Gonçalo M. Correia

Machine Learning/NLP Ph.D. candidate supervised by André Martins, Vlad Niculae, and Wilker Aziz. ELLIS Ph.D. between Lisbon and Amsterdam. Interested in any applications of Machine Learning; while his strengths are currently in NLP applications, he's positive that he has the ability to learn quickly the finer details of any other application. Click here to download my single-page Resumé.

Education

  1. TitleB.Sc. in Biomedical Engineering

    InstitutionInstituto Superior Técnico (University of Lisbon)

    Description

    In my B.Sc., I learned the foundations of engineering, mathematics, and physics, along with medical related courses. During these years, I earned two Academic Merit awards.

  2. TitleM.Sc. in Artificial Intelligence (Machine Learning)

    InstitutionThe University of Edinburgh

    Thesis TitleMaking Machines More Human: A Multitask Learning Approach to VQA and Human Attention Prediction
    Description

    In my M.Sc., I learned the foundations of Machine Learning, Deep Learning, and NLP. Frank Keller advised my thesis, which was about using human attention as an additional supervision signal to improve the performance of a Visual Question Answering model. I finished the degree with First Honors with Distinction.

  3. TitlePhD in Electrical and Computer Engineering (Machine Learning)

    InstitutionInstituto Superior Técnico (University of Lisbon) and University of Amsterdam

    Thesis TitleLearnable Sparsity and Weak Supervision for Data-Efficient, Transparent, and Compact Neural Models
    Description

    My Ph.D. journey has not only taught me valuable skills in Machine Learning, Deep Learning, and Natural Language Generation but also to be self-taught and autonomous. I have published in top-tier venues during these years.

Work Experience

  1. RoleSummer Intern

    CompanyUnbabel

    Description

    Development of a Language Identification Tool capable of distinguishing between several languages and particularly between language varieties. Development of a web scraper to extract sentence pairs from the web in order to train machine translation models. I was supervised by the CTO of Unbabel, João Graça, Ph.D.

  2. RoleResearch Assistant

    CompanyInstituto de Telecomunicações/Unbabel

    Description

    Development of low-resource solutions for Automatic Post-Editing of Machine Translations. It set the stage for a publication at ACL 2019.

Publications

  1. TitleA Simple and Effective Approach to Automatic Post-Editing with Transfer Learning

    ConferenceIn Proceedings of ACL, 2019

    AuthorsGonçalo M Correia, André FT Martins

    Comment📄 Presented as a Poster.

  2. TitleUnbabel's Submission to the WMT2019 APE Shared Task: BERT-based Encoder-Decoder for Automatic Post-Editing

    ConferenceIn Proceedings of WMT, 2019

    AuthorsAntónio V. Lopes, M. Amin Farajian, Gonçalo M. Correia, Jonay Trenous, André F.T. Martins

    Comment🏆 Winner of the Shared Task!

  3. TitleAdaptively Sparse Transformers

    ConferenceIn Proceedings of EMNLP, 2019

    AuthorsGonçalo M. Correia, Vlad Niculae and André F.T. Martins

    Comment🤖 Presented as a Talk in Machine Learning Session III.

  4. TitleEfficient Marginalization of Discrete and Structured Latent Variables via Sparsity

    ConferenceIn Proceedings of NeurIPS, 2020

    AuthorsGonçalo M. Correia, Vlad Niculae, Wilker Aziz and André F.T. Martins

    Comment⭐ Spotlight paper.

Teaching

  1. RoleSummer School Monitor

    PlaceLxMLS

    Description

    I was a monitor on the 2018 edition of the Lisbon Machine Learning School. Besides helping out in the day-to-day organization of the school, I was also responsible of answering questions and helping out students during the Lab sessions.

  2. RoleInstructor of Text Classification Specialization

    PlaceLisbon Data Science Starters Academy

    Description

    During July 2018, Catarina Silva, Pedro Ferreira and I were responsible for the 4th Specialization of the Lisbon Data Science Starters Academy, releasing 3 Learning Units and organizing an hackaton related to the topic of Text Classification. Here's the links for the Learning Notebooks:

  3. RoleSummer School Monitor

    PlaceLxMLS

    Description

    I was a monitor on the 2019 edition of the Lisbon Machine Learning School. Besides helping out in the day-to-day organization of the school, I was also responsible of answering questions and helping out students during the Lab sessions.

  4. RoleTeaching Assistant in Deep Structured Learning

    PlaceInstituto Superior Técnico (University of Lisbon)

    Description

    As a TA, I helped to grade and create the homeworks of the PhD course Deep Structured Learning. I also gave one of the lectures — the PyTorch Tutorial!

Skills

Proficient (Tools)

  • Python
  • PyTorch
  • PyTorch Lightning
  • HuggingFace's Transformers
  • Weights & Biases and TensorBoard
  • Git
  • NumPy
  • LaTeX
  • Matplotlib

I've also used

  • Tensorflow
  • Scikit-learn
  • Pandas
  • Bash and Shell Scripting

Knowledge

  • Machine Learning and Statistics
  • Numerical/Statistical Programming
  • Natural Language Processing
  • Deep Learning
  • Neural Networks for NLP
  • Transformers and Pre-trained Language Models
  • Natural Language Generation
  • Generative and Latent Variable Models
  • Weakly and Semi-Supervised Learning