David Macêdo, PhD

David Macêdo, PhD

Senior AI Research Scientist & Engineer. Tech Lead

Artificial Intelligence Biography

Doctor of Philosophy (PhD) in Computer Science (Artificial Intelligence) [GPA 10.00/10.00], Center of Informatics, Federal University of Pernambuco (UFPE), Brazil. First Place in the Admission Process. “Towards Robust Deep Learning using Entropic Losses”.

https://arxiv.org/abs/2208.03566

Visiting Researcher with Montreal Institute for Learning Algorithms (MILA), University of Montreal (UdeM), Quebec, Canada.

https://mila.quebec/en/person/david-macedo

Authored one book and around fifty articles on Artificial Intelligence published in Top Conferences and Journals. More than a half thousand citations.

https://scholar.google.com/citations?user=hypWII4AAAAJ&hl=en

Top Main Conferences (NeurIPS, ICLR, ICML) and IEEE Reviewer.

https://neurips.cc/Conferences/2022/ProgramCommittee

https://iclr.cc/Conferences/2022/Reviewers

https://icml.cc/Conferences/2022/Reviewers

Co-creator and Collaborator Professor of the Artificial Intelligence course of the Computer Science Master and Doctorate Programs at the Center for Informatics (CIn), Federal University of Pernambuco (UFPE), Brazil.

https://dlmacedo.com/courses/deeplearning/

Participation in more than a hundred Research and Extension Projects. Google Research Award: “Robust Deep Learning”. UFPE, Brazil. Microsoft Research Award: “Deep Learning for Speaker Recognition”. UFPE, Brazil. Co-orientation of Masters Dissertations.

https://deeplearning.cin.ufpe.br/

https://deeplearning.cin.ufpe.br/research/

Master of Science (MSc) in Computer Science (Artificial Intelligence) [GPA 10.00/10.00], Center of Informatics, Federal University of Pernambuco (UFPE), Brazil. First Place in the Admission Process.

Bachelor of Science (BSc) in Electronic Engineering with Highest Academic Distinction (Best Overall Student) [GPA 9,54/10,00] from UFPE, Brazil. First Place in the Admission Process.

Interests

  • Artificial Intelligence (AI)
  • Computer Vision and Understanding
  • Natural Language Processing
  • Audio and Speech Processing
  • Tabular Data and Time Series
  • Attention and Transformers
  • Graph Neural Networks
  • Large Language Models
  • Foundation Models
  • Diffusion Models
  • Generative AI
  • Applications

Education

  • PhD Computer Science (Artificial Intelligence) [GPA 10.0/10.00], 2022

    Universidade Federal de Pernambuco (First Place in the Admission Process)

  • MSc Computer Science (Artificial Intelligence) [GPA 10.0/10.00], 2018

    Universidade Federal de Pernambuco (First Place in the Admission Process)

  • BSc Eletronic Engineering [GPA 9.56/10.00] with Highest Academic Distinction (Best Overall Student), 1996

    Universidade Federal de Pernambuco (First Place in the Admission Process)

Artificial Intelligence Course

Welcome to our Artificial Intelligence Course!

Follow the Artificial Intelligence Course for lecture videos (portuguese) and slides (english), codes, and additional resorces.

This is the syllabus for the course:

Artificial Intelligence Publications

*
All Foundations Computer Vision Natural Language Processing Audio and Speech Other Applications Dissertation Thesis Books Codes

Distinction Maximization Loss

Deep Learning Code. A project to improve out-of-distribution detection (open set recognition) and uncertainty estimation to your deep learning project.
Follow
Distinction Maximization Loss

Enhancing Deep Learning Performance Using Displaced Rectifier Linear Unit

Deep Learning Book. Published by Editora Dialética.
Follow
Enhancing Deep Learning Performance Using Displaced Rectifier Linear Unit

Enhancing deep learning performance using displaced rectifier linear unit

Deep Learning Masters Dissertation. Center of Informatics (CIn), Federal University of Pernambuco (UFPE).
Follow
Enhancing deep learning performance using displaced rectifier linear unit

Entropic Out-of-Distribution Detection

Deep Learning Code. A project to add scalable state-of-the-art out-of-distribution detection (open set recognition) support to your deep learning model.
Follow
Entropic Out-of-Distribution Detection

SVM-CNN

Deep Learning Code. A project to build a feature extractor based on Tensorflow and Scikit-learn created to improve the accuracy of SVM to classify MNIST dataset fast and with more accuracy.
Follow
SVM-CNN

The Robust Deep Learning Library

Deep Learning Code. A project to train your model from scratch or fine-tune a pretrained model using the losses provided in this library to improve out-of-distribution detection and uncertainty estimation performances.
Follow
The Robust Deep Learning Library

Towards Robust Deep Learning using Entropic Losses

Deep Learning PhD Thesis. Center of Informatics (CIn), Federal University of Pernambuco (UFPE).
Follow
Towards Robust Deep Learning using Entropic Losses

A Fast Fully Octave Convolutional Neural Network for Document Image Segmentation

International Joint Conference on Neural Networks 2020.
Follow
A Fast Fully Octave Convolutional Neural Network for Document Image Segmentation

Additive Margin SincNet for Speaker Recognition

International Joint Conference on Neural Networks 2019.
Follow
Additive Margin SincNet for Speaker Recognition

AM-MobileNet1D: A Portable Model for Speaker Recognition

International Joint Conference on Neural Networks 2020.
Follow
AM-MobileNet1D: A Portable Model for Speaker Recognition

