Sami Bourouis

A Real Time Adaptive Intrusion Detection Alert Classifier for High Speed Networks

With the emergence of High Speed Network (HSN), the manual intrusion alert detection become an extremely laborious and time-consuming task since it requires an experienced skilled staff in security fields and need a deep analysis. In addition, the batch model of alert management is no longer adequate given that labeling is a continuous time process since incoming intrusion alerts are often collected continuously in time. Furthermore, the static model is no longer appropriate due to the fluctuation nature of the number of alerts incurred by Internet traffic fluctuation nature. This paper proposes an efficient real time adaptive intrusion detection alert classifier dedicated for high speed network. Our classifier is based an online self-trained SVM algorithm with several learning strategies and execution modes. We evaluate our classifier against three different data-sets and the performance study shows an excellent results in term of accuracy and efficiency. The predictive local learning strategy presents a good tradeoff between accuracy and time processing. In addition, it does not involve a human intervention which make it an excellent solution that satisfy high speed network alert management challenges.

Image and video denoising by combining unsupervised bounded generalized gaussian mixture modeling and spatial information

In recent years, a great deal of effort has been expended on developing robust solutions for images quality degradation caused mainly by noise. In this paper, we explore this area of research and we propose a new unsupervised algorithm for both image and video denoising. Our solution is based on a flexible statistical mixture model driven by a finite mixtures of bounded generalized Gaussian distributions (BGGMD) which offers more flexibility in data modeling than the well known classical gaussian distributions which fail to fit the shape of heavy-tailed data produced by the presence of noise or outliers. The proposed framework takes into account also spatial information between neighboring pixels to be more robust and flexible, and able to provide smooth and accurate denoising results. For model’s parameters estimation, we investigate the unsupervised expectation-maximization (EM) algorithm. In order to evaluate the performance of the proposed model, we conducted a series of extensive experiments. Obtained results are more encouraging than those obtained using similar approaches. These results show the robustness and flexibility of the proposed method to adapt different shapes of observed data and bounded support data in the case of noisy images and videos.

A hierarchical Dirichlet process mixture of generalized Dirichlet distributions for feature selection

Display Omitted A statistical framework based on hierarchical Dirichlet processes and generalized Dirichlet distribution is developed.The framework simultaneously performs model parameters estimations as well as model complexity determination.The learning of the model is done via variational Bayes inference.The efficiency of the proposed algorithm is validated via challenging applications. This paper addresses the problem of identifying meaningful patterns and trends in data via clustering (i.e. automatically dividing a data set into meaningful homogenous sub-groups such that the data within the same sub-group are very similar, and data in different sub-groups are very different). The clustering framework that we propose is based on the generalized Dirichlet distribution, which is widely accepted as a flexible modeling approach, and a hierarchical Dirichlet process mixture prior. A main advantage of the adopted hierarchical Dirichlet process is that it provides a principled elegant nonparametric Bayesian approach to model selection by supposing that the number of mixture components can go to infinity. In addition to capturing the structure of the data, the combination of hierarchical Dirichlet process and generalized Dirichlet distribution is shown to offer a natural efficient solution to the feature selection problem when dealing with high-dimensional data. We develop two variational learning approaches (i.e. batch and incremental) for learning the parameters of the proposed model. The batch algorithm examines the entire data set at once while the incremental one learns the model one step at a time (i.e. update the models parameters each time new data are introduced). The utility of the proposed approach is demonstrated on real applications namely face detection, facial expression recognition, human gesture recognition, and off-line writer identification. The obtained results show clearly the merits of our statistical framework.

Web Service Intrusion Detection Using a Probabilistic Framework

In this paper, we propose an anomaly-based approach to detect intrusions attempts that may target web services. These intrusions (or attacks) are modeled as outliers (or noise) within a principled probabilistic framework. The proposed framework is based on finite Gaussian mixtures and allows the detection of both previously seen and unknown attacks against web services. The main idea of our framework is based on the consideration of malicious requests as outliers within our finite mixture model. Using this idea the intrusion detection problem is reduced to an adversarial classification problem. The merits of the proposed approach are shown using a data set containing both normal and intrusive requests, which were collected from a large real-life web service.

