Mateus Roder
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Mateus Roder has a total of 49 co-authors. Here are the top co-authors by number of shared publications:
- João Paulo Papa
São Paulo State University (UNESP)
18 shared publications - Gustavo Henrique de Rosa
Sao Paulo State University
12 shared publications - Leandro Aparecido Passos
São Paulo State University (UNESP)
12 shared publications - André Luis Debiaso Rossi
Sao Paulo State University
4 shared publications João Paulo Papa
3 shared publications- Luiz Carlos Felix Ribeiro
Sao Paulo State University
3 shared publications
Latest Publications
- Open Access
- 15 September 2026
- Published by Elsevier in SSRN Electronic Journal
An unsupervised workflow for pre-annotation of borehole breakout from acoustic image logs
Borehole image (BHI) logs are extensively used for the characterization of fractures, borehole breakouts, and other geological and geomechanical features. Recent advances in supervised machine learning have demonstrated promising results for automated BHI interpretation; however, these approaches remain highly dependent on large manually annotated datasets, whose generation is labor-intensive, time-consuming, and often difficult to standardize. This study presents an interpretable, unsupervised pre-annotation workflow designed to accelerate the creation of training datasets for breakout detection in acoustic borehole image logs. Rather than treating breakout recognition as a conventional image classification problem, the proposed approach combines unsupervised image segmentation with azimuthal signal parameterization to identify spatially coherent breakout candidates. Low-amplitude acoustic textures are first segmented using an unsupervised clustering workflow, after which their circumferential organization is summarized into a one-dimensional azimuthal signal describing the vertical persistence and lateral concentration of segmented features. Candidate breakout regions are subsequently identified through a series of geologically interpretable signal- and geometry-based validation criteria before being projected back onto the original borehole image as pre-annotations. The workflow was tested on 757 acoustic BHI image segments from six wells encompassing a broad range of carbonate reservoir conditions, including intervals with complex dissolution features and natural fractures. Across all evaluated datasets, 91.3% of the generated annotations were classified by expert human visual inspection as directly usable or requiring only minor manual refinement, substantially reducing the amount of manual labeling required for supervised learning workflows. Error analysis further demonstrated that the dominant failure modes are systematic and geologically interpretable, primarily involving artifact leakage, oversegmentation, and missed breakouts in highly heterogeneous intervals. The proposed workflow provides a computationally lightweight and fully interpretable alternative for generating high-quality breakout pre-annotations while preserving expert supervision, thereby facilitating the development of larger labeled datasets for future supervised BHI interpretation.
- 15 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessAn unsupervised workflow for pre-annotation of borehole breakout from acoustic image logs
Borehole image (BHI) logs are extensively used for the characterization of fractures, borehole breakouts, and other geological and geomechanical features. Recent advances in supervised machine learning have demonstrated promising results for automated BHI interpretation; however, these approaches remain highly dependent on large manually annotated datasets, whose generation is labor-intensive, time-consuming, and often difficult to standardize. This study presents an interpretable, unsupervised pre-annotation workflow designed to accelerate the creation of training datasets for breakout detection in acoustic borehole image logs. Rather than treating breakout recognition as a conventional image classification problem, the proposed approach combines unsupervised image segmentation with azimuthal signal parameterization to identify spatially coherent breakout candidates. Low-amplitude acoustic textures are first segmented using an unsupervised clustering workflow, after which their circumferential organization is summarized into a one-dimensional azimuthal signal describing the vertical persistence and lateral concentration of segmented features. Candidate breakout regions are subsequently identified through a series of geologically interpretable signal- and geometry-based validation criteria before being projected back onto the original borehole image as pre-annotations. The workflow was tested on 757 acoustic BHI image segments from six wells encompassing a broad range of carbonate reservoir conditions, including intervals with complex dissolution features and natural fractures. Across all evaluated datasets, 91.3% of the generated annotations were classified by expert human visual inspection as directly usable or requiring only minor manual refinement, substantially reducing the amount of manual labeling required for supervised learning workflows. Error analysis further demonstrated that the dominant failure modes are systematic and geologically interpretable, primarily involving artifact leakage, oversegmentation, and missed breakouts in highly heterogeneous intervals. The proposed workflow provides a computationally lightweight and fully interpretable alternative for generating high-quality breakout pre-annotations while preserving expert supervision, thereby facilitating the development of larger labeled datasets for future supervised BHI interpretation.