An Adapted GRASP Approach for Hyperparameter Search on Deep Networks Applied to Tabular Data

International Joint Conference on Neural Networks 2022.
Follow
An Adapted GRASP Approach for Hyperparameter Search on Deep Networks Applied to Tabular Data

Convolution Optimization in Fire Classification

IEEE Access 2022.
Follow
Convolution Optimization in Fire Classification

Detecting Malicious HTTP Requests Without Log Parser Using RequestBERT-BiLSTM

Brazilian Conference on Intelligent Systems 2022.
Follow
Detecting Malicious HTTP Requests Without Log Parser Using RequestBERT-BiLSTM

Distantly-Supervised Neural Relation Extraction with Side Information using BERT

International Joint Conference on Neural Networks 2020.
Follow
Distantly-Supervised Neural Relation Extraction with Side Information using BERT

Enhancing Batch Normalized Convolutional Networks using Displaced Rectifier Linear Units: A Systematic Comparative Study

Expert Systems with Applications 2019.
Follow
Enhancing Batch Normalized Convolutional Networks using Displaced Rectifier Linear Units: A Systematic Comparative Study

Entropic Out-of-Distribution Detection

International Joint Conference on Neural Networks 2021.
Follow
Entropic Out-of-Distribution Detection

Entropic Out-of-Distribution Detection: Seamless Detection of Unknown Examples

IEEE Transactions on Neural Networks and Learning Systems 2021.
Follow
Entropic Out-of-Distribution Detection: Seamless Detection of Unknown Examples

Heartbeat Anomaly Detection using Adversarial Oversampling

International Joint Conference on Neural Networks 2019.
Follow
Heartbeat Anomaly Detection using Adversarial Oversampling

Hierarchical Attentional Hybrid Neural Networks for Document Classification

International Conference on Artificial Neural Networks 2019.
Follow
Hierarchical Attentional Hybrid Neural Networks for Document Classification

Identification of Microorganism Colony Odor Signature using InceptionTime

IEEE International Conference on Systems, Man, and Cybernetics 2021.
Follow
Identification of Microorganism Colony Odor Signature using InceptionTime

Improving Universal Language Model Fine-Tuning using Attention Mechanism

International Joint Conference on Neural Networks 2019.
Follow
Improving Universal Language Model Fine-Tuning using Attention Mechanism

Intrusion Detection for Cyber–Physical Systems Using Generative Adversarial Networks in Fog Environment

IEEE Internet of Things Journal 2020.
Follow
Intrusion Detection for Cyber–Physical Systems Using Generative Adversarial Networks in Fog Environment

Kutralnet: A Portable Deep Learning Model for Fire Recognition

International Joint Conference on Neural Networks 2020.
Follow
Kutralnet: A Portable Deep Learning Model for Fire Recognition

LogBERT-BiLSTM: Detecting Malicious Web Requests

International Conference on Artificial Neural Networks 2022.
Follow
LogBERT-BiLSTM: Detecting Malicious Web Requests

Multi-Class Mobile Money Service Financial Fraud Detection by Integrating Supervised Learning with Adversarial Autoencoders

International Joint Conference on Neural Networks 2021.
Follow
Multi-Class Mobile Money Service Financial Fraud Detection by Integrating Supervised Learning with Adversarial Autoencoders

Multi-human Fall Detection and Localization in Videos

Computer Vision and Image Understanding 2022.
Follow
Multi-human Fall Detection and Localization in Videos

PictoBERT: Transformers for Next Pictogram Prediction

Expert Systems with Applications 2022.
Follow
PictoBERT: Transformers for Next Pictogram Prediction

Reducing Squeezenet Storage Size with Depthwise Separable Convolutions

International Joint Conference on Neural Networks 2018.
Follow
Reducing Squeezenet Storage Size with Depthwise Separable Convolutions

SegNetRes-CRF: A Deep Convolutional Encoder-Decoder Architecture for Semantic Image Segmentation

International Joint Conference on Neural Networks 2018.
Follow
SegNetRes-CRF: A Deep Convolutional Encoder-Decoder Architecture for Semantic Image Segmentation

Spatial-Temporal Graph Convolutional Networks for Sign Language Recognition

International Conference on Artificial Neural Networks 2019.
Follow
Spatial-Temporal Graph Convolutional Networks for Sign Language Recognition

Squeezed Deep 6DoF Object Detection Using Knowledge Distillation

International Joint Conference on Neural Networks 2020.
Follow
Squeezed Deep 6DoF Object Detection Using Knowledge Distillation

Squeezed Very Deep Convolutional Neural Networks for Text Classification

International Conference on Artificial Neural Networks 2019.
Follow
Squeezed Very Deep Convolutional Neural Networks for Text Classification

Towards Optimizing Convolutional Neural Networks for Robotic Surgery Skill Evaluation

International Joint Conference on Neural Networks 2019.
Follow
Towards Optimizing Convolutional Neural Networks for Robotic Surgery Skill Evaluation

Training Aware Sigmoidal Optimization

Brazilian Conference on Intelligent Systems 2022.
Follow
Training Aware Sigmoidal Optimization

Unsupervised Multi-view Multi-person 3D Pose Estimation Using Reprojection Error

International Conference on Artificial Neural Networks 2022.
Follow
Unsupervised Multi-view Multi-person 3D Pose Estimation Using Reprojection Error