In this paper, we propose EDHP-FKP, an advanced framework for finger knuckle print (FKP) recognition that combines edge-enhanced preprocessing, deep feature extraction, and hierarchical classification. The key contributions of this work are: (1) an optimized preprocessing stage using Canny, Sobel, and Laplacian of Gaussian (LoG) edge detection to enhance discriminative FKP features; (2) multiple deep feature extractors (VGG-16, AlexNet, ResNet-50) and Gabor filter descriptors are investigated to capture both global and local texture information, and (3) classification using the Hierarchical Prototype (HP) classifier and Nearest Neighbor (NN) classifier to improve recognition accuracy and robustness. Experimental results demonstrate the effectiveness of the proposed framework, achieving competitive Rank-1 accuracy and low EER across multiple FKP modalities.

Autism spectrum disorder (ASD) comprises a group of neurodevelopmental conditions that affect brain development in early childhood and persist throughout life. Its primary symptoms include impaired communication and social interaction, restricted and repetitive behaviors, attention difficulties, and aggression. Early screening is crucial for controlling this condition. In recent years, with the evolution of artificial intelligence (AI), there has been huge interest in early diagnosis of ASD using multiple AI techniques based on various biological and behavioral markers. However, this diversity has led to confusion, as no study to date has provided a clear, comprehensive framework for data modalities that simplifies their utilization and demonstrates their potential for developing a novel ASD detection system. In this paper, we present a new study of recent advances in ASD detection using AI approaches. We detail the methods, technologies, and detection processes employed in current ASD detection. Furthermore, we propose a novel taxonomy, in which the data modalities are organized into four main categories, incorporating the most interesting markers. In addition, an empirical study is established, enabling us to focus on the challenges encountered and to propose solutions for the future in order to improve the quality and accuracy of diagnosis. To ensure the robustness and quality of the insights and perspectives arising from this study, a robust approach for article collection and selection is implemented to identify the most important contributions, and to vary the markers used for detection. Finally, the review highlights practical aspects of deploying data modalities suited for clinical and low-resource settings, and defines the essential future directions that should be focused on.

The rapid spread and daily usage of social networks have made them vulnerable to spammers. Therefore, detecting and eliminating spam and spammers has become more than necessary to reduce the risks that it poses to users’ security. In order to achieve this goal, it is crucial to determine the exact features that help identify and classify whether a user is spam or not. This paper proposes a wrapper-based method for selecting the most important features. It is based on combining two recent metaheuristic algorithms, the Henry Gas Solubility Optimization Algorithm (HGSO) and the Equilibrium Optimizer Algorithm (EO), with the goal of choosing a small and most influential subset of features that give good performance and help in the spammer profile detection process. For the purpose of showing the ability of the proposed method to achieve the desired goals, several comparisons are conducted on a modified Social Honeypot dataset. The first comparison is made between HGSOEO and the two algorithms (HGSO and EO) that were used to develop the proposed algorithm to prove the power of hybridization. The two next comparisons are made against some classical filter- and wrapper-based feature selection methods. The last comparison is carried out against some well-known metaheuristic algorithms for feature selection. Experiments and analysis of the results show that the proposed model is more accurate than the algorithms and methods that we compared it to.

Human beings are exposed to an infinite number of illnesses that can affect their health in multiple ways. The early detection of a disease can help doctors to manage it better by treating or at least controlling it. This research aims to help detect diseases from medical imaging. In this paper, we propose a new approach for disease detection from medical imaging based on VGG19 transfer learning with firefly optimization and Deep Rule Based (DRB) classifier. First, a set of features is extracted from the images using the pre-trained model VGG19. Next, we perform an optimization of the feature set by employing Firefly Algorithm (FA) to select the best subset of features that enhances the classifier performance and reduce the dataset complexity. Finally, DRB model is used to classify the samples. This method is tested on two image datasets, which concern two different diseases; Autism spectrum disorder (ASD) and Melanoma one of the most dangerous types of skin cancer. The first dataset is Autism_Image_Data, which comprises 2940 images of autistic and non-autistic children. The second dataset is a collection of 10605 images of benign and malignant skin lesions, called the “Melanoma Skin Cancer Dataset of 10000 Images”. To highlight the effectiveness of the system, it has been compared to several previous works. Experimental results have shown the best performance of the proposed approach compared to existing works. The proposed approach reached respectively 91.8 % and 83.33% of accuracy on the melanoma skin cancer dataset and the ASD dataset.

Feature selection (FS) is a crucial step in the data preprocessing phase of machine learning. FS is the process of selecting the relevent subset features from a large set of original features. In this study, we introduce an approach for feature selection in clustering problem relied on the Equilibrium Optimizer and the Gaussian Mixture Model. The main objective is to identify the subset that provides high-quality clustering, assessed using the Adjusted Rand Score (ARI) as the fitness function. The approach is tested on six available benchmarks with varying numbers of features and instances. The approach achieves superior results according to the ARI value and number of features. The comparison between the results with all features and the selected features highlights the effectiveness of the presented approach and the importance of feature selection. Further comparison with the Particle Swarm Optimization for feature selection confirms its superiority.