- Open Access
- 31 October 2025
- Published by Elsevier in Applied Computing and Geosciences
- Vol. 28
Vision transformers as SOTA models for lithological classification of Brazilian pre-salt rocks
- 31 October 2025
- Published by Elsevier in Applied Computing and Geosciences
- Vol. 28
Open AccessVision transformers as SOTA models for lithological classification of Brazilian pre-salt rocks
- Open Access
Adapting LLMs for the Medical Domain in Portuguese: A Study on Fine-Tuning and Model Evaluation
This study evaluates the performance of large language models (LLMs) as medical agents in Portuguese, aiming to develop a reliable and relevant virtual assistant for healthcare professionals. The HealthCareMagic-100k-en and MedQuAD datasets, translated from English using GPT-3.5, were used to fine-tune the ChatBode-7B model using the PEFT-QLoRA method. The InternLM2 model, with initial training on medical data, presented the best overall performance, with high precision and adequacy in metrics such as accuracy, completeness and safety. However, DrBode models, derived from ChatBode, exhibited a phenomenon of catastrophic forgetting of acquired medical knowledge. Despite this, these models performed frequently or even better in aspects such as grammaticality and coherence. A significant challenge was low inter-rater agreement, highlighting the need for more robust assessment protocols. This work paves the way for future research, such as evaluating multilingual models specific to the medical field, improving the quality of training data, and developing more consistent evaluation methodologies for the medical field.
Open AccessAdapting LLMs for the Medical Domain in Portuguese: A Study on Fine-Tuning and Model Evaluation
This study evaluates the performance of large language models (LLMs) as medical agents in Portuguese, aiming to develop a reliable and relevant virtual assistant for healthcare professionals. The HealthCareMagic-100k-en and MedQuAD datasets, translated from English using GPT-3.5, were used to fine-tune the ChatBode-7B model using the PEFT-QLoRA method. The InternLM2 model, with initial training on medical data, presented the best overall performance, with high precision and adequacy in metrics such as accuracy, completeness and safety. However, DrBode models, derived from ChatBode, exhibited a phenomenon of catastrophic forgetting of acquired medical knowledge. Despite this, these models performed frequently or even better in aspects such as grammaticality and coherence. A significant challenge was low inter-rater agreement, highlighting the need for more robust assessment protocols. This work paves the way for future research, such as evaluating multilingual models specific to the medical field, improving the quality of training data, and developing more consistent evaluation methodologies for the medical field.
- Open Access
Feature Selection and Hyperparameter Fine-tuning in Artificial Neural Networks for Wood Quality Classification
Quality classification of wood boards is an essential task in the sawmill industry, which is still usually performed by human operators in small to median companies in developing countries. Machine learning algorithms have been successfully employed to investigate the problem, offering a more affordable alternative compared to other solutions. However, such approaches usually present some drawbacks regarding the proper selection of their hyperparameters. Moreover, the models are susceptible to the features extracted from wood board images, which influence the induction of the model and, consequently, its generalization power. Therefore, in this paper, we investigate the problem of simultaneously tuning the hyperparameters of an artificial neural network (ANN) as well as selecting a subset of characteristics that better describes the wood board quality. Experiments were conducted over a private dataset composed of images obtained from a sawmill industry and described using different feature descriptors. The predictive performance of the model was compared against five baseline methods as well as a random search, performing either ANN hyperparameter tuning and feature selection. Experimental results suggest that hyperparameters should be adjusted according to the feature set, or the features should be selected considering the hyperparameter values. In summary, the best predictive performance, i.e., a balanced accuracy of $0.80$, was achieved in two distinct scenarios: (i) performing only feature selection, and (ii) performing both tasks concomitantly. Thus, we suggest that at least one of the two approaches should be considered in the context of industrial applications.Open AccessFeature Selection and Hyperparameter Fine-tuning in Artificial Neural Networks for Wood Quality Classification
