Publications

You can also find my articles on my Google Scholar profile.

Journal Articles

Few-shot and continuous online learning for forecasting in the energy industry

Published in Energy, 2025

Forecasting in the energy sector is critical for planning and efficiency, but existing methods require extensive historical data and struggle with changing conditions. This work presents a few-shot forecasting method for energy time series prediction under nonstationary conditions and data scarcity. The solution “plugs into” any existing regressor, combining ideas to create a flexible data-efficient tool.

Recommended citation: Gabriel Cirac, Vinicius Eduardo Botechia, Denis José Schiozer, Víctor Martínez, Rafael de Oliveira Werneck, and Anderson Rocha. Few-shot and continuous online learning for forecasting in the energy industry. Energy, 336:138470, 2025.
DOI | Download Bibtex

Rock-type classification: A (critical) machine-learning perspective

Published in Computers & Geosciences, 2024

We investigate machine-learning techniques for rock-type classification. A throughout literature review (considering the machine-learning technique, number of classes, rock types, and image types) presents a diversity of datasets employed and a wide range of classification results as well as multiple problem formulations. Throughout the discussion of the literature, we highlight some common machine-learning pitfalls and criticize the decisions taken by some authors on the problem formulation.

Recommended citation: Pedro Ribeiro Mendes, Soroor Salavati, Oscar Linares, Maiara Moreira Gonçalves, Marcelo Ferreira Zampieri, Vitor Hugo de Sousa Ferreira, Manuel Castro, Rafael de Oliveira Werneck, Renato Moura, Elayne Morais, Ahmed Esmin, Leopoldo Lusquino, Denis José Schiozer, Alexandre Ferreira, Alessandra Davólio, and Anderson Rocha. Rock-type classification: A (critical) machine-learning perspective. Computers Geosciences, 193:105730, 2024.
DOI | Download Bibtex

A multi-modal approach for mixed-frequency time series forecasting

Published in Neural Computing and Applications, 2024

This study proposes a novel multimodal approach for mixed-frequency time series forecasting in the oil industry, enabling the use of high-frequency (HF) data in their original frequency. We specifically address the challenge of integrating HF data streams, such as pressure and temperature measurements, with daily time series without introducing noise.

Recommended citation: Leopoldo Lusquino Filho, Rafael de Oliveira Werneck, Manuel Castro, Pedro Ribeiro Mendes Júnior, Augusto Lustosa, Marcelo Zampieri, Oscar Linares, Renato Moura, Elayne Morais, Murilo Amaral, Soroor Salavati, Ashish Loomba, Ahmed Esmin, Maiara Gonçalves, Denis José Schiozer, Alexandre Ferreira, Alessandra Davólio, and Anderson Rocha. A multi-modal approach for mixed-frequency time series forecasting. Neural Computing and Applications, 36(34):21581–21605, 2024.
DOI | Download Bibtex

Time series causal relationships discovery through feature importance and ensemble models

Published in Scientific Reports, 2023

Inferring causal relationships from observational data is a key challenge in understanding the interpretability of Machine Learning models. Given the ever‑increasing amount of observational data available in many areas, Machine Learning algorithms used for forecasting have become more complex, leading to a less understandable path of how a decision is made by the model. To address this issue, we propose leveraging ensemble models, e.g., Random Forest, to assess which input features the trained model prioritizes when making a forecast and, in this way, establish causal relationships between the variables.

Recommended citation: Manuel Castro, Pedro Ribeiro Mendes Júnior, Aurea Soriano-Vargas, Rafael de Oliveira Werneck, Maiara Moreira Gonçalves, Leopoldo Lusquino Filho, Renato Moura, Marcelo Zampieri, Oscar Linares, Vitor Ferreira, Alexandre Ferreira, Alessandra Davólio, Denis Schiozer, and Anderson Rocha. Time series causal relationships discovery through feature importance and ensemble models. Scientific Reports, 13(1):11402, 2023.
DOI | Download Bibtex

Data-driven deep-learning forecasting for oil production and pressure

Published in Journal of Petroleum Science and Engineering, 2022

Production forecasting plays an important role in oil and gas production, aiding engineers to perform field management. However, this can be challenging for complex reservoirs such as the highly heterogeneous carbonate reservoirs from Brazilian Pre-salt fields. We propose a new setup for forecasting multiple outputs using machine-learning algorithms and evaluate a set of deep-learning architectures suitable for time-series forecasting.

