Hanane Azzag

Incremental Construction of Neighborhood Graphs Using the Ants Self-Assembly Behavior

This paper presents a novel head detection algorithm based on contour analysis on depth images. A sequence of depth-valued images is used as the system input using a 3D time-of-flight depth sensor. The background and foreground layers of the image are segmented using a straightforward depth thresholding technique. Moving regions are further processed in each frame, and contour analysis is performed on the depth maps to extract the curves of moving regions. Finally, ellipse fitting is performed to determine the objective head targets in the image. This information will be passed to the tracker in order to accomplish tracking the targets in the scene. Experimental results demonstrate the efficiency of the proposed method.In this paper we present a new incremental algorithm for building neighborhood graphs between data. It is inspired from the self-assembly behavior observed in real ants where ants progressively become attached to an existing support and then successively to other attached ants. Each artificial ant represents one data. The way ants move and build a graph depends on the similarity between the data. We have compared our results to those obtained by the relative neighborhood algorithm on several databases (either artificial or real), and we show that our method is competitive especially with respect to execution times.

SOM Clustering Using Spark-MapReduce

In this paper, we consider designing clustering algorithms that can be used in MapReduce using Spark platform, one of the most popular programming environment for processing large datasets. We focus on the practical and popular serial Self-organizing Map clustering algorithm (SOM). SOM is one of the famous unsupervised learning algorithms and its useful for cluster analysis of large quantities of data. We have designed two scalable implementations of SOM-MapReduce algorithm. We report the experiments and demonstrated the performance in terms of classification accuracy, rand, speedup using real and synthetic data with 100 millions of points, using different cores.

2013 Special Issue: Growing self-organizing trees for autonomous hierarchical clustering

This paper presents a new unsupervised learning method based on growing processes and autonomous self-assembly rules. This method, called Growing Self-organizing Trees (GSoT), can grow both network size and tree topology to represent the topological and hierarchical dataset organization, allowing a rapid and interactive visualization. Tree construction rules draw inspiration from elusive properties of biological organization to build hierarchical structures. Experiments conducted on real datasets demonstrate good GSoT performance and provide visual results that are generated during the training process.

A new bi-clustering approach using topological maps

In this paper, we propose a new bi-clustering algorithm based on self-organizing maps titled BiTM (Bi-clustering using Topological Map). BiTM provides a simultaneous clustering of rows and columns of the data matrix in order to increase the homogeneity of bi-clusters by respecting neighborhood relationship and using a single map. BiTM maps provide a new topological visualization of the bi-clusters. Experimental results and comparison studies show that BiTM improves the results in term of bi-clustering and visualization.

Self-Organizing map and tree topology for graph summarization

In this paper, we present a novel approach called SOM-tree to summarize a given graph into a smaller one by using a new decomposition of original graph. The proposed approach provides simultaneously a topological map and a tree topology of data using self-organizing maps. Unlike other clustering methods, the tree-structure aim to preserve the strengths of connections between graph vertices. The hierarchical nature of the summarization data structure is particularly attractive. Experiments evaluated by Accuracy and Normalized Mutual Information conducted on real data sets demonstrate the good performance of SOM-tree.

How to use ants for data stream clustering

We present in this paper a new bio-inspired algorithm that dynamically creates groups of data. This algorithm is based on the concept of artificial ants that move together in a complex manner with simple localization rules. Each ant represents one datum in the algorithm. The moves of ants aim at creating homogeneous groups of data that evolve together in a graph environment. We also suggest an extension to this algorithm to treat data streaming. The extended algorithm has been tested on real-world data. Our algorithms yielded competitive results as compared to K-means and Ascending Hierarchical Clustering (AHC), two well known methods.

Biclustering using Spark-MapReduce

Biclustering approaches are more complex compared to the traditional clustering particularly those requiring large dataset and Mapreduce platforms. We propose a new approach of biclustering based on popular self-organizing maps, which is one of the famous unsupervised learning algorithms. We have designed scalable implementations of the new topological biclustering algorithm using MapReduce with the Spark platform.

Clustering using chemical and colonial odors of real ants

We suggest in this paper a new automatic data clustering model based on the behavior of real ants. Drawing on a simulation of colonial odors and pheromone mechanisms, we set up complete dynamic graphs to solve the problem of data clustering. Using graph we will clarify the relationships between clusters of data.

Self-organizing trees for visualizing protein dataset

Clustering and visualizing multidimensional or structured data are important tasks for data analysis, especially in bioinformatics. Self-organizing models are often used to address both of these issues. In this paper we introduce a hierarchical and topological visualization technique called Self-organizing Trees (SoT) which is able to represent data in hierarchical and topological structure. The experiment is conducted on a real-world protein data set.

Growing Self-organizing Trees for knowledge discovery from data

In this paper, we propose a new unsupervised learning method based on growing neural gas and using self-assembly rules to build hierarchical structures. Our method named GSoT (Growing Self-organizing Trees) depicts data in topological and hierarchical organization. This makes GSoT a good tool for data clustering and knowledge discovery. Experiments conducted on real data sets demonstrate the good performance of GSoT.