Quality classification of wood boards is an essential task in the sawmill industry, which is still usually performed by human operators in small to median companies in developing countries. Machine learning algorithms have been successfully employed to investigate the problem, offering a more affordable alternative compared to other solutions. However, such approaches usually present some drawbacks regarding the proper selection of their hyperparameters. Moreover, the models are susceptible to the features extracted from wood board images, which influence the induction of the model and, consequently, its generalization power. Therefore, in this paper, we investigate the problem of simultaneously tuning the hyperparameters of an artificial neural network (ANN) as well as selecting a subset of characteristics that better describes the wood board quality. Experiments were conducted over a private dataset composed of images obtained from a sawmill industry and described using different feature descriptors. The predictive performance of the model was compared against five baseline methods as well as a random search, performing either ANN hyperparameter tuning and feature selection. Experimental results suggest that hyperparameters should be adjusted according to the feature set, or the features should be selected considering the hyperparameter values. In summary, the best predictive performance, i.e., a balanced accuracy of $0.80$, was achieved in two distinct scenarios: (i) performing only feature selection, and (ii) performing both tasks concomitantly. Thus, we suggest that at least one of the two approaches should be considered in the context of industrial applications. - 12 October 2023
- Published by Springer Nature
- p. 323-337
Feature Selection and Hyperparameter Fine-Tuning in Artificial Neural Networks for Wood Quality Classification
- 12 October 2023
- Published by Springer Nature
- p. 323-337
Feature Selection and Hyperparameter Fine-Tuning in Artificial Neural Networks for Wood Quality Classification
- 22 August 2023
- Published by Springer Nature
- p. 85-104
Metaheuristics for Feature Selection: A Comprehensive Comparison Using Opytimizer
- 22 August 2023
- Published by Springer Nature
- p. 85-104
Metaheuristics for Feature Selection: A Comprehensive Comparison Using Opytimizer
- Open Access
Facial Point Graphs for Amyotrophic Lateral Sclerosis Identification
Identifying Amyotrophic Lateral Sclerosis (ALS) in its early stages is essential for establishing the beginning of treatment, enriching the outlook, and enhancing the overall well-being of those affected individuals. However, early diagnosis and detecting the disease's signs is not straightforward. A simpler and cheaper way arises by analyzing the patient's facial expressions through computational methods. When a patient with ALS engages in specific actions, e.g., opening their mouth, the movement of specific facial muscles differs from that observed in a healthy individual. This paper proposes Facial Point Graphs to learn information from the geometry of facial images to identify ALS automatically. The experimental outcomes in the Toronto Neuroface dataset show the proposed approach outperformed state-of-the-art results, fostering promising developments in the area.
Open AccessFacial Point Graphs for Amyotrophic Lateral Sclerosis Identification
Identifying Amyotrophic Lateral Sclerosis (ALS) in its early stages is essential for establishing the beginning of treatment, enriching the outlook, and enhancing the overall well-being of those affected individuals. However, early diagnosis and detecting the disease's signs is not straightforward. A simpler and cheaper way arises by analyzing the patient's facial expressions through computational methods. When a patient with ALS engages in specific actions, e.g., opening their mouth, the movement of specific facial muscles differs from that observed in a healthy individual. This paper proposes Facial Point Graphs to learn information from the geometry of facial images to identify ALS automatically. The experimental outcomes in the Toronto Neuroface dataset show the proposed approach outperformed state-of-the-art results, fostering promising developments in the area.