Recommended citation: Rafael de Oliveira Werneck, Raphael Prates, Renato Moura, Maiara Moreira Gonçalves, Manuel Castro, Aurea Soriano-Vargas, Pedro Ribeiro Mendes Júnior, M. Manzur Hossain, Marcelo Ferreira Zampieri, Alexandre Ferreira, Alessandra Davólio, Denis Schiozer, and Anderson Rocha. Data-driven deep-learning forecasting for oil production and pressure. Journal of Petroleum Science and Engineering, 210:109937, 2022.
DOI | Download Paper | Download Bibtex

A visual analytics approach to anomaly detection in hydrocarbon reservoir time series data

Published in Journal of Petroleum Science and Engineering, 2021

Detecting anomalies in time series data of hydrocarbon reservoir production is crucially important. Anomalies can result for different reasons: gross errors, system availability, human intervention, or abrupt changes in the series. We have developed a visual analytics approach based on an interactive visualization of time series data involving machine learning approaches for anomaly identification.

Recommended citation: Aurea Soriano-Vargas, Rafael Werneck, Renato Moura, Pedro Mendes Júnior, Raphael Prates, Manuel Castro, Maiara Gonçalves, Manzur Hossain, Marcelo Zampieri, Alexandre Ferreira, Alessandra Davólio, Bernd Hamann, Denis José Schiozer, and Anderson Rocha. A visual analytics approach to anomaly detection in hydrocarbon reservoir time series data. Journal of Petroleum Science and Engineering, 206:108988, 2021.
DOI | Download Paper | Download Bibtex

Learning cost function for graph classification with open-set methods

Published in Pattern Recognition Letters, 2019

In several pattern recognition problems, effective graph matching is of paramount importance. In this paper, we introduce a novel framework to learn discriminative cost functions. These cost functions are embedded into a graph matching-based classifier.

Recommended citation: Rafael Werneck, Romain Raveaux, Salvatore Tabbone, and Ricardo da Silva Torres. Learning cost function for graph classification with open-set methods. Pattern Recognition Letters, 128:8 – 15, 2019.
DOI | Download Paper | Download Bibtex

Exploiting ConvNet Diversity for Flooding Identification

Published in IEEE Geoscience Remote Sensing Letters, 2018

Flooding is the world’s most costly type of natural disaster in terms of both economic losses and human causalities. A first and essential procedure toward flood monitoring is based on identifying the area most vulnerable to flooding, which gives authorities relevant regions to focus. In this letter, we propose several methods to perform flooding identification in high-resolution remote sensing images using deep learning.

Recommended citation: Keiller Nogueira, Samuel G. Fadel, Ícaro C. Dourado, Rafael de Oliveira Werneck, Javier A. V. Muñoz, Otávio A. B. Penatti, Rodrigo Tripodi Calumby, Lin Tzy Li, Jefersson A. dos Santos, and Ricardo da Silva Torres. Exploiting convnet diversity for flooding identification. IEEE Geosci. Remote Sensing Lett., 15(9):1446–1450, 2018.
DOI | Download Paper | Download Bibtex

Graph-based bag-of-words for classification

Published in Pattern Recognition, 2018

This paper introduces the Bag of Graphs (BoG), a Bag-of-Words model that encodes in graphs the local structures of a digital object. Two BoG-based methods, Bag of Singleton Graphs (BoSG) and Bag of Visual Graphs (BoVG), are defined and evaluated for graph and image classification.

Recommended citation: F. B. Silva, R. de O. Werneck, S. Goldenstein, S. Tabbone, and R. da S. Torres. Graph-based bag-of-words for classification. Pattern Recognition, 74(Supplement C):266 – 285, February 2018.
DOI | Download Paper | Download Bibtex

Kuaa: A unified framework for design, deployment, execution, and recommendation of machine learning experiments

Published in Future Generation Computer Systems, 2018

In this work, we propose Kuaa, a workflow-based framework that can be used for designing, deploying, and executing machine learning experiments in an automated fashion. This framework is able to provide a standardized environment for exploratory analysis of machine learning solutions, as it supports the evaluation of feature descriptors, normalizers, classifiers, and fusion approaches in a wide range of tasks involving machine learning.

