Heitor R. Guimarães

heitor.guimaraes at inrs dot ca


Ph.D. student & Senior Data Scientist

Montreal, QC, Canada

I’m a Ph.D. student in Telecommunications at INRS under the supervision of Prof. Tiago Falk. My research interest lies in efficient deep learning (e.g., Model compression and Efficient architectures), representation learning, and self-supervised learning for speech and language processing (SLP). In addition, I aim to build efficient speech representation learning algorithms robust to distribution shifts and malicious users.

Previously, I worked as a senior data scientist at Itaú-Unibanco, where I was responsible for developing a wide range of ML solutions to handle tabular data, speech, and NLP, from conception to large-scale deployment, impacting more than 20 million users. In the past, I also had some internship experiences with investment funds and in the O&G fields. On the academic side, I have received my M.Sc. degree in Electrical Engineering at USP, a Specialization in Data Science at ITA, and my B.S. in Computer and Information Engineering from UFRJ.

For more details, please take a look at my CV (last update: 2023/10/18).

Apart from my professional interests, I like to practice Wushu (lineage: Wu Ji Tang Lang Men), and I enjoy the Beat Generation literary movement, mainly the works of Kerouac, Ginsberg, and Cassady.


Nov 8, 2023 Our paper “Adapting Self-Supervised Features for Background Speech Detection in Beehive Audio Recordings” received the Best Paper Presented by a Young Researcher Award at IEEE MetroAgriFor
Nov 3, 2023 Received the IEEE Signal Processing Society (SPS) Scholarship!
Sep 4, 2023 We got two papers accepted at IEEE MetroAgriFor on Beehive Acoustics Monitoring
Jul 15, 2023 Received the CIFAR Inclusive AI Scholarship to attend the CIFAR DLRL Summer School!
Jun 2, 2023 We got one paper accepted at the AI and (cyber)security workshop @ IEEE SMC 2023

selected publications

  1. Conference
    VIC-KD: Variance-Invariance-Covariance Knowledge Distillation to Make Keyword Spotting More Robust Against Adversarial Attacks
    Heitor R. Guimarães, Arthur Pimentel, Anderson R. Avila, and 1 more author
    In 2024 International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2024
  2. Conference
    RobustDistiller: Compressing Universal Speech Representations for Enhanced Environment Robustness
    Heitor R. Guimarães, Arthur Pimentel, Anderson R. Avila, and 3 more authors
    In 2023 International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2023
  3. Conference
    Improving the Robustness of DistilHuBERT to Unseen Noisy Conditions via Data Augmentation, Curriculum Learning, and Multi-Task Enhancement
    Heitor R. Guimarães, Arthur Pimentel, Anderson R. Avila, and 2 more authors
    In NeurIPS 2022 Efficient Natural Language and Speech Processing Workshop, 2022
  4. Conference
    A Perceptual Loss Based Complex Neural Beamforming for Ambix 3D Speech Enhancement
    Heitor R. Guimarães, Wesley Beccaro, and Miguel A. Ramirez
    In Proc. L3DAS22: Machine Learning for 3D Audio Signal Processing, 2022
  5. Journal
    Monaural speech enhancement through deep wave-U-net
    Heitor R. Guimarães, Hitoshi Nagano, and Diego W. Silva
    Expert Systems with Applications, 2020