Repositorio Digital de la UPF
Publicación
Contribución de congreso (ConferenceOutput)
"The red one!": on learning to refer to things based on discriminative properties
Comunicació presentada a: 54th Annual Meeting of the Association for Computational Linguistics celebrat del 7 al 12 d'agost de 2016 a Berlín, Alemanya., As a first step towards agents learning to communicate about their visual environment, we propose a system that, given visual representations of a referent (CAT) and a context (SOFA), identifies their discriminative attributes, i.e., properties that distinguish them (has_tail). Moreover, although supervision is only provided in terms of discriminativeness of attributes for pairs, the model learns to assign plausible attributes to specific objects (SOFA-has_cushion). Finally, we present a preliminary experiment confirming the referential success of the predicted discriminative attributes., This work was supported by ERC 2011 Starting Independent Research Grant n. 283554 (COMPOSES).
Proyecto:
EC / FP7 283554