Recommended citation: Rafael Werneck, Waldir Rodrigues de Almeida, Bernardo Vecchia Stein, Daniel Vatanabe Pazinato, Pedro Ribeiro Mendes Júnior, Otávio Augusto Bizetto Penatti, Anderson Rocha, and Ricardo da Silva Torres. Kuaa: A unified framework for design, deployment, execution, and recommendation of machine learning experiments. Future Generation Computer Systems, 2018.
DOI | Download Paper | Download Bibtex

Nearest neighbors distance ratio open-set classifier

Published in Machine Learning, 2017

In this paper, we propose a novel multiclass classifier for the open-set recognition scenario. This scenario is the one in which there are no a priori training samples for some classes that might appear during testing.

Recommended citation: P. R. Mendes Júnior, R. M. de Souza, R. de O. Werneck, B. V. Stein, D. V. Pazinato, W. R. de Almeida, O. A. B. Penatti, R. da S. Torres, and A. Rocha. Nearest neighbors distance ratio open-set classifier. Machine Learning, 106(3):359–386, 3 2017.
DOI | Download Paper | Download Bibtex

Where is my puppy? Retrieving lost dogs by facial features

Published in Multimedia Tools and Applications, 2017

A pet that goes missing is among many people’s worst fears: a moment of distraction is enough for a dog or a cat wandering off from home. Some measures help matching lost animals to their owners; but automated visual recognition is one that — although convenient, highly available, and low-cost — is surprisingly overlooked. In this paper, we inaugurate that promising avenue by pursuing face recognition for dogs.

Recommended citation: Thierry Pinheiro Moreira, Mauricio Lisboa Perez, Rafael de Oliveira Werneck, and Eduardo Valle. Where is my puppy? retrieving lost dogs by facial features. Multimedia Tools and Applications, 76(14):15325–15340, 2017.
DOI | Download Paper | Download Bibtex

Pixel-Level Tissue Classification for Ultrasound Images

Published in IEEE Journal of Biomedical and Health Informatics, 2016

Pixel-level tissue classification for ultrasound images, commonly applied to carotid images, is usually based on defining thresholds for the isolated pixel values. We propose the use of descriptors that take into account information about a neighborhood of a pixel when classifying it.

Recommended citation: D. V. Pazinato, B. V. Stein, W. R. de Almeida, R. de O. Werneck, P. R. M. JÃžnior, O. A. B. Penatti, R. d. S. Torres, F. H. Menezes, and A. Rocha. Pixel-level tissue classification for ultrasound images. IEEE Journal of Biomedical and Health Informatics, 20(1):256–267, 1 2016.
DOI | Download Paper | Download Bibtex

Mid-level image representations for real-time heart view plane classification of echocardiograms

Published in Computers in Biology and Medicine, 2015

In this paper, we explore mid-level image representations for real-time heart view plane classification of 2D echocardiogram ultrasound images. The proposed representations rely on bags of visual words, successfully used by the computer vision community in visual recognition problems.

Recommended citation: Otávio A.B. Penatti, Rafael de O. Werneck, Waldir R. de Almeida, Bernardo V. Stein, Daniel V. Pazinato, Pedro R. Mendes Júnior, Ricardo da S. Torres, and Anderson Rocha. Mid-level image representations for real-time heart view plane classification of echocardiograms. Computers in Biology and Medicine, 66:66 – 81, 2015.
DOI | Download Paper | Download Bibtex

Conference Papers

Watch the reservoir! Improving short-term production forecast through transformers

Published in SPE Europe Energy Conference and Exhibition, 2024

Data-driven methodologies have been used in reservoir management and production forecasting, particularly demonstrating remarkable efficacy in short-term oil production forecasts. However, there is space to improve its prediction, especially in tackling the complexities of challenging reservoirs, such as the heterogeneous carbonate reservoirs from Brazilian Pre-salt fields. In this paper, we propose a new strategy to improve short-term forecasting for oil production through attention mechanisms that boost state-of-the-art methods.

