| Contact information:
andre.t.martins AT tecnico DOT ulisboa DOT pt
Instituto de Telecomunicacões
Torre Norte - Sala 9.07
Av. Rovisco Pais, 1 1049-001 Lisboa - Portugal
Phone: +351 218418454
I am an Associate Professor at Instituto Superior Técnico, Senior Researcher at the Instituto de Telecomunicações, and VP of AI Research at Unbabel in Lisbon, Portugal. I also do scientific consulting for Priberam Labs. I work on natural language processing and machine learning.
Until 2012, I was a PhD student in the joint CMU-Portugal program in Language Technologies, at Carnegie Mellon University and Instituto Superior Técnico. My advisors were Mário Figueiredo, Noah Smith, Pedro Aguiar and Eric Xing.
- Chryssa Zerva (Post-doc at IT, 2021-)
- Vlad Niculae (Post-doc at IT, 2018-2020, now Assistant Professor at University of Amsterdam)
- Erick Fonseca (Post-doc at IT, 2018-2020), now Research Scientist at Cleverly
Students and Interns
- Haau-Sing Li (ELLIS PhD at TU Darmstadt University, 2021-; co-advised with Iryna Gurevych)
- Nuno Guerreiro (PhD at IST, 2020-)
- Patrick Fernandes (PhD at CMU/IST, 2020-; co-advised with Graham Neubig)
- Taisiya Glushkova (PhD at IST, 2020-)
- Nuno Guerreiro (PhD at IST, 2020-)
- António Farinhas (MSc at IST, 2020)
- Pedro Coelho (MSc at IST, 2020; co-advised with Christine Maroti at Unbabel)
- João Alves (MSc at IST, 2020; co-advised with Amin Farajian at Unbabel)
- João Moura (MSc at IST, 2020; co-advised with Fábio Kepler at Unbabel)
- Rita Costa (MSc at IST, 2020; co-advised with Luisa Coheur)
- Nikita Nangia (Summer Intern at IT, 2019)
- Michael Zhang (Research Intern at Unbabel, 2019)
- Daan van Stigt (Research Intern at Unbabel, 2019)
- Marcos Treviso (PhD at IST, 2019-, co-advised with Mário Figueiredo)
- Tsvetomila Mihaylova (PhD at IST, 2018-, co-advised with Vlad Niculae))
- Ben Peters (PhD at IST, 2018-)
- Gonçalo Correia (ELLIS PhD at IST, 2018-, co-advised with Vlad Niculae)
- Pedro Martins (PhD at IST, 2018-; co-advised with Zita Marinho)
- Sameen Maruf (PhD at Monash University, 2016-: co-advised with Reza Haffari)
- Zita Marinho (PhD at CMU/IST, 2013-2018; co-advised with Geoff Gordon and Sidd Srinivasa, now Head of Research at Priberam Labs)
- Telmo Pires (Research Intern at IT and Unbabel, 2017, now at Google AI)
- Pedro Ferreira (MSc at IST, 2017-)
- Chaitanya Malaviya (Research Intern at Unbabel, 2017, now MSc at CMU)
- Julia Kreutzer (Research Intern at Unbabel, 2016, now PhD student at Heidelberg University)
- Daniel Ferreira (MSc at IST, 2015; co-advised with Mariana Almeida; now PhD at Vienna Technical University)
- Renato Negrinho (Research Intern at Priberam, 2014, now PhD at CMU)
- Vasco Marcelino (MSc at IST, 2014; co-advised with Miguel Almeida)
- Marta Quintão (MSc at IST, 2014)
- Pedro Ramos (MSc at IST, 2013)
- I am giving a keynote talk at the TRITON conference.
- I gave a keynote talk at TALN where I presented some of the work we did in the DeepSPIN project. Here are the slides.
- I am co-organizing LxMLS 2021 (Lisbon Machine Learning School), which will be a fully remote school this year. See here for details!
