Search

Feature selection is an active research area in data mining and machine learning, especially with the increase in the amount of numerical data. FS is a search strategy to find the best subset of features among a large number of subsets of features. Thus, FS is applied in most modern applications and in various domains, which requires the search for a powerful FS technique to process and classify high-dimensional data. In this paper, we propose a new technique for dimension reduction in feature selection. This approach is based on a recent metaheuristic called Archimedes’ Optimization Algorithm (AOA) to select an optimal subset of features to improve the classification accuracy. The idea of the AOA is based on the steps of Archimedes' principle in physics. It explains the behavior of the force exerted when an object is partially or fully immersed in a fluid. AOA optimization maintains a balance between exploration and exploitation, keeping a population of solutions and studying a large area to find the best overall solution. In this study, AOA is exploited as a search technique to find an optimal feature subset that reduces the number of features to maximize classification accuracy. The K-nearest neighbor (K-NN) classifier was used to evaluate the classification performance of selected feature subsets. To demonstrate the superiority of the proposed method, 16 benchmark datasets from the UCI repository are used and also compared by well-known and recently introduced meta-heuristics in this context, such as: sine-cosine algorithm (SCA), whale optimization algorithm (WOA), butterfly optimization algorithm (BAO), and butterfly flame optimization algorithm (MFO). The results prove the effectiveness of the proposed algorithm over the other algorithms based on several performance measures used in this paper.

Abstract Vehicle identification systems are vital components that enable many aspects of contemporary life, such as safety, trade, transit, and law enforcement. They improve community and individual well-being by increasing vehicle management, security, and transparency. These tasks entail locating and extracting license plates from images or video frames using computer vision and machine learning techniques, followed by recognizing the letters or digits on the plates. This paper proposes a new license plate detection and recognition method based on the deep learning YOLO v8 method, image processing techniques, and the OCR technique for text recognition. For this, the first step was the dataset creation, when gathering 270 images from the internet. Afterward, CVAT (Computer Vision Annotation Tool) was used to annotate the dataset, which is an open-source software platform made to make computer vision tasks easier to annotate and label images and videos. Subsequently, the newly released Yolo version, the Yolo v8, has been employed to detect the number plate area in the input image. Subsequently, after extracting the plate the k-means clustering algorithm, the thresholding techniques, and the opening morphological operation were used to enhance the image and make the characters in the license plate clearer before using OCR. The next step in this process is using the OCR technique to extract the characters. Eventually, a text file containing only the character reflecting the vehicle's country is generated. To ameliorate the efficiency of the proposed approach, several metrics were employed, namely precision, recall, F1-Score, and CLA. In addition, a comparison of the proposed method with existing techniques in the literature has been given. The suggested method obtained convincing results in both detection as well as recognition by obtaining an accuracy of 99% in detection and 98% in character recognition.

Data clustering is a pivotal aspect of data mining, attracting significant attention in the exploration of cluster analysis through optimization algorithms. However, challenges arise in the application of optimization-based techniques, stemming from the non-linear nature of the objective function and the complexities inherent in the search space. Clustering, serving as an unsupervised data analysis technique, is employed to discern homogeneous groups of objects by evaluating the values of their attributes. To overcome the above-mentioned drawbacks, we have proposed in this article a new metaheuristic based on a mathematical model called Sinh Cosh Optimizer (SCHO). This optimizer is based on the geometric functions Sinh and Cosh and consists of four key stages: the exploitation and exploration phases, the limited search strategy and the implementation of a switching mechanism. To evaluate the effectiveness of the suggested approach, we conduct a comparative analysis against four widely acknowledged metaheuristic methods found in the literature. The assessment is performed using ten standard datasets sourced from the UCI Machine Learning Repository. These artificial intelligence algorithms are renowned for their promising capabilities in addressing machine learning problems, particularly those related to data clustering. Comprehensive computational experiments and result analysis demonstrate that the proposed algorithm consistently outperforms the comparison methods while exhibiting remarkable stability. Specifically, the experimental results show that the algorithm achieves superior F-score performance on 9 out of 10 datasets and outperforms in accuracy on 8 out of 10 datasets.

In this work, we propose a novel approach to enhancing the security of passwords before storing them in databases. Our method utilizes Braille transformation to encrypt the password after generating the corresponding hash. The hash is divided into multiple blocks, each representing a character treated as a transformation unit. Each character is then associated with its corresponding Braille code, which consists of 6 digits. To further enhance security, we randomly replace each occurrence of “0” in the generated string with one of the digits 7, 8, or 9. The final string, six times larger than the original hash, is then stored in the database. To evaluate our approach, we conducted several experiments and comparisons. The results demonstrate that Braille transformation is resistant to brute-force attacks, statistical attacks, and differential attacks. These results were justified using various evaluation criteria, such as execution time and memory space occupied. Braille transformation is susceptible to any modification made to the hash or the generated string, further reinforcing its security. Our Braille-based approach offers an effective solution to strengthen the security of database passwords. It provides advantages in terms of protection against different attacks and offers a robust evaluation based on relevant criteria.

