Universidade Federal do Rio de Janeiro COPPE

Instituto Alberto Luiz Coimbra de Pós-Graduação e Pesquisa de Engenharia

Instituto de Matemática

 
Visualizar Meses
Visualizar Meses
Visualizar Flat
Visualizar Flat
Visualizar Semanas
Visualizar Semanas
Visualizar Dias
Visualizar Dias
Categorias
Categorias
Procurar
Procurar

Evento: ' IEEE: Palestra de Massimo de Gregorio Istituto di Cibernetica - Itália'

Geral
Data: Tuesday, November 14, 2006 At 15:00
Duração: 2 Horas
Contato - Info:
Pietro Maris Ferreira Presidente do Ramo Estudantil IEEE da UFRJ maris@ieee.org Felipe M. G. França, PhD Associate Professor felipe@ieee.org
Email:
URL:

O Ramo Estudantil IEEE da UFRJ e o Prof. Felipe M. G. França têm o prazer de convida-los para a palestra sobre o Agent WiSARD: a neurosymbolic approach to image reconstruction and understanding com Massimo de Gregorio.

Atenção a palestra será em inglês!

Centro de Tecnologia, Bloco H, Sala H215a
14 de Novembro de 2006, 15:00

Massimo de Gregorio is, since 1994, a Researcher at the Istituto di Cibernetica "E. Caianiello" - CNR. Presently, his research interests are mainly focused on the integration of artificial neural networks and symbolic reasoning systems, from the unified and classical points of view. He is author of nearly 50 scientific papers and in the last four years he has been the scientific coordinator of three different research projects.


Abstract:

High-level vision, which involves both top-down application of stored knowledge and bottom-up image processing, is a suitable testing domain for the effectiveness of explanations integrating pictures with verbal accounts of problem-solving processes. Indeed, high-level visual tasks may involve problem-solving processes that require the kinds of explanation that are appropriate for knowledge-based systems. Neurosymbolic computational systems can exploit largely complementary advantages of both neural and symbolic information processing. In view of their learning and associative capacities, neural networks are known to perform well in visual pattern recognition, whereas symbol manipulation algorithms are tailored for formal reasoning. Combining these features in a neurosymbolic system is clearly appealing for modelling tasks that involve both reasoning and low-level visual processing.
An additional motive of interest for neurosymbolic systems, more specifically related to the explanation problem, is emphasised here: a form of bidirectional associative behaviour of neural networks is exploited to produce mental images.
Furthermore, most of the reasoning carried out by the symbolic module can be implemented with the NSP (NeuroSymbolic Processor) that is a massively parallel interpreter of logic programs.
The proposed approach, that we refer to as the Agent WiSARD approach, has been tested in various situations and domains producing encouraging results. It was adopted for understanding and explaining portal shapes (ARCS – Arches Recall and Classification System), for describing 2D and 3D scenes formed by geometrical features, to help mobile agents (robots) in the auto location problem, and in "active" video surveillance systems.


Atenciosamente,

Pietro Maris Ferreira
Presidente do Ramo Estudantil IEEE da UFRJ

Felipe M. G. França, PhD
Associate Professor


Procurar no Calendário

Powered by ExtCalendar 2

© 2017 PESC/COPPE - Programa de Engenharia de Sistemas e Computação

Cidade Universitária, Centro de Tecnologia, Bloco H, Sala 319
Caixa Postal: 68511 CEP: 21941-972 Fones: +55 21 3938-8672 / +55 21 3938-8673 Fax: +55 21 3938-8676
Rio de Janeiro - RJ - Brasil
Horário de atendimento da Secretaria: 2a. a 6a. de 7:00 às 16:00 horas (exceto feriados escolares)