William Berrios

I’m William, a bachelor in Mechatronics Engineering (Robotics 🤖) from UNI. Currently, I am working as a research intern at the The Center for Brains, Minds & Machines - MIT working with Arturo Deza examining the potential of Visual Transformers and adversarial training on the modeling of the ventral stream processing. Previously, I worked as a computer vision research intern developing a deep active learning framework to help curators efficiently label biomedical images at the Electronic Visualization Laboratory under the supervision of Professor Elisabeta Marai.

My research interests lie in building robust and generalizable computer vision systems with minimal or no supervision. Specific areas of interest are representation learning, generative modeling, few-shot learning, adversarial robustness and robotics.

Eager to contribute to science development in Peru, I am actively organizing meetings with graduate students in computer science at REPU for peruvian undergraduates. This year, I am looking forward to pursuing a PhD in Computer Science and becoming a first-generation graduate student in my family.

Email / Resume / LinkedIn / GitHub / Twitter

News

02/22 Got second place 🥈 at Brain-Score Competition organized by MIT.
10/21 Obtained a silver medal🥈 at Ventilator Pressure Prediction Competition - Kaggle
08/21 Honored by the Ministry of Education of Peru 🇵🇪 in recognition for outstanding participations at international machine learning competitions. Interview (Spanish).
07/21 Pleased to help as Finance & Sponsor Chair of LatinX in AI Workshop @ ICML’ 21

Publications

Sunflower

Joint Rotational Invariance and Adversarial Training of a dual-stream Transformer yields state of the art Brain-Score for Area V4
William Berrios, Arturo Deza
Workshop on Attention and Transformers in Vision - CVPR 2022 | Under Review at NeurIPS
[Paper][Code]

Sunflower

BI-LAVA: Biocuration Image Labeling through Active Learning and Visual Analysis
Juan Trelles, William Berrios, Hagit Shatkay and G.Elisabeta Marai
Under Review at IEEE VIS 2022
[Paper Coming]

Competitions

  • 1st Place at 24h Hackaton Rimac - April.22
    Tabular competition, boosting models, big data
    Rank 1/80 participants - LATAM.
    Awarded by $4000.

  • 2nd Place at Brain-Score Competition - Feb.22
    Neuroscience, Primate Object Recognition, Vision Transformers
    Organized by Brain-Score Team (MIT) and sponsored by MIT Quest for Intelligence, C-BRIC and MIT-IBM Watson AI Lab
    Awarded by $1250
    [Competition][Code]

  • 1st Place at BCI Machine Learning Competition - Dec.21
    Tabular competition, boosting models, big data
    Rank 1/400 participants (15+ countries).
    Awarded by $3600.
    [Competition] [Code]

  • Silver Medal at Ventilator Pressure Prediction Competition - Oct.21
    Time series competition, Deep Learning, LSTM's
    Top 5% - Organized by Google Brain and Kaggle.
    [Competition] [Code]

  • 1st Place at BNP Machine Learning Competition - May.21
    Text Data, Deep Learning, Transformers
    Rank 1/100 data scientists from LATAM
    Organized by BNP Paribas Cardiff and Domino DataLab
    [Description] [Code]

  • 2nd Place at International Interbank Datathon - Feb.21
    Tabular competition, boosting models
    Rank 2/229 data scientists. Organized by Interbank - Peru
    Awarded by $6000 dollars

  • 1st Place at BanColombia Datathon - Jan.21
    Tabular competition, boosting models
    Rank 1/80 teams - LATAM. Organized by Group BanColombia
    Awarded by $2000 dollars