurons with training MSE of 0.465, respectively. With the help of the prediction models, it would be possible to measure current and voltage output accurately, allowing simultaneous data acquisition while avoiding the need for a separate measuring device. [ABSTRACT FROM AUTHOR]

The exponential rise in spam and phishing emails presents a critical challenge to the privacy, security, and efficiency of users. This research introduces a deep learning model with enhanced performance over existing top-tier studies. The model's strength lies in its ability to precisely classify emails into three distinct categories: legitimate (ham), unsolicited (spam), and malicious (phishing). This research employs two sophisticated feature selection techniques to enhance classification accuracy: Principal Component Analysis (PCA) and Particle Swarm Optimization (PSO). These techniques are instrumental in identifying and extracting the most informative features from the data, which are critical for the task of categorizing emails effectively. Rigorous testing has elevated the PSO-enhanced model to a position of excellence, with an accuracy rate of 99.60%. This high degree of accuracy is a testament to the strength of deep learning in the arena of email filtering. The research confirms the value of feature selection in augmenting deep learning models, laying the groundwork for innovative defenses against email threats. The study's insights offer optimistic prospects for the advancement of more resilient email systems. Utilizing the substantial computational prowess of deep learning and the precision of feature selection techniques like PCA and PSO, the research charts a course for significantly reducing spam and phishing email incidents. As such, this research marks a significant stride in digital security, equipping stakeholders with a powerful asset in the ongoing effort to safeguard against cyber threats. [ABSTRACT FROM AUTHOR]

Deep learning has widespread use in various domains, including computer vision, audio processing, and natural language processing. The hyperparameters of deep learning algorithms have a significant impact on the performance of these algorithms. However, it can be challenging to calculate the hyperparameters of complicated machine learning models like deep neural networks due to the nature of the models. This research suggested a strategy for hyperparameter optimization utilizing the Long Short-Term Memory with Sparrow Search Algorithm (LSTM-SSA) model. The model that has been presented uses a deep neural network, which can recognize and classify instances of hate speech as either hate speech or neither. Experiments are conducted to validate the suggested technique in both straightforward and intricate network environments. The LSTM-SSA model is validated using a dataset consisting of hate speech, and an experimental investigation into the model's sensitivity, accuracy, and specificity is carried out. The outcomes of the experiments demonstrated that the suggested model might be improved upon, as it had an accuracy of 0.936. [ABSTRACT FROM AUTHOR]

Brain tumors were the most common kind of tumor in humans. Brain tumors can be detected from various imaging technologies. The proposed research work strives to improve the prediction accuracy of brain tumor detection and segmentation from MRI of human head scans by using a novel activation function E-Tanh. The role of activation functions is to perform computations and make decisions in artificial neural networks (ANN). We developed three ANN models for brain tumor detection by modifying the hidden layers. We have trained these ANN models using the E-Tanh activation function and evaluated their performance. This novel activation function achieved 98% prediction accuracy for the MRI brain tumor image detection neural network model, which was higher than the existing activation functions. We also have segmented brain tumors from the BraTS2020 dataset by using this activation function in U-Net-based architecture. We attained dice scores of 83%, 95%, and 85% for the whole, core, and enhancing tumors, which are significantly higher than the ReLU activation function. [ABSTRACT FROM AUTHOR]

Nutrient deficiency in Citrus reticulata (mandarin orange) plants is a critical agricultural issue that has far-reaching implications. Such deficiencies not only compromise the overall health and vigor of the mandarin orange trees but also render them more susceptible to diseases and pest infestations. Addressing this concern is of principal importance to ensure sustainable citrus cultivation and enhance crop yield. In this research, we propose a comprehensive approach to tackle the pervasive problem of nutrient deficiencies in Citrus Reticulata var. Fremont leaves. Our approach leverages cutting-edge techniques in image processing and machine learning to provide accurate and efficient detection of these deficiencies. The image data is divided into four classes: N Deficiency, P Deficiency, K Deficiency and Normal. The file sizes are compressed using a lossless compression method, resulting in an average file size reduction of 96.99%. The second stage involves applying segmentation processes to the images using the Canny and Sauvola methods. Third stage involves extracting colour and texture features from the images. The feature values will be used for the classification process in the next stage. The segmentation process employs two methods, namely Canny and Sauvola, which effectively separate the leaves from the background. The detection process is evident in the feature extraction phase, which utilizes two features: colour and texture. The fourth stage involves the classification process based on the segmentation results methods, performed separately using Artificial Neural Network (ANN) and Support Vector Machine (SVM) methods. These process results in four datasets: Canny-ANN, Canny-SVM, Sauvola-ANN, and Sauvola-SVM. The highest accuracy is achieved by the Sauvola-ANN method, with a value of 93.75%. [ABSTRACT FROM AUTHOR]