- Open Access
Improving Pre-Trained Weights Through Meta-Heuristics Fine-Tuning
Machine Learning algorithms have been extensively researched throughout the last decade, leading to unprecedented advances in a broad range of applications, such as image classification and reconstruction, object recognition, and text categorization. Nonetheless, most Machine Learning algorithms are trained via derivative-based optimizers, such as the Stochastic Gradient Descent, leading to possible local optimum entrapments and inhibiting them from achieving proper performances. A bio-inspired alternative to traditional optimization techniques, denoted as meta-heuristic, has received significant attention due to its simplicity and ability to avoid local optimums imprisonment. In this work, we propose to use meta-heuristic techniques to fine-tune pre-trained weights, exploring additional regions of the search space, and improving their effectiveness. The experimental evaluation comprises two classification tasks (image and text) and is assessed under four literature datasets. Experimental results show nature-inspired algorithms' capacity in exploring the neighborhood of pre-trained weights, achieving superior results than their counterpart pre-trained architectures. Additionally, a thorough analysis of distinct architectures, such as Multi-Layer Perceptron and Recurrent Neural Networks, attempts to visualize and provide more precise insights into the most critical weights to be fine-tuned in the learning process.
Open AccessImproving Pre-Trained Weights Through Meta-Heuristics Fine-Tuning
Machine Learning algorithms have been extensively researched throughout the last decade, leading to unprecedented advances in a broad range of applications, such as image classification and reconstruction, object recognition, and text categorization. Nonetheless, most Machine Learning algorithms are trained via derivative-based optimizers, such as the Stochastic Gradient Descent, leading to possible local optimum entrapments and inhibiting them from achieving proper performances. A bio-inspired alternative to traditional optimization techniques, denoted as meta-heuristic, has received significant attention due to its simplicity and ability to avoid local optimums imprisonment. In this work, we propose to use meta-heuristic techniques to fine-tune pre-trained weights, exploring additional regions of the search space, and improving their effectiveness. The experimental evaluation comprises two classification tasks (image and text) and is assessed under four literature datasets. Experimental results show nature-inspired algorithms' capacity in exploring the neighborhood of pre-trained weights, achieving superior results than their counterpart pre-trained architectures. Additionally, a thorough analysis of distinct architectures, such as Multi-Layer Perceptron and Recurrent Neural Networks, attempts to visualize and provide more precise insights into the most critical weights to be fine-tuned in the learning process.
- Open Access
From Actions to Events: A Transfer Learning Approach Using Improved Deep Belief Networks
In the last decade, exponential data growth supplied machine learning-based algorithms' capacity and enabled their usage in daily-life activities. Additionally, such an improvement is partially explained due to the advent of deep learning techniques, i.e., stacks of simple architectures that end up in more complex models. Although both factors produce outstanding results, they also pose drawbacks regarding the learning process as training complex models over large datasets are expensive and time-consuming. Such a problem is even more evident when dealing with video analysis. Some works have considered transfer learning or domain adaptation, i.e., approaches that map the knowledge from one domain to another, to ease the training burden, yet most of them operate over individual or small blocks of frames. This paper proposes a novel approach to map the knowledge from action recognition to event recognition using an energy-based model, denoted as Spectral Deep Belief Network. Such a model can process all frames simultaneously, carrying spatial and temporal information through the learning process. The experimental results conducted over two public video dataset, the HMDB-51 and the UCF-101, depict the effectiveness of the proposed model and its reduced computational burden when compared to traditional energy-based models, such as Restricted Boltzmann Machines and Deep Belief Networks.
Open AccessFrom Actions to Events: A Transfer Learning Approach Using Improved Deep Belief Networks
In the last decade, exponential data growth supplied machine learning-based algorithms' capacity and enabled their usage in daily-life activities. Additionally, such an improvement is partially explained due to the advent of deep learning techniques, i.e., stacks of simple architectures that end up in more complex models. Although both factors produce outstanding results, they also pose drawbacks regarding the learning process as training complex models over large datasets are expensive and time-consuming. Such a problem is even more evident when dealing with video analysis. Some works have considered transfer learning or domain adaptation, i.e., approaches that map the knowledge from one domain to another, to ease the training burden, yet most of them operate over individual or small blocks of frames. This paper proposes a novel approach to map the knowledge from action recognition to event recognition using an energy-based model, denoted as Spectral Deep Belief Network. Such a model can process all frames simultaneously, carrying spatial and temporal information through the learning process. The experimental results conducted over two public video dataset, the HMDB-51 and the UCF-101, depict the effectiveness of the proposed model and its reduced computational burden when compared to traditional energy-based models, such as Restricted Boltzmann Machines and Deep Belief Networks.