arXiv subjects

Sebastiano Stramaglia

Publications and source records attributed to Sebastiano Stramaglia.

67 records · Page 4Linked to original sources

arXiv2007

Identification of network modules by optimization of ratio association

We introduce a novel method for identifying the modular structures of a network based on the maximization of an objective function: the ratio association. This cost function arises when the communities detection problem is described in the probabilistic autoencoder frame. An analogy with kernel k-means methods allows to develop an efficient optimization algorithm, based on the deterministic annealing scheme. The performance of the proposed method is shown on a real data set and on simulated networks.

cond-mat.dis-nn↗

arXiv2006

Semi-supervised learning by search of optimal target vector

We introduce a semi-supervised learning estimator which tends to the first kernel principal component as the number of labelled points vanishes. Our approach is based on the notion of optimal target vector, which is defined as follows. Given an input data-set of ${\bf x}$ values, the optimal target vector $\mathbf{y}$ is such that treating it as the target and using kernel ridge regression to model the dependency of $y$ on ${\bf x}$, the training error achieves its minimum value. For an unlabeled data set, the first kernel principal component is the optimal vector. In the case one is given a partially labeled data set, still one may look for the optimal target vector minimizing the training error. We use this new estimator in two directions. As a substitute of kernel principal component analysis, in the case one has some labeled data, to produce dimensionality reduction. Second, to develop a semi-supervised regression and classification algorithm for transductive inference. We show application of the proposed method in both directions.

cond-mat.dis-nn↗

arXiv2006

Causal interactions and delays in a neuronal ensemble

We analyze a neural system which mimics a sensorial cortex, with different input characteristics, in presence of transmission delays. We propose a new measure to characterize collective behavior, based on the nonlinear extension of the concept of Granger causality, and an interpretation is given of the variation of the percentage of the causally relevant interactions with transmission delays.

cond-mat.dis-nn↗

arXiv2006

Nonlinear parametric model for Granger causality of time series

We generalize a previously proposed approach for nonlinear Granger causality of time series, based on radial basis function. The proposed model is not constrained to be additive in variables from the two time series and can approximate any function of these variables, still being suitable to evaluate causality. Usefulness of this measure of causality is shown in a physiological example and in the study of the feed-back loop in a model of excitatory and inhibitory neurons.

cond-mat.dis-nn↗

arXiv2005

Kernel method for clustering based on optimal target vector

We introduce the notion of optimal target vector, and describe how it creates a link between supervised and unsupervised learning. We exploit this notion to construct Ising models, for dichotomic clustering, whose couplings are (i) both ferro- and anti-ferromagnetic (ii) depending on the whole data-set and not only on pairs of samples. The effectiveness of the method is shown in the case of the well known iris data-set and in benchmarks of gene expression levels.

cond-mat.dis-nn↗

arXiv2005

An invariance property of kernel based predictors

We consider kernel based learning methods for regression and analyze what happens to the risk minimizer when new variables, statistically independent of input and target variables, are added to the set of input variables; this problem arises, for example, in the detection of causality relations between two time series. We find that the risk minimizer remains unchanged if we constrain the risk minimization to hypothesis spaces induced by suitable kernel functions. We show that not all kernel induced hypothesis spaces enjoy this property. We present sufficient conditions ensuring that the risk minimizer does not change, and show that they hold for inhomogeneous polynomial and Gaussian RBF kernels. We also provide examples of kernel induced hypothesis spaces whose risk minimizer changes if independent variables are added as input.

cond-mat.dis-nn↗

arXiv2004

Extending Granger causality to nonlinear systems

We consider extension of Granger causality to nonlinear bivariate time series. In this frame, if the prediction error of the first time series is reduced by including measurements from the second time series, then the second time series is said to have a causal influence on the first one. Not all the nonlinear prediction schemes are suitable to evaluate causality, indeed not all of them allow to quantify how much the knowledge of the other time series counts to improve prediction error. We present a novel approach with bivariate time series modelled by a generalization of radial basis functions and show its application to a pair of unidirectionally coupled chaotic maps and to a physiological example.

physics.data-an↗

arXiv2001

Diluted neural networks with adapting and correlated synapses

We consider the dynamics of diluted neural networks with clipped and adapting synapses. Unlike previous studies, the learning rate is kept constant as the connectivity tends to infinity: the synapses evolve on a time scale intermediate between the quenched and annealing limits and all orders of synaptic correlations must be taken into account. The dynamics is solved by mean-field theory, the order parameter for synapses being a function. We describe the effects, in the double dynamics, due to synaptic correlations.

cond-mat.dis-nn↗

arXiv2000

Phase Ordering in Chaotic Map Lattices with Additive Noise

We present some result about phase separation in coupled map lattices with additive noise. We show that additive noise acts as an ordering agent in this class of systems. In particular, in the weak coupling region, a suitable quantity of noise leads to complete ordering. Extrapolating our results at small coupling, we deduce that this phenomenon could take place also in the limit of zero coupling.

cond-mat.dis-nn↗

arXiv1999

Phase ordering in chaotic map lattices with conserved dynamics

Dynamical scaling in a two-dimensional lattice model of chaotic maps, in contact with a thermal bath, is numerically studied. The model here proposed is equivalent to a conserved Ising model with coupligs which fluctuate over the same time scale as spin moves. When couplings fluctuations and thermal fluctuations are both important, this model does not belong to the class of universality of a Langevin equation known as model B; the scaling exponents are continuously varying with the temperature and depend on the map used. The universal behavior of model B is recovered when thermal fluctuations are dominant.

cond-mat.stat-mech↗

arXiv1998

Persistence exponent in a superantiferromagnetic quenching

We measure the persistence exponent in a phase separating two-dimensional spin system with non-conserved dynamics quenched in a region with four coexisting stripe phases. The system is an Ising model with nearest neighbor, next-to-the-nearest neighbor and plaquette interactions. Due the particular nature of the ground states, the order parameter is defined in terms of blocks of spins. Our estimate of the persistence exponent, $θ=0.42$, differs from those of the two-dimensional Ising and four state Potts models. Our procedure allows the study of persistence properties also at finite temperature $T$: our results are compatible with the hypothesis that $θ$ does not depend on $T$ below the critical point.

cond-mat.stat-mech↗

arXiv1996

Polymerization in a Ferromagnetic Spin Model with Threshold

We propose a spin model with a new kind of ferromagnetic interaction, which may be called {\it ferromagnetic coupling with threshold}. In this model the contribute of a given spin to the total energy has only two possible values and depends on the number of parallel spins among its nearest and next to the nearest neighbors. By mapping the model onto the Ising version of the isotropic eight vertex model, we obtain some evidence of a low temperature phase made of alternate parallel pluses and minuses polymers.

cond-mat↗

arXiv1995

Universality of the Tearing Phase in Matrix Models

The spontaneous symmetry breaking associated to the tearing of a random surface, where large dynamical holes fill the surface, was recently analized obtaining a non-universal critical exponent on a border phase. Here the issue of universality is explained by an independent analysis. The one hole sector of the model is useful to manifest the origin of the (limited) non-universal behaviour, that is the existence of two inequivalent critical points.

hep-th↗