- We have new papers accepted at NAACL 2021 and ACL 2021.
- I am AC for NAACL 2021, NeurIPS 2021, EMNLP 2021, ICLR 2022.
- I am Action Editor for TACL.
- I am co-directing the ELLIS NLP program with Iryna Guleyvich and Ivan Titov, with an amazing list of fellows and scholars!
- We're launching LUMLIS, the Lisbon unit of the ELLIS network!
- We have new papers accepted at EMNLP 2020 and NeurIPS 2020.
- I am SAC for EACL 2021 and AC for ICLR 2021.
- We have an opening for a Post-Doc position in the DeepSPIN project. See here for details!
- We have new papers accepted at ACL 2020, ICML 2020, and EAMT 2020.
- We have a new JMLR paper on Learning with Fenchel-Young Losses (work done with collaboration with Mathieu Blondel).
- I am AC for NeurIPS 2020 and EMNLP 2020.
- I was SAC for ACL 2020 and AC for ICLR 2020.
- We presented a tutorial on Latent Structure Models for NLP at ACL 2019 in Florence.
- I am giving an invited talk at the First EurNLP Summit in London.
- I am giving invited talks at 3 Summer schools: LxMLS 2019 in Lisbon, AthNLP 2019 in Athens, and MLRS 2019 in Bangkok.
- I was part of a discussion panel at Culturgest about AI. Here's the video (in Portuguese).
- We have new papers accepted at AISTATS, NAACL, ACL, and EMNLP 2019.
- Unbabel won the WMT 2019 Shared Task on Quality Estimation! Check the results here!
- We received the best system demo paper award for OpenKiwi, a Pytorch-based software toolkit for translation quality estimation. Check the repo and the demo paper at ACL 2019!
- We have a new EMNLP paper where we propose SparseMAP to build dynamic computation graphs via sparse latent structure.
- I gave an invited talk in the ACL 2018 Workshop on Neural Machine Translation and Generation. Here are the slides.
- We have a new ICML paper where we propose SparseMAP as a new inference procedure for sparse structured prediction (work done in collaboration with Vlad Niculae, Mathieu Blondel, and Claire Cardie).
- We have a new ACL short paper where we use new forms of sparse and constrained attention within neural machine translation (work done in collaboration with Chaitanya Malaviya and Pedro Ferreira).
- Great news: I got an ERC (European Research Council) Starting Grant on "Deep Structured Prediction in Natural Language Processing (DeepSPIN)". I am now looking for Post-Doc and PhD Students.
- I'm giving an invited talk in the EMNLP 2017 Workshop on Structured Prediction for NLP.
- I gave an invited talk in the ICML 2017 Workshop on Learning to Generate Natural Language and co-organized the ICML 2017 Workshop on Deep Structured Prediction.
- Our paper on end-to-end differentiable neural-easy first decoders made it to EMNLP 2017!
- We have a new TACL paper where we further improve our quality estimation system by using automatic post-editing as a auxiliary task.
- Our word-level quality estimation system achieved the top score in the First Conference on Machine Translation (WMT 2016). Check the paper here.
- Our paper on sparsemax was accepted at ICML 2016. It's a sparse, differentiable, alternative to the softmax activation.
- I joined Unbabel! We do AI powered human quality translation.
- I presented a tutorial on "Linear Programming Decoders in NLP" at EMNLP 2014, which covers integer linear programming, message-passing algorithms, and dual decomposition. The slides are available here and the videos here (part 1, part 2).
- Our TurboSemanticParser system achieved the top score in the open track of the SemEval 2014 task on "Broad-Coverage Semantic Dependency Parsing." Check the paper here.
- I presented a tutorial on "Structured Sparsity in NLP" at EACL 2014. The slides are available here (see here for an older version of the same tutorial presented at NAACL 2012).
- I received the Portuguese IBM Scientific Prize for my work in structured prediction and "Turbo Parsing."
- I successfully defended my thesis in May 11, 2012, at CMU!