Multimodal Bone Cancer Detection Using Fuzzy Classification and Variational Model

Precise segmentation of bone cancer is an important step for several applications. However, the achievement of this task has proven problematic due to lack of contrast and the non homogeneous intensities in many modalities such as MRI and CT-scans. In this paper we investigate this line of research by introducing a new method for segmenting bone cancer. Our segmentation process involves different steps: a registration step of different image modalities, a fuzzy-possibilistic classification FPCM step and a final segmentation step based on a variational model. The registration and the FPCM algorithms are used to locate and to initialize accurately the deformable model that will evolve smoothly to delineate the expected tumor boundaries. Preliminary results show accurate and promising detection of the cancer region.

Visual Scene Reconstruction Using a Bayesian Learning Framework

In this paper, we focus on constructing new flexible and powerful parametric framework for visual data modeling and reconstruction. In particular, we propose a Bayesian density estimation method based upon mixtures of scaled Dirichlet distributions. The consideration of Bayesian learning is interesting in several respects. It allows simultaneous parameters estimation and model selection, it permits also taking uncertainty into account by introducing prior information about the parameters and it allows overcoming learning problems related to over- or under-fitting. In this work, three key issues related to the Bayesian mixture learning are addressed which are the choice of prior distributions, the estimation of the parameters, and the selection of the number of components. Moreover, a principled Metropolis-within-Gibbs sampler algorithm for scaled Dirichlet mixtures is developed. Finally, the proposed Bayesian framework is tested on a challenging real-life application namely visual scene reconstruction.

Spatially Constrained Mixture Model with Feature Selection for Image and Video Segmentation

In this paper we propose to improve image and video sequences segmentation through the integration of feature selection process into an unsupervised learning approach based on a finite mixture of bounded generalized Gaussian distributions (BGGMD). The proposed algorithm is less sensitive to over-segmentation, more flexible to data modeling and leading to better characterization and localization of object of interest in high-dimensional spaces since it is able to automatically reject irrelevant visual features. In order to determine adequately and automatically the number of regions in each image or frame, spatial information is incorporated as a prior information between neighboring pixels. Experimental results which are performed on a several real world images and videos demonstrate the effectiveness of the proposed framework with respect to other conventional Gaussian-based mixture models.

Hybrid Statistical Framework for Diabetic Retinopathy Detection

We present in this paper a novel hybrid statistical framework for retinal image classification and diabetic retinopathy detection. Our purpose here is to develop a probabilistic SVM-based kernel combined with a finite mixture of Scaled Dirichlet distributions. The developed method offers more flexibility in data modeling and classification since it takes advantage of both generative and discriminative models. Quantitative results obtained from a large dataset of real retinal images confirm the effectiveness of the proposed framework.

EP-Based Infinite Inverted Dirichlet Mixture Learning: Application to Image Spam Detection

We propose in this paper a new fully unsupervised model based on a Dirichlet process prior and the inverted Dirichlet distribution that allows the automatic inferring of clusters from data. The main idea is to let the number of mixture components increases as new vectors arrive. This allows answering the model selection problem in a elegant way since the resulting model can be viewed as an infinite inverted Dirichlet mixture. An expectation propagation (EP) inference methodology is developed to learn this model by obtaining a full posterior distribution on its parameters. We validate the model on a challenging application namely image spam filtering to show the merits of the framework.

Variational learning of hierarchical infinite generalized Dirichlet mixture models and applications

Data clustering is a fundamental unsupervised learning task in several domains such as data mining, computer vision, information retrieval, and pattern recognition. In this paper, we propose and analyze a new clustering approach based on both hierarchical Dirichlet processes and the generalized Dirichlet distribution, which leads to an interesting statistical framework for data analysis and modelling. Our approach can be viewed as a hierarchical extension of the infinite generalized Dirichlet mixture model previously proposed in Bouguila and Ziou (IEEE Trans Neural Netw 21(1):107–122, 2010). The proposed clustering approach tackles the problem of modelling grouped data where observations are organized into groups that we allow to remain statistically linked by sharing mixture components. The resulting clustering model is learned using a principled variational Bayes inference-based algorithm that we have developed. Extensive experiments and simulations, based on two challenging applications namely images categorization and web service intrusion detection, demonstrate our model usefulness and merits.