Publications by Martin Hasler | EPFL Graph Search

Distributed machine learning in networks by consensus

We propose an algorithm to learn from distributed data on a network of arbitrarily connected machines without exchange of the data-points. Parts of the dataset are processed locally at each machine, and then the consensus communication algorithm is employe ...

Elsevier2014

Dynamics of Stochastically Blinking Systems. Part I: Finite Time Properties

Martin Hasler, Igor Belykh

We consider dynamical systems whose parameters are switched within a discrete set of values at equal time intervals. Similar to the blinking of the eye, switching is fast and occurs stochastically and independently for different time intervals. There are t ...

Siam Publications2013

Dynamics of Stochastically Blinking Systems. Part II: Asymptotic Properties

Martin Hasler, Igor Belykh

We study stochastically blinking dynamical systems as in the companion paper (Part I). We analyze the asymptotic properties of the blinking system as time goes to infinity. The trajectories of the averaged and blinking system cannot stick together forever, ...

Siam Publications2013

Multistability of twisted states in non-locally coupled Kuramoto-type models

Martin Hasler

A ring of N identical phase oscillators with interactions between L-nearest neighbors is considered, where L ranges from 1 (local coupling) to N/2 (global coupling). The coupling function is a simple sinusoid, as in the Kuramoto model, but with a minus sig ...

American Institute of Physics2012

Mesoscale and clusters of synchrony in networks of bursting neurons

Martin Hasler, Igor Belykh

We study the role of network architecture in the formation of synchronous clusters in synaptically coupled networks of bursting neurons. We give a simple combinatorial algorithm that finds the largest synchronous clusters from the network topology. We demo ...

American Institute of Physics2011

Training Distributed Neural Networks by Consensus

Martin Hasler, Leonidas Georgopoulos

We address the problem of learning a classifier from distributed data over a number of arbitrarily connected machines without exchange of the datapoints. Our purpose is to train a neural network at each machine as if the entire dataset was locally availabl ...

2011

Network community-detection enhancement by proper weighting

Martin Hasler, Ali Ajdari Rad, Alireza Khadivi

In this article, we show how proper assignment of weights to the edges of a complex network can enhance the detection of communities and how it can circumvent the resolution limit and the extreme degeneracy problems associated with modularity. Our general ...

American Physical Society2011

A lower bound for algebraic connectivity based on the connection-graph-stability method

Martin Hasler, Mahdi Jalili, Ali Ajdari Rad

This paper introduces the connection-graph-stability method and uses it to establish a new lower bound on the algebraic connectivity of graphs (the second smallest eigenvalue of the Laplacian matrix of the graph) that is sharper than the previously publish ...

Elsevier2011

A Weighting Scheme for Enhancing Community Detection in Networks

Martin Hasler, Alireza Khadivi

Many algorithms have recently been proposed for finding communities in networks. By definition, a community is a subset of vertices with a high number of connections among the vertices, but only few connections with other vertices. The worst drawback of mo ...

IEEE2010

Information Processing in Complex Networks

Martin Hasler, Ali Ajdari Rad, Alireza Khadivi

2010

Automatic skill acquisition in Reinforcement Learning using connection graph stability centrality

Martin Hasler, Ali Ajdari Rad

Reinforcement Learning (RL) is an approach for training agent's behavior through trial-and-error interactions with a dynamic environment. An important problem of RL is that in large domains an enormous number of decisions are to be made. Hence, instead of ...

Ieee Service Center, 445 Hoes Lane, Po Box 1331, Piscataway, Nj 08855-1331 Usa2010

Community detection enhancement in networks using proper weighting and partial synchronization

Martin Hasler, Ali Ajdari Rad, Alireza Khadivi

A community in a network is a subset of vertices densely connected to each other, but less connected to the vertices outside. Many different approaches have been developed to find such structures in a given network, but the main drawback of most of the ava ...

Ieee Service Center, 445 Hoes Lane, Po Box 1331, Piscataway, Nj 08855-1331 Usa2010

Reservoir optimization in recurrent neural networks using properties of Kronecker product

Martin Hasler, Mahdi Jalili, Ali Ajdari Rad

Recurrent neural networks based on reservoir computing are increasingly being used in many applications. Optimization of the topological structure of the reservoir and the internal connection weights for a given task is one of the most important problems i ...

2010

Topography of EEG multivariate phase synchronization in early Alzheimer's disease

Martin Hasler, Reto Meuli, Mahdi Jalili

Alzheimer's disease (AD) is likely to disrupt the synchronization of the bioelectrical processes in the distributed cortical networks underlying cognition. We analyze the surface topography of the multivariate phase synchronization (MPS) of multichannel EE ...

2010

Fire Detection and Localization Using Wireless Sensor Networks

Martin Hasler, Alireza Khadivi

A Wireless Sensor Network (WSN) is a network of usually a large number of small sensor nodes that are wirelessly connected to each other in order to remotely monitor an environment or phenomena. Sensor nodes use the data aggregation method as an effective ...

Springer, 233 Spring Street, New York, Ny 10013, United States2010

Attenuated asymmetry of functional connectivity in schizophrenia: A high-resolution EEG study

Martin Hasler, Reto Meuli, Mahdi Jalili, Kim Do

The interhemispheric asymmetries that originate from connectivity-related structuring of the cortex are compromised in schizophrenia (SZ). Under the assumption that such abnormalities affect functional connectivity, we analyzed its correlate-EEG synchroniz ...

Wiley-Blackwell2010

Turbo-Like Structures for Chaos Encoding and Decoding

Martin Hasler, Slobodan Kozic, Francisco Javier Escribano Aparicio

In this paper, we explain how to build a turbo-like structure with binary inputs and chaotic outputs for efficient coding and decoding in additive white Gaussian noise (AWGN). We analyze the convergence of the decoding algorithm, the performance in the err ...

2009

Early consensus in complex networks under variable graph topology

Martin Hasler, Leonidas Georgopoulos

Analysis of the linear algorithm for consensus on complex networks shows the existence of two distinct phases, the asymptotic and the transient. The network reaches the asymptotic when the components corresponding to the spectral gap eigenvalue of the weig ...

IEEE2009

Multiple task learning in recurrent neural networks using layered reservoirs

Martin Hasler, Mahdi Jalili, Ali Ajdari Rad

In this paper, to learn multiple tasks sharing same inputs, a two-layer architecture for a reservoir based recurrent neural network is proposed. The inputs are fed into the general workspace layer where the weights are adapted to provide maximum informatio ...

2009

Low-Density Codes Based on Chaotic Systems for Low-Density Codes Based on Chaotic Systems for Simple Encoding

Martin Hasler, Slobodan Kozic

This paper proposes a new class of low-density generator-matrix codes (LDGM) based on chaotic dynamical systems. The codes are designed by controlling symbolic dynamics and using linear convolutional codes. Analyzing the complex structure of chaotic system ...

2009