Acessibilidade / Reportar erro

Aprendizado não-supervisionado em redes neurais pulsadas de base radial: um estudo da capacidade de agrupamento para a classificação de pixels

Pulsed neural networks - networks that encode information in the timing of spikes - has been studied as a new and promising approach in the artificial neural networks paradigm, emergent from the cognitive science. One of these new models is the pulsed neural network with radial basis function, a network able to store information in the axonal propagation delay of neurons. A learning algorithm was successfully applied to this pulsed network, which was able to map a sequence of input pulses into a sequence of output pulses. More recently, a method based on the use of Gaussian receptive fields was proposed to encode constant data into a temporal sequence of spikes. This method allowed this network to deal with computational data. The learning process of this new network is not completely understood and deeper investigations are necessary in order to situate this model in the machine learning context and also to establish the network abilities and limitations. This work investigates this new classifier and presents a study of the network capability in solving the three-dimensional clustering task, particularly looking for establishing its application domains and horizons in the computer vision field.

Pulsed neural networks; unsupervised learning; clustering; computer vision


Sociedade Brasileira de Automática Secretaria da SBA, FEEC - Unicamp, BLOCO B - LE51, Av. Albert Einstein, 400, Cidade Universitária Zeferino Vaz, Distrito de Barão Geraldo, 13083-852 - Campinas - SP - Brasil, Tel.: (55 19) 3521 3824, Fax: (55 19) 3521 3866 - Campinas - SP - Brazil
E-mail: revista_sba@fee.unicamp.br