In this paper we propose a biologically inspired mathematical model to simulate the personalized interactions of users with cultural heritage objects. The main idea is to measure the interests of a spectator w.r.t. an artwork by means of a model able to describe the behaviour dynamics. In this approach, the user is assimilated to a computational neuron, and its interests are deduced by counting potential spike trains, generated by external currents. The key idea of this paper consists in comparing a strengthened validation approach for neural networks based on classification with our novel proposal based on clustering; indeed, clustering allows to discover natural groups in the data, which are used to verify the neuronal response and to tune the computational model. Preliminary experimental results, based on a phantom database and obtained from a real world scenario, are shown. They underline the accuracy improvements achieved by the clustering-based approach in supporting the tuning of the model parameters. © Springer International Publishing Switzerland 2015.

Validation approaches for a biological model generation describing visitor behaviours in a cultural heritage scenario

Ponti, G.
2015

Abstract

In this paper we propose a biologically inspired mathematical model to simulate the personalized interactions of users with cultural heritage objects. The main idea is to measure the interests of a spectator w.r.t. an artwork by means of a model able to describe the behaviour dynamics. In this approach, the user is assimilated to a computational neuron, and its interests are deduced by counting potential spike trains, generated by external currents. The key idea of this paper consists in comparing a strengthened validation approach for neural networks based on classification with our novel proposal based on clustering; indeed, clustering allows to discover natural groups in the data, which are used to verify the neuronal response and to tune the computational model. Preliminary experimental results, based on a phantom database and obtained from a real world scenario, are shown. They underline the accuracy improvements achieved by the clustering-based approach in supporting the tuning of the model parameters. © Springer International Publishing Switzerland 2015.
Clustering;Data mining;Computational neural models;User profiling
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/20.500.12079/4110
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
social impact