Hermes Senger
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Hermes Senger has a total of 90 co-authors. Here are the top co-authors by number of shared publications:
- Fabricio Alves Barbosa da Silva
Scientific Computing Program, Oswaldo Cruz Foundation (Fiocruz), Rio de Janeiro, Brazil
16 shared publications - Veronica Gil-Costa
Facultad de Ciencias Físico Matemáticas y Naturales, Universidad Nacional de San Luis, San Luis 5700, Argentina
7 shared publications Cesar A. C. Marcondes
Aeronautics Institute of Technology – ITA
5 shared publications- Edson Satoshi Gomi
Universidade de São Paulo
4 shared publications - Eduardo Raul Hruschka
Department of Computer Science, University of Sao Paulo, Sao Carlos, Brazil
3 shared publications - Guilherme Weigert Cassales
School of Mathematical and Computational Sciences, Massey University, Auckland, New Zealand
3 shared publications
Latest Publications
- 21 October 2025
- Published by Springer Nature in Machine Learning
- Vol. 114 (11) , 1-31
Pdarts: projected differentiable architecture search for seismic inversion
- 21 October 2025
- Published by Springer Nature in Machine Learning
- Vol. 114 (11) , 1-31
Pdarts: projected differentiable architecture search for seismic inversion
- Open Access
- 1 September 2025
- Published by Elsevier in Applied Computing and Geosciences
- Vol. 27
Simwave: A finite difference simulator for acoustic waves propagation
- 1 September 2025
- Published by Elsevier in Applied Computing and Geosciences
- Vol. 27
Open AccessSimwave: A finite difference simulator for acoustic waves propagation
- Open Access
- 15 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 72953-72983
A Survey of Data Stream-Based Intrusion Detection Systems
Detecting malicious activities in network environments poses a challenge that attracts significant attention due to its complexity and importance. Advances in the field have led to the development of several algorithms that approach the problem under the view of a data stream machine learning task. This task involves a set of steps: data collection or choice of public datasets, data pre-processing, data reduction, development or application of data mining techniques, and evaluation methodology. However, these steps must address the inherent issues of dynamic environments such as data streams and intrusion detection systems. These issues include, but are not limited to, the continuous influx of data, changes in both normal and attack class distributions, the emergence of new attack types, and the scarcity of labeled data examples to update the decision models. This survey provides an overview of intrusion detection systems (IDS) using data stream machine learning techniques, characterizing the literature approaches according to the classic steps of the data mining task. In addition, we discuss recommendations for practical IDS development and highlight datasets and tools that can aid in detecting malicious behavior. Finally, we outline potential avenues for future research and open questions in the field.
- 15 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 72953-72983
Open AccessA Survey of Data Stream-Based Intrusion Detection Systems
Detecting malicious activities in network environments poses a challenge that attracts significant attention due to its complexity and importance. Advances in the field have led to the development of several algorithms that approach the problem under the view of a data stream machine learning task. This task involves a set of steps: data collection or choice of public datasets, data pre-processing, data reduction, development or application of data mining techniques, and evaluation methodology. However, these steps must address the inherent issues of dynamic environments such as data streams and intrusion detection systems. These issues include, but are not limited to, the continuous influx of data, changes in both normal and attack class distributions, the emergence of new attack types, and the scarcity of labeled data examples to update the decision models. This survey provides an overview of intrusion detection systems (IDS) using data stream machine learning techniques, characterizing the literature approaches according to the classic steps of the data mining task. In addition, we discuss recommendations for practical IDS development and highlight datasets and tools that can aid in detecting malicious behavior. Finally, we outline potential avenues for future research and open questions in the field.