In this study, we introduce a refined approach to encrypting 32-bit color images, leveraging the potential of four 1D chaotic maps – the Logistic map, Tent map, Chebyshev map, and Sine map. These chaotic maps intricately populate the four matrices within our encryption system, assigning exclusive integers ranging from 0 to 255. Our proposed methodology employs 16 × 16 matrices to represent the four channels (red, green, blue, and alpha) of a 32-bit color image, strategically utilizing specific grids for channel encryption. The top-left and bottom-right grids facilitate the encryption of the red and alpha channels, respectively, while the top-right and bottom-left grids are employed for encrypting the green and blue channels. The algorithm initiates by extracting decimal values from each pixel in the source image, mapping them to their corresponding positions in the matrices. A subsequent right circular shift operation on each pixel, determined by its row and column coordinates, is performed to prevent the encryption of areas with uniform color. To enhance security further, we employ the Four-square cipher method to encrypt the decimal values of the pixels. In the confusion stage, we apply the Arnold Cat Map transformation to strategically rearrange the position of all pixels, introducing an additional layer of complexity. Rigorous assessments using various security criteria were conducted to evaluate our algorithm's performance against common attacks, yielding consistently excellent results. Our method demonstrated superior outcomes, including a 25 % to 44 % increase in resistance to common attacks compared to existing methods.

Abstract In the present study, we aim to propose an effective and robust ensemble-learning approach with stacked generalization for image segmentation. Initially, the input images are processed for feature extraction and edge detection using the Gabor filter and the Canny algorithms, respectively; our main goal is to determine the most feature descriptions. Subsequently, we applied the stacking generalization technique, which is generally built with two main learning levels. The first level is composed of two algorithms that give good results in the literature, namely: LightGBM (Light Gradient Boosting Machine) and SVM (support vector machine). The second level is the meta-model in which we use a predictor model that takes the base-level predictions to improve the accuracy of the final prediction. In the stacked generalization process, we use the Extreme Gradient Boosting (XGBoost); it takes as input the sub-models’ outputs to better classify each pixel of the image to give the final prediction. Today, several research works exist in the literature using different machine learning algorithms; in fact, instead of trying to find a single efficient and optimal learner, ensemble-based techniques take the advantage of each basic model; they integrate their outputs to obtain a more consistent and reliable learner. The result obtained from the models of individuals and our proposed approach is compared using a set of evaluation measures for image quality such as IoU, DSC, CC, SSIM, SAM, and UQI. The evaluation and a comparison of the results obtained showed more consistent predictions for the proposed model. Thus, we have made a comparison with some recent deep learning-based unsupervised segmentation methods. The evaluation and a comparison of the results obtained showed more coherent predictions for our stacked generalization in terms of precision, robustness, and consistency.

With the growing need for secure multimedia data transmission, image encryption has become an important research area. Traditional encryption algorithms like RSA are not well-suited for this purpose, leading researchers to explore new approaches such as chaotic maps. The present study introduces a new image encryption algorithm that utilizes an improved Rossler system as a keystream generator. The improved Rossler system is an enhanced version of the original Rossler system, which has been optimized for better chaotic behavior and improved security. For the confusion part, we combine DNA encoding techniques with Baker maps to ensure high levels of security. Various performance metrics, including NPCR, UACI, correlation coefficient, histogram analysis, and key sensitivity analysis, were used to evaluate the proposed scheme. The results showed that the proposed method surpassed several existing image encryption methods in terms of both security and efficiency.

This paper proposes a new and efficient color image encryption algorithm based on multiple chaotic maps (Logistic map, Sine map, and Chebyshev map) and the intersecting planes method inside a cube. All maps were used to fill the three adjacent faces of the cube. These faces represent the three channels of the color image (red, green, and blue). The first phase of our method begins with extracting all the pixels from the original image and looking for the corresponding values of each pixel on the three faces of the cube. Then, we used a circular rotation operation based on the position of each pixel (row, column). This rotation prevents two identical pixels to have the same encrypted value. Afterward, we used the intersecting planes method with the corresponding face to encrypt the pixels. For the confusion part, we chose the 2D transformation Arnold Cat Map to shuffle all the pixels and change their positions according to the parameters calculated from all the pixels of the original image. We evaluated the performance of our algorithm against various differential and statistical attacks and based on multiple factors, such as histogram, entropy, correlation coefficient, UACI, NPCR, and PSNR. We compared the proposed method with several efficient methods in the field. The various results got by our method show its performance and reliability in terms of safety, precision, confidentiality, and robustness.