Biometric recognition systems are widely being used in several applications due to its distinctiveness and reliability. In recent years, hand-based person recognition has received much momentum due to its stability, feature richness, reliability and higher user acceptability. In this paper, we propose a multimodal hand biometric system based on Finger Knuckle Print (FKP) and Palmprint. In particular, the PCANet deep learning method is employed to extract distinctive features from each modality. Then, multiclass SVM is utilized to compute a matching score for each individual modality. Finally, score level fusion is performed to combine the matching scores via different rules such as, min, sum, max and multiplication. The performance of the proposed system is evaluated on the publicly available database known as PolyU. First, we conducted several experiments on single FKP and Palmprint traits. Next, score level fusion based multimodal experiments were performed. The proposed framework was able to achieve 0.00% of EER (equal error rates) and 100% of rank-1 performance. In addition, the proposed system based on PCANet with score level fusion of FKP and Palmprint outperformed existing multimodal methods.

Examining information concerning users of Location-Based Social Networks (LBSNs) and predicting a user's activity becomes very difficult to process.However recently, biclustering algorithms have shown their effective plan to uncover sub-matrices showing unique patterns.In this paper, we present a novel scheme based on biclustering algorithms with the Apriori algorithm to extract association rules from LBSNs data The idea focuses on generating a model that explains and analyzes the user's activities to get a lot of information about his behavior.Next, we select the sub-matrices using the BCX motif algorithm.After that, the Apriori algorithm is only applied to these sub-matrices, allowing for a meaningful extraction and useful rules for classifying user profiles The proposed technique is applied to a large real dataset called Gowalla.Thus, we have extracted rules such as (X→Y [support%, Confidence% and Lift]).Furthermore, we have presented a compression study of the results by biclustering the Apriori method and the Apriori algorithm.Additionally, the extracted rules provided relevant information for analyzing users' activities.we show that biclustering with Apriori method exhibits the best performance.In summary, the model of association rule briefly described user activities and it is considered a novel tool, which can be included in the offered state-of-the-art methods.

One of the most critical processes is feature selection, which eliminates features that may decrease classification performance and increase computational time. In this paper, we introduce and provide a comparison study between two algorithms, which are Equilibrium Optimizer (EO) and Henry Gas Solubility Optimization (HGSO) for Feature Selection (FS). The function objective of both algorithms are based on two main objectives, such as Error Rate (ER) and feature Reduction Rates (RR). In this comparative study, three classifiers (Naive Bayes NB, k-Nearest Neighbor KNN, and Random Forest RF) have been employed. The evaluation of the work was conducted on ten datasets, including Iris, Lung Cancer, Spambase, and Musk. The two algorithms show higher performances according to the accuracy and number of features, especially HGSOFS, which in turn shows its effectiveness and provides good results in the two tasks of FS when we compare it to the PSOFS (Particle Swarm Optimization for Feature Selection) and FAFS (Fire Fly for Feature Selection).

In this paper, we proposed a novel finger knuckle pattern (FKP) based personal authentication system using multilayer deep rule based (DRB) classifier. The presented approach is completely data-driven and fully automatic. However, the DRB classifier is generic and can be used in variety of classification or prediction problems. In particular, from the input finger knuckle, two kinds of features (i.e., Binarized Statistical Image Features and Gabor Filer bank) are extracted, which are then fed to fuzzy rules based DRB classifier to determine whether the user is genuine or impostor. Experimental results in the form of accuracy, error equal rate (EER) and receiver operating characteristic (ROC) curves demonstrate that presented DRB classifier is a powerful tool in FKP based biometric identification system. Experiments are reported using publicly available FKP PolyU database provided by University of Hong Kong. Experiments using this database show that the presented framework, in this study, can attain performance better than previously proposed methods. Moreover, score level fusion of all FKP modalities with BSIF + DRB yielded an equal error rate of 0.19% and an accuracy of 99.65%.

Diabetic Retinopathy (DR) is one of the mainly causes of visual loss worldwide. In fact, DR is leading source of impaired vision in people between 25 and 74 years old. DR exists in wide ranged and its detection is a challenging problem. The gradual deterioration of retina leads to DR with several types of lesions, including hemorrhages, exudates, micro aneurysms, etc. Early detection and diagnosis can prevent and save the vision of diabetic patients or at least the progression of DR can be slowed down. The manual diagnosis and analysis of fundus images to substantiate morphological changes in micro aneurysms, exudates, blood vessels, hemorrhages, and macula are usually time-consuming and monotonous task. It can be made easy and fast with the help of computer-aided system based on advanced machine learning techniques that can greatly help doctors and medical practitioners. Thus, the main focus of this paper is to provide a summary of the numerous methods designed for discovering hemorrhages, microaneurysms and exudates are discussed for eventual recognition of non-proliferative diabetic retinopathy. This survey will help the budding researchers, scientists, and practitioners in the field.

