Federico Cecconi

Defining and identifying communities in networks

The investigation of community structures in networks is an important issue in many domains and disciplines. This problem is relevant for social tasks (objective analysis of relationships on the web), biological inquiries (functional studies in metabolic and protein networks), or technological problems (optimization of large infrastructures). Several types of algorithms exist for revealing the community structure in networks, but a general and quantitative definition of community is not implemented in the algorithms, leading to an intrinsic difficulty in the interpretation of the results without any additional nontopological information. In this article we deal with this problem by showing how quantitative definitions of community are implemented in practice in the existing algorithms. In this way the algorithms for the identification of the community structure become fully self-contained. Furthermore, we propose a local algorithm to detect communities which outperforms the existing algorithms with respect to computational cost, keeping the same level of reliability. The algorithm is tested on artificial and real-world graphs. In particular, we show how the algorithm applies to a network of scientific collaborations, which, for its size, cannot be attacked with the usual methods. This type of local algorithm could open the way to applications to large-scale technological and biological systems.

Econets: Neural networks that learn in an environment

Ecological networks are networks that learn in an environment. It is the environment, and not the researcher, that determines the conditions in which learning takes place such as which input patterns are seen, what the teaching input is, etc. Furthermore, input patterns at time N+1 are often a function of the output of the network at time N. Two hypotheses are explored with reference to ecological networks. One is that predicting the sensory consequences (input) for an organism of the organisms actions (output) on the environment is one of the basic tasks of this type of network—basic for constructing an environmental map or world model. The other is that learning to predict the sensory consequences of the organisms actions favourably predisposes the organism to learn to attain goals with those actions. Some data from simulations that support these two hypotheses are reported.

Comparison of voter and Glauber ordering dynamics on networks.

We study numerically the ordering process of two very simple dynamical models for a two-state variable on several topologies with increasing levels of heterogeneity in the degree distribution. We find that the zero-temperature Glauber dynamics for the Ising model may get trapped in sets of partially ordered metastable states even for finite system size, and this becomes more probable as the size increases. Voter dynamics instead always converges to full order on finite networks, even if this does not occur via coherent growth of domains. The time needed for order to be reached diverges with the system size. In both cases the ordering process is rather insensitive to the variation of the degree distribution from sharply peaked to scale free.

Cultural Change in Spatial Environments

A cellular automata model is used to study aspects of cultural change in spatial environments. Cultures are represented as bit strings in individual cells. Cultures may change because they become more similar to prevailing nearby cultures, are subject to intrinsic random changes, or expand to previously empty cells. Extending Axelrods (1997) results, the authors show that assimilation does not lead to a single homogeneous culture even if, unlike in Axelrods model, cultural assimilation may take place even between neighboring cells with zero similarity; intrinsic changes decrease rather than increase the number of stable cultural regions; and expansion of a single culture in a previously unoccupied territory does not result in a single culture in the entire territory. Geographical features (such as mountains) that are an obstacle to contact between cells increase the number of different cultural regions.

Learning in the Active Mode

The paper distinguishes between two different modes of learning by neural networks. Traditional networks learn in the passive mode by incorporating in their internal structure the regularities present in the input and teaching input they passively receive from outside. Networks that live in a physical environment (ecological networks) can learn in the active mode by acting on the environment and learning to predict what changes in the environment or in their relation to the environment are caused by their actions. Being able to predict the consequences of ones own actions is useful when one wants to cause desired consequences with these actions. The paper contrasts learning to predict the consequences of ones actions with learning to predict environmental changes that are independent from the networks actions. It then discusses how perceptually ‘hidden’ properties of the environment such as the weight of objects are better learned in the active rather than in the passive mode and how learning in the active mode can be particularly useful in a social environment and in learning by imitating others. Learning in the active mode appears to be a crucial component of the human adaptive pattern and is tightly linked to another component of this pattern, i.e., the human tendency to modify the external environment rather than adapt to the environment as it is.

Social Aggregations in Evolving Neural Networks

Sociality is related to space because it can only develop inside spatial aggregations of individuals that can physically interact with each other. We present simulations of populations of simple organisms living together in the same environment. The simulations use genetic algorithms to model the evolution of neural networks behaving in the environment. Spatial aggregations emerge evolutionarily (a) as an indirect by-product of the spatial distribution of resources in the environment and of the actions of the organisms on these resources, (b) as an advantageous adaptation of living inside social groups that function as “information centers”, (c) as a pre-condition for learning from others.

PSO (Particle Swarm Optimization): One Method, Many Possible Applications

Particle Swarm Optimization (PSO ) is an optimization technique, deriving from the EO [5]: the main features are the natural inspiration and the possibility to implement PSO onto different levels. This chapter is divided in three section: (1) the PSO definitions and relationship with MAS (Multi Agent Systems) framework; (2) three applications of PSO methods; (3) some general conclusions and perspectives. We try to show that PSO has a marked multidisciplinary character since systems with swarm characteristics can be observed in a variety of domains: the main argument in favor to PSO is proper the multidisciplinary character. Besides, POS can resolve multiobjective otpimization problems in efficient way, because POS naturally incorporates some concepts from Pareto-Optimal framework.

Norm and Social Compliance: A Computational Study

The necessity to model the mental ingredients of norm compliance is a controversial issue within the study of norms. So far, the simulation-based study of norm emergence has shown a prevailing tendency to model norm conformity as a thoughtless behavior, emerging from social learning and imitation rather than from specific, norm-related mental representations. In this article, the opposite stance-namely, a view of norms as hybrid, two-faceted phenomena, including a behavioral/social and an internal/mental side-is taken. Such a view is aimed at accounting for the difference between norms, on one hand, and either behavioral regularities conventions on the other. After a brief presentation of a normative agent architecture, the preliminary results of agent-based simulations testing the impact of norm recognition and the role of normative beliefs in the emergence and stabilization of social norms are presented and discussed. We focused our attention on the effects which the use of a cognitive architecture namely a norm recognition module produces on the environment.

On the Interplay between Extortion and Punishment. An Agent Based Model of "Camorra"

This paper presents an Agent-Based Model aimed to reproduce the demographics, economic and employment variables of a Southern Italian region (Campania) where one specific variant of Extortion Racketeering Systems (Erss), camorra, is highly active and prosperous. Preliminary results of a set of simulations show the effects of varying levels of extortion and punishment on the rates of inactivity, employment, etc. of a population of agents endowed with social learning mechanisms

R&D Subsidization Effect and Network Centralization: Evidence from an Agent-Based Micro-Policy Simulation

This paper presents an agent-based micro-policy simulation model assessing public RD then, we provide a simulation experiment where the pattern of the total level of R&D activated by a fixed amount of public support is analysed as function of companies’ network topology. More specifically, the suggested simulation experiment shows that a larger “hubness†of the network is more likely accompanied with a decreasing median of the aggregated total R&D performance of the system. Since the aggregated firm idiosyncratic R&D (i.e., the part of total R&D independent of spillovers) is slightly increasing, we conclude that positive cross-firm spillover effects - in the presence of a given amount of support - have a sizeable impact within less centralized networks, where fewer hubs emerge. This may question the common wisdom suggesting that larger R&D externality effects should be more likely to arise when few central champions receive a support.