Iacer Calixto :: education

2013 - Present  ::  PhD. candidate in ADAPT Centre.

Dublin City University, Trinity College Dublin (Ireland)

I work on multi-modal neural and statistical machine translation. I specifically study how to incorporate multimodal data (e.g., videos, speech recordings, images) into a state-of-the-art MT pipeline, and believe that multimodal data can be beneficial to machine translation systems in several distinct scenarios, such as in e-commerce localisation or the automatic transcription of conferences for a multilingual audience.

I have just got a paper accepted at ACL 2017 where I propose a multi-modal, doubly-attentive decoder for Neural Machine Translation; one paper accepted at EMNLP 2017 where I study different strategies to incorporate global visual featurs into Neural Machine Translation; and also one more paper accepted at RANLP 2017 (a short version of the paper is available here), where I propose multilingual and multi-modal sentence embeddings.

2011 - 2013  ::  Multiple Masters in Natural Language Processing and Human Language Technology.

University of Wolverhampton (England), Université de Franche-Comté (France), Universidade do Algarve (Portugal)

Thesis: Towards the generation of a database for scientific research in Natural Language Processing with an Information Extraction system.

2009 - 2011  ::  MSc. in Computer Science.

Federal University of Goiás, UFG (Brazil)

Thesis: An Origin-Destination Matrix Estimation Method Based on Fuzzy Linear Programming with focus on Brazilian Congested Networks (pdf in Portuguese).

2003 - 2007  ::  BSc. in Information Systems.

Faculdades Alves Faria, ALFA (Brazil)

Software requirements, analysis, design and implementation. Calculus, algebra and statistics. Software, process and IT management.