Recommended citation: R. Werneck, L. A. Lusquino Filho, A. Lustosa, A. Loomba, M. M. Gonçalves, A. Esmin, S. Salavati, E. Morais, P. Ribeiro Mendes Junior, M. Zampieri, M. Amaral, O. C. Linares, M. Castro, R. Moura, D. J. Schiozer, A. Mello Ferreira, A. Davolio, and A. Rocha. Watch the reservoir! Improving short-term production forecast through transformers. volume SPE Europe Energy Conference and Exhibition of SPE Europec featured at EAGE Conference and Exhibition, page D031S016R002, 06 2024.
DOI | Download Bibtex

Enhancing Short-Term Production Forecast in Oil Fields: Integrating Data-Driven and Model-Based Approaches to Reduce Uncertainty

Published in SPE Europe Energy Conference and Exhibition, 2024

Reservoir simulation models are usually applied to optimize oil field production across its life cycle but face challenges in short-term forecasting. Data-driven techniques (DD) show promise for short-term predictions but lack reliability over extended periods. This study introduces a Hybrid Methodology (HyM) combining the optimal features of model-based (MB) with DD approaches to select the best simulation models to make short-term decisions, effectively reducing uncertainty in short-term production forecasts for a real field.

Recommended citation: M. M. Gonçalves, R. Werneck, M. Castro, M. Amaral, P. Ribeiro Mendes, L. A. Lusquino Filho, A. Esmin, R. Moura, E. Morais, O. C. Linares, A. Lustosa, S. Salavati, D. J. Schiozer, A. Mello Ferreira, A. Rocha, and A. Davolio. Enhancing short-term production forecast in oil fields: Integrating data-driven and model-based approaches to reduce uncertainty. volume SPE Europe Energy Conference and Exhibition of SPE Europec featured at EAGE Conference and Exhibition, page D031S019R002, 06 2024.
DOI | Download Bibtex

A New Hybrid Data-Driven and Model-Based Methodology for Improved Short-Term Production Forecasting

Published in Offshore Technology Conference, 2023

Model-based (MB) solutions are widely used in reservoir management and production forecasting throughout the life-cycle of oil fields. However, such approaches are not often used for short-term (up to six months) forecasting due to the immediate-term productivity missmatch and the large number of models required to honor uncertainties.This work, proposes and investigates a hybrid methodology (HM) that combines MB and data-driven (DD) techniques focusing on improving the short-term production forecast.

Recommended citation: Vitor Hugo de Sousa Ferreira, Manuel Castro, Renato Moura, Rafael de Oliveira Werneck, Marcelo Ferreira Zampieri, Maiara Moreira Gonçalves, Oscar Linares, Soroor Salavati, Leopoldo Andre Dutra Lusquino Filho, Pedro Ribeiro Mendes Júnior, Alexandre Mello Ferreira, Alessandra Davolio, Denis José Schiozer, and Anderson Rocha. A new hybrid data-driven and model-based methodology for improved short-term production forecasting. volume Offshore Technology Conference of OTC Offshore Technology Conference, page D041S049R007, 05 2023.
DOI | Download Bibtex

Oil production and pressure multimodal forecasting integrating high-frequency production data

Published in Rio Oil & Gas 2022: Sessões Técnicas Digitais e Presenciais | Technical Sessions (Video Presentations + Technical Papers in PDF), 2022

Production forecasts (oil, gas, water) are important for the management of petroleum reservoirs. Previous results in the literature indicate the effectiveness of using machine learning models in short-term predictive analysis; however, such models are dependent on a large amount of input data. This work proposes a multimodal approach in which high-frequency and daily series are treated separately.