The large dimension of datasets influences on the classification performances and computational time. For that, feature selection is among process that avoid of those problems by selecting the relevance and eliminate the redundancy features.In this paper, we propose a new algorithm for feature selection FAFS (Firefly Algorithm for Feature Selection) which is based on the firefly algorithm. FAFS uses two objectives, which are Accuracy Rate and Reduction Rate. We propose a new formula to calculate the distance r and attractive A in Firefly algorithm.The experimental results show the capability of the proposed algorithm with three classifiers (KNN, NB, and LDA) and their outperformance against PSO-FS (Particle Swarm Optimization for feature selection).

The manipulation of a large number of features has become a critical problem in Intrusion Detection Systems(IDS). Therefore, Feature Selection (FS) is integrated to select the significant features, in order to avoid the computational complexity, and improve the classification performance. In this paper, we present a new multi-objective feature selection algorithm MOEDAFS (Multi-Objective Estimation of Distribution Algorithms (EDA) for Feature Selection). The MOEDAFS is based on EDA and Mutual Information (MI). EDA is used to explore the search space and MI is integrated as a probabilistic model to guide the search by modeling the redundancy and relevance relations between features. Therefore, we propose four probabilistic models for MOEDAFS. MOEDAFS selects the better feature subsets (non-dominated solutions) that have a better detection accuracy and smaller number of features. MOEDAFS uses two objective functions (minimizing classification Error Rate (ER) and minimizing the Number of Features(NF)). In order to demonstrate the performance of MOEDAFS, a comparative study is designed by internal and external comparison on NSL-KDD dataset. Internal comparison is performed between the four versions of MOEDAFS. External comparison is organized against some well-known deterministic, metaheuristic, and multi-objective feature selection algorithms that have a single and Multi-solution. Experimental results demonstrate that MOEDAFS outperforms recent algorithms.

Regarding to the huge number of connections and the large flow of data on the Internet, Intrusion Detection System (IDS) has a difficulty to detect attacks. Moreover, irrelevant and redundant features influence on the quality of IDS precisely on the detection rate and processing cost. Feature Selection (FS) is the important technique, which gives the issue for enhancing the performance of detection. There are different works have been proposed, but a map for understanding and constructing a state of the FS in IDS is still need more investigation. In this paper, we introduce a survey of feature selection algorithms for intrusion detection system. We describe the well-known approaches that have been proposed in FS for IDS. Furthermore, we provide a classification with a comparative study between different contribution according to their techniques and results. We identify a new taxonomy for future trends and existing challenges.

Image restoration is known as enhancement and recovery of images. Personal pictures captured by varied digital cameras will simply be manipulated by a range of dedicated image process algorithms .The aim of this paper is to implement a model of neural network with Fillingin technique to resolve the problem of image restoration, which is retrieving the original image degraded by invariant blur. The algorithm is proposed in this paper implements a general RBF neural network model with Probabilistic approach which differentiates the pixels of image according to their level of corruption and employees different ways to correct it. Less corrupted are corrected by remaining part of pixel using Filling-in technique, while others are corrected by using RBF neural network image restoration resulting in better signal to blur noise and better visual quality.

We want via this paper to show effectiveness of Discrete Simulation (DS) paradigm to improve and enhance Uncertain Decision Making. Our main goal is to allow modeling and simulation of Decision-making problems under uncertainty. Our simulation system implements two DS techniques: First order Simulation and second order simulation (characterized by its population size and number of replications). We propose in this paper to adapt simulation routines of these two techniques to the principles of decision tree representation formalism, to implement a forward inference mechanism that allows optimal decision strategy determination under uncertainty. Several comparative tests of the proposed simulation techniques with mathematical measurements have been done. Our system allows multiple types of users (Decision makers, analysts, students, ...) to create Decision models, access to existing models, run simulations, compare measures and share experiences. All created models and calculated measures are saved in XML format to permit their easy reuse.

An ontology is an explicit specification of a conceptualization, the term is often linked with the Semantic Web, ontologies are used like representations of knowledge, to annotate web resources and also to communication between systems. This make them very important, but unfortunately their construction is expensive, and because they are representations of knowledge, we thought of using the enormous amount of information available under textual format to automate the process of ontology building, and since we deal with texts, the NLP (Natural Language Processing) is considered as the base for the ontology construction from text. In this paper our goal is to propose an agent-based approach to build ontology from text, and implement a multi-agent system guided by this approach, which start from a set of textual resources to give us an ontology in OWL (Ontology Web Language), using the Formal Concept Analysis FCA and Relational Concept Analysis RCA to move from the syntactic level to the semantic level.

The aim of this paper is to implement an optimal neural network model to resolve the problem of colour image restoration which consists of retrieving original image degraded by invariant blur and corrupted by random white additive noise.We propose in this paper an algorithm which implements a general network of reduced neural networks model and adaptive regularization.The developed model is based on the original model of zhou and the modified model of Paik and katsaggelos.The adaptive regularization parameter is used in our case when degradation model contains an additive component (additive noise) in order to obtain a compromise between image sharpness and noise elimination.It is chosen using an iterative algorithm which calculates the best value that maximizes the PSNR of the restored image.Our model presents some improvements in terms of complexity and quality of restored images.It is shown by experiments that restored images obtained by the proposed model are better in terms of both numerical measurement and visual quality.