**Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?**

Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur GraphSearch.

Personne# Alireza Khadivi

Cette page est générée automatiquement et peut contenir des informations qui ne sont pas correctes, complètes, à jour ou pertinentes par rapport à votre recherche. Il en va de même pour toutes les autres pages de ce site. Veillez à vérifier les informations auprès des sources officielles de l'EPFL.

Unités associées

Chargement

Cours enseignés par cette personne

Chargement

Domaines de recherche associés

Chargement

Publications associées

Chargement

Personnes menant des recherches similaires

Chargement

Cours enseignés par cette personne

Aucun résultat

Domaines de recherche associés (8)

Réseau de capteurs sans fil

Un réseau de capteurs sans fil est un réseau ad hoc d'un grand nombre de nœuds, qui sont des micro-capteurs capables de recueillir et de transmettre des données d'une manière autonome. La position de

Feu

vignette|upright=1.7|Feu de bois.
Le feu est la production d'une flamme et la dégradation visible d'un corps par une réaction chimique exothermique d'oxydo-réduction appelée combustion.
De manière gé

Sensor node

A sensor node (also known as a mote in North America), consists of an individual node from a sensor network that is capable of performing a desired action such as gathering, processing or communicati

Publications associées (17)

Chargement

Chargement

Chargement

Unités associées (3)

Personnes menant des recherches similaires (105)

, , , , , , , , ,

Coupled neural networks, the Internet, World Wide Web, social networks and interacting biological networks are few examples of systems which consist of a large number of interacting dynamical units. Collective behavior of such systems is a consequence of the network structure as well as their dynamic properties. In this thesis, we study some aspects of structural properties of complex networks. We use the term "complex networks" informally to mean large networks without regular structure. Furthermore, we have exploited complex dynamical networks in order to solve three important distinct problems. The first problem that is tackled in two different ways is community detection in complex networks. This problem has recently attracted many researchers whereas many advantageous potential applications can be launched by knowing the community structure of the networks of interest. In our first work, we proposed to modify the network such that the performance of the existing community detection algorithms would be enhanced. More precisely, we have argued that a convenient weighting scheme improves the intrinsic drawbacks of the modularity measure, i.e. resolution limit and extreme degeneracies, and consequently; suitably weighting the network prior to applying a modularity optimization based community detection algorithm improves the overall performance. Then, we proposed a proper weighting scheme to verify our claim. The algorithm shows good results on a variety of networks. We have also extended our proposal for weighted networks . In our second work on community detection, we proposed a general appropriate graph-generative-model based likelihood function so that the actual community structure of the graph maximizes this function. Then, a dynamics based algorithm is proposed for optimizing this function. Indeed, we map a network into a dynamical network to disclose its community structure. This algorithm is very fast and accurate. In addition, it delivers the overlapping communities as well as the hierarchical structure. The second problem that is addressed in this thesis is finding abnormalities of neural synchronization patterns in different regions of the cortex for Alzheimer's disease patients. In fact, we study the interdependencies within the time series of the sources of electroencephalography (EEG) and give the relevant specific maps for patients suffering from Alzheimer's disease. In order to find the sources of EEG in the cortex, the LAURA inverse solution method is exploited. The synchronization within multivariate time series of EEG sources is estimated to capture the collective coherent spatiotemporal activity of neuronal populations. The results shows that the main effect of Alzheimer's disease progression type was clearly visible in the left hemisphere and mainly limited to the higher EEG frequencies. Finally, the last problem that is studied is fire forest surveillance using wireless sensor networks. We have proposed two algorithms to address this problem. The first algorithm is actually a fire detection method which spreads an alarm across the network in the case of a fire outbreak in the field. The second algorithm localizes the fire by finding a circle which surrounds the fire. In fact, the first algorithm declares whether there is a fire in the field and the second algorithm reports how big and where the fire is. For fire localization, the wireless sensor network uses a distributed average consensus algorithm to find the parameters of the circle. Indeed, the wireless sensor network works as a dynamical network which aims to do some arithmetic calculations in a non-centralized fashion. In order to verify our methods, we simulated them be Castalia, wireless sensor networks specialized software, while the required simulation input data, i.e. temperature of the environment, were provided by FARSITE, i.e. fire simulation software. The combination of these two software packages allowed us to simulate our methods in realistic conditions. The results reveal that the fire detection method is pretty fast while the fire localization method is accurate.

Cristian Carmeli, Alireza Khadivi, Reto Meuli

Alzheimer's disease (AD) disrupts functional connectivity in distributed cortical networks. We analyzed changes in the S-estimator, a measure of multivariate intraregional synchronization, in electroencephalogram (EEG) source space in 15 mild AD patients versus 15 age-matched controls to evaluate its potential as a marker of AD progression. All participants underwent 2 clinical evaluations and 2 EEG recording sessions on diagnosis and after a year. The main effect of AD was hyposynchronization in the medial temporal and frontal regions and relative hypersynchronization in posterior cingulate, precuneus, cuneus, and parietotemporal cortices. However, the S-estimator did not change over time in either group. This result motivated an analysis of rapidly progressing AD versus slow-progressing patients. Rapidly progressing AD patients showed a significant reduction in synchronization with time, manifest in left frontotemporal cortex. Thus, the evolution of source EEG synchronization over time is correlated with the rate of disease progression and should be considered as a cost-effective AD biomarker. (c) 2013 Elsevier Inc. All rights reserved.

, ,

Mechanisms on automatic discovery of macro actions or skills in reinforcement learning methods are mainly focused on subgoal discovery methods. Among the proposed algorithms, those based on graph centrality measures demonstrate a high performance gain. In this paper, we propose a new graph theoretic approach for automatically identifying and evaluating subgoals. Moreover, we propose a method for providing some useful prior knowledge for corresponding policy of developed skills based on two graph centrality measures, namely node connection graph stability and co-betweenness centrality. Investigating some benchmark problems, we show that the proposed approach improves the learning performance of the agent significantly.

2012