Recommended citation: Leopoldo André Dutra Lusquino Filho, Rafael de Oliveira Werneck, Pedro Ribeiro Mendes Júnior, Manuel Castro, Eduardo dos Santos Pereira, Renato Moura, Vítor Hugo de Sousa Ferreira, Alexandre Mello Ferreira, Alessandra Davolio Gomes, and Anderson de Rezende Rocha. Oil production and pressure multimodal forecasting integrating high-frequency production data. In IBP, editor, Rio Oil & Gas 2022: Sessões Técnicas Digitais e Presenciais | Technical Sessions (Video Presentations + Technical Papers in PDF), number 308, Rio de Janeiro | Brasil, September 2022.
DOI | Download Paper | Download Bibtex

Detecção de anomalias em dados de produção usando técnicas de aprendizado de máquina

Published in Rio Oil & Gas 2022: Sessões Técnicas Digitais e Presenciais | Technical Sessions (Video Presentations + Technical Papers in PDF), 2022

Oil production data may present anomalous behavior that does not reflect the actual reservoir dynamics. Some causes are human interventions, abrupt increase of water, severe slugging, flow instability, amongst others. This work describes the use of two unsupervised anomaly detection techniques (DBSCAN and GMM) and one supervised strategy based on recurrent neural networks underpinned by machine learning to discover observations that do not behave as expected.

Recommended citation: Aurea Rossy Soriano Vargas, Rafael de Oliveira Werneck, Maiara Moreira Gonçalves, Eduardo dos Santos Pereira, Leopoldo André Dutra Lusquino Filho, Soroor Salavati, M. Manzur Hossain, Alexandre Mello Ferreira, Alessandra Davolio Gomes, Denis José Schiozer, and Anderson de Rezende Rocha. Detecção de anomalias em dados de produção usando técnicas de aprendizado de máquina. In IBP, editor, Rio Oil & Gas 2022: Sessões Técnicas Digitais e Presenciais | Technical Sessions (Video Presentations + Technical Papers in PDF), number 298, Rio de Janeiro | Brasil, September 2022.
DOI | Download Paper | Download Bibtex

Graph-Based Early-Fusion for Flood Detection

Published in IEEE International Conference on Image Processing, 2018

Flooding is one of the most harmful natural disasters, as it poses danger to both buildings and human lives. Therefore, it is fundamental to monitor these disasters to define prevention strategies and help authorities in damage control. In this paper, we propose a fusion-based recognition system for detecting flooding events in images extracted from social media. We propose two new graph-based early-fusion methods, which consider multiple descriptions and modalities to generate an effective image representation.

Recommended citation: R. Werneck, Í. C. Dourado, S. G. Fadel, S. Tabbone, and R. da S. Torres. Graph-Based Early-Fusion for Flood Detection. In IEEE International Conference on Image Processing, pages 1048–1052, 2018.
DOI | Download Paper | Download Bibtex

Learning Cost Functions for Graph Matching

Published in Structural, Syntactic, and Statistical Pattern Recognition, 2018

During the last decade, several approaches have been proposed to address detection and recognition problems, by using graphs to represent the content of images. In this perspective, we propose an original approach to learn the matching cost functions between graphs’ nodes.

Recommended citation: Rafael Werneck, R. Raveaux, S. Tabbone, and R. da S. Torres. Learning Cost Functions for Graph Matching. In Xiao Bai, Edwin R. Hancock, Tin Kam Ho, Richard C. Wilson, Battista Biggio, and Antonio Robles-Kelly, editors, Structural, Syntactic, and Statistical Pattern Recognition, pages 345–354, Cham, 2018. Springer International Publishing.
DOI | Download Paper | Download Bibtex

Data-Driven Flood Detection using Neural Networks

Published in Working Notes Proc. MediaEval Workshop, 2017

This paper describes the approaches used by our team (MultiBrasil) for the Multimedia Satellite Task at MediaEval 2017. For both disaster image retrieval and flood-detection in satellite images, we employ neural networks for end-to-end learning.

Recommended citation: K. Nogueira, S. G. Fadel, Í. C. Dourado, R. de O. Werneck, J. A.V. Muñoz, O. A.B. Penatti, R. T. Calumby, L. T. Li, J. A. dos Santos, and R. da S. Torres. Data-Driven Flood Detection using Neural Networks. In Working Notes Proc. MediaEval Workshop, 2017.
DOI | Download Paper | Download Bibtex