Giuliano Grossi

A two-level social mobility model for trace generation

We propose a synthetic trace generator that, although based on a simple mobility model, generates traces with statistical properties (like inter-contact time and contact duration) resembling those of well-known real traces. The proposed model is based on a waypoint scheme, with some modifications: the introduction of two categories of nodes, steady and nomadic, with different mobility rates; the grouping of nodes in communities sharing the same location preferences, and the definition of a micro-mobility model inside each location. The statistical properties of the traces generated with this model are compared to those obtained from publicly available real traces.

ECG compression retaining the best natural basis k-coefficients via sparse decomposition

Abstract A novel and efficient signal compression algorithm aimed at finding the sparsest representation of electrocardiogram (ECG) signals is presented and analyzed. The idea behind the method relies on basis elements drawn from the initial transitory of a signal itself, and the sparsity promotion process applied to its subsequent blocks grabbed by a sliding window. The saved coefficients rescaled in a convenient range, quantized and compressed by a lossless entropy-based algorithm. Experiments on signals extracted from the MIT-BIH Arrhythmia database show that the method achieves in most of the cases very high performance.

An approximation algorithm for the maximum cut problem and its experimental analysis

Abstract An approximation algorithm for the maximum cut problem is designed and analyzed; its performance is experimentally compared with that of a neural algorithm and that of Goemans and Williamsons algorithm. Although the guaranteed quality of our algorithm in the worst-case analysis is poor, we give experimental evidence that its average behavior is better than that of Goemans and Williamsons algorithm.

A Genetic Model and the Hopfield Networks

In this paper a genetic model is presented and the dynamics in the thermodynamic limit is derived. Analogies and differences with neural networks are discussed and attractors of the genetic model are characterized as equilibria points of Hopfields networks. The neural network and the genetic system are experimentally compared as approximate algorithms for the MAX-CUT problem.

A fixed-point iterative schema for error minimization in k-sparse decomposition

Analogously to the well known greedy strategy called Orthogonal Matching Pursuit (OMP), we present a new algorithm to solve the sparse approximation problem over redundant dictionaries where the input signal is restricted to be a linear combination of k atoms or fewer from a fixed dictionary. The basic strategy of our method rests on a family of nonlinear mappings which results to be contractive in a interval close to zero. By iterating contractions and projections the method is able to extract the most significant components also for noisy signal which subsumes an ideal underlying signal having sufficiently sparse representation. For reasonable error level, the fixed point solution of such a iterative schema provides a sparse approximation containing only the nonzero terms characterizing the unique sparsest representation of the ideal noiseless sparse signal. The heuristic method so derived has been applied both to synthetic and real data. The former was generated by combining exact signals drawn by usual Bernoulli-Gaussian model and Gaussian noise; the later is taken by electrocardiogram (ECG) signals with application to the dictionary learning problem. In both cases the proposed method outperforms OMP method both regarding sparse approximation error and computation time.

A neural algorithm for MAX-2SAT: performance analysis and circuit implementation

Abstract A neural algorithm for solving approximately the maximum 2-satisfiability problem is presented and its performance is analysed: the worst case relative error is 0.25 and the computation time is bounded by nm 4 , where n is the number of variables and m the number of clauses of a problem instance. Simulation experiments show a very good average case performance. We design a uniform family of circuits of small size and depth to implement the algorithm and present an efficient realization on field programmable gate arrays.

A Genetic Model: Analysis and Application to MAXSAT

In this paper, a genetic model based on the operations of recombination and mutation is studied and applied to combinatorial optimization problems. Results are: The equations of the deterministic dynamics in the thermodynamic limit (infinite populations) are derived and, for a sufficiently small mutation rate, the attractors are characterized; A general approximation algorithm for combinatorial optimization problems is designed. The algorithm is applied to the Max Ek-Sat problem, and the quality of the solution is analyzed. It is proved to be optimal for k3 with respect to the worst case analysis; for Max E3-Sat the average case performances are experimentally compared with other optimization techniques.

Robust Face Recognition Providing the Identity and Its Reliability Degree Combining Sparse Representation and Multiple Features

For decades, face recognition (FR) has attracted a lot of attention, and several systems have been successfully developed to solve this problem. However, the issue deserves further research effort so as to reduce the still existing gap between the computer and human ability in solving it. Among the others, one of the human skills concerns his ability in naturally conferring a “degree of reliability” to the face identification he carried out. We believe that providing a FR system with this feature would be of great help in real application contexts, making more flexible and treatable the identification process. In this spirit, we propose a completely automatic FR system robust to possible adverse illuminations and facial expression variations that provides together with the identity the corresponding degree of reliability. The method promotes sparse coding of multi-feature representations with LDA projections for dimensionality reduction, and uses a multistage classifier. The method has been evaluated in t...

2008 Special Issue: FPGA implementation of a stochastic neural network for monotonic pseudo-Boolean optimization

In this paper a FPGA implementation of a novel neural stochastic model for solving constrained NP-hard problems is proposed and developed. The model exploits pseudo-Boolean functions both to express the constraints and to define the cost function, interpreted as energy of a neural network. A wide variety of NP-hard problems falls in the class of problems that can be solved by this model, particularly those having a quadratic pseudo-Boolean penalty function. The proposed hardware implementation provides high computation speed by exploiting parallelism, as the neuron update and the constraint violation check can be performed in parallel over the whole network. The neural system has been tested on random and benchmark graphs, showing good performance with respect to the same heuristic for the same problems. Furthermore, the computational speed of the FPGA implementation has been measured and compared to software implementation. The developed architecture featured dramatically faster computation, with respect to the software implementation, even adopting a low-cost FPGA chip.

Sparse representation based classification for face recognition by k -limaps algorithm

In this paper, we present a new approach for face recognition that is robust against both poorly defined and poorly aligned training and testing data even with few training samples. Working in the conventional feature space yielded by the Fishers Linear Discriminant analysis, it uses a recent algorithm for sparse representation, namely k-LiMapS, as general classification criterion. Such a technique performs a local l0 pseudo-norm minimization by iterating suitable parametric nonlinear mappings. Thanks to its particular search strategy, it is very fast and able to discriminate among separated classes lying in the low-dimension Fisherspace. Experiments are carried out on the FRGC version 2.0 database showing good classification capability even when compared with the state-of-the-art l1 norm-based sparse representation classifier (SRC).