Identificador persistente para citar o vincular este elemento: http://hdl.handle.net/10553/106990
Título: Design and independent training of composable and reusable neural modules
Autores/as: Castillo Bolado, David Alejandro 
Guerra-Artal, Cayetano 
HernÁndez-Tejera, Mario 
Clasificación UNESCO: 120304 Inteligencia artificial
Palabras clave: Compositionality
Learning by role
Methodology
Modular training
Modularity, et al.
Fecha de publicación: 2021
Publicación seriada: Neural Networks 
Resumen: Monolithic neural networks and end-to-end training have become the dominating trend in the field of deep learning, but the steady increase in complexity and training costs has raised concerns about the effectiveness and efficiency of this approach. We propose modular training as an alternative strategy for building modular neural networks by composing neural modules that can be trained independently and then kept for future use. We analyse the requirements and challenges regarding modularity and compositionality and, with that information in hand, we provide a detailed design and implementation guideline. We show experimental results of applying this modular approach to a Visual Question Answering (VQA) task parting from a previously published modular network and we evaluate its impact on the final performance, with respect to a baseline trained end-to-end. We also perform compositionality tests on CLEVR.
URI: http://hdl.handle.net/10553/106990
ISSN: 0893-6080
DOI: 10.1016/j.neunet.2021.03.034
Fuente: Neural Networks [ISSN 0893-6080], v. 139, p. 294-304, (Julio 2021)
Colección:Artículos
Adobe PDF (853,53 kB)
Vista completa

Google ScholarTM

Verifica

Altmetric


Comparte



Exporta metadatos



Los elementos en ULPGC accedaCRIS están protegidos por derechos de autor con todos los derechos reservados, a menos que se indique lo contrario.