- Open Access
- 7 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 51-60
Mini-batching with Fused Training and Testing for Data Streams Processing on the Edge
Edge Computing (EC) has emerged as a solution to reduce energy demand and greenhouse gas emissions from digital technologies. EC supports low latency, mobility, and location awareness for delay-sensitive applications by bridging the gap between cloud computing services and end-users. Machine learning (ML) methods have been applied in EC for data classification and information processing. Ensemble learners have often proven to yield high predictive performance on data stream classification problems. Mini-batching is a technique proposed for improving cache reuse in multi-core architectures of bagging ensembles for the classification of online data streams, which benefits application speedup and reduces energy consumption. However, the original mini-batching presents limited benefits in terms of cache reuse and it hinders the accuracy of the ensembles (i.e., their capacity to detect behavior changes in data streams). In this paper, we improve mini-batching by fusing continuous training and test loops for the classification of data streams. We evaluated the new strategy by comparing its performance and energy efficiency with the original mini-batching for data stream classification using six ensemble algorithms and four benchmark datasets. We also compare mini-batching strategies with two hardware-based strategies supported by commodity multi-core processors commonly used in EC. Results show that mini-batching strategies can significantly reduce energy consumption in 95% of the experiments. Mini-batching improved energy efficiency by 96% on average and 169% in the best case. Likewise, our new mini-batching strategy improved energy efficiency by 136% on average and 456% in the best case. These strategies also support better control of the balance between performance, energy efficiency, and accuracy.
- 7 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 51-60
Open AccessMini-batching with Fused Training and Testing for Data Streams Processing on the Edge
Edge Computing (EC) has emerged as a solution to reduce energy demand and greenhouse gas emissions from digital technologies. EC supports low latency, mobility, and location awareness for delay-sensitive applications by bridging the gap between cloud computing services and end-users. Machine learning (ML) methods have been applied in EC for data classification and information processing. Ensemble learners have often proven to yield high predictive performance on data stream classification problems. Mini-batching is a technique proposed for improving cache reuse in multi-core architectures of bagging ensembles for the classification of online data streams, which benefits application speedup and reduces energy consumption. However, the original mini-batching presents limited benefits in terms of cache reuse and it hinders the accuracy of the ensembles (i.e., their capacity to detect behavior changes in data streams). In this paper, we improve mini-batching by fusing continuous training and test loops for the classification of data streams. We evaluated the new strategy by comparing its performance and energy efficiency with the original mini-batching for data stream classification using six ensemble algorithms and four benchmark datasets. We also compare mini-batching strategies with two hardware-based strategies supported by commodity multi-core processors commonly used in EC. Results show that mini-batching strategies can significantly reduce energy consumption in 95% of the experiments. Mini-batching improved energy efficiency by 96% on average and 169% in the best case. Likewise, our new mini-batching strategy improved energy efficiency by 136% on average and 456% in the best case. These strategies also support better control of the balance between performance, energy efficiency, and accuracy.
- Open Access
- 17 October 2023
- Published by Sociedade Brasileira de Computação
- p. 109-120
Data mapping strategies for multi-GPU implementation of a seismic application
Seismic imaging applications are computationally costly, and the industry’s demand is continuously increasing due to the availability of better data, larger data, and the need for better resolution images. It means that the computational capacity needed tends to increase both in terms of FLOPS calculation and memory. Nowadays, many HPC clusters have nodes with multiple GPUs (e.g., 2, 4, and 8). In this paper, we investigate mechanisms and strategies for the data exchange (of the halo zones) of a finite differences grid of a wave simulator implemented in OpenMP. We compare the performance and programming effort of four data mapping mechanisms supported by OpenMP and CUDA. Our best strategy has achieved speedups of 3.87 on four V100 GPUs with NVLink.
- 17 October 2023
- Published by Sociedade Brasileira de Computação
- p. 109-120
Open AccessData mapping strategies for multi-GPU implementation of a seismic application
Seismic imaging applications are computationally costly, and the industry’s demand is continuously increasing due to the availability of better data, larger data, and the need for better resolution images. It means that the computational capacity needed tends to increase both in terms of FLOPS calculation and memory. Nowadays, many HPC clusters have nodes with multiple GPUs (e.g., 2, 4, and 8). In this paper, we investigate mechanisms and strategies for the data exchange (of the halo zones) of a finite differences grid of a wave simulator implemented in OpenMP. We compare the performance and programming effort of four data mapping mechanisms supported by OpenMP and CUDA. Our best strategy has achieved speedups of 3.87 on four V100 GPUs with NVLink.
- Open Access
- 5 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network and Service Management
- Vol. 20 (3) , 3038-3054
Balancing Performance and Energy Consumption of Bagging Ensembles for the Classification of Data Streams in Edge Computing
In recent years, the Edge Computing (EC) paradigm has emerged as an enabling factor for developing technologies like the Internet of Things (IoT) and 5G networks, bridging the gap between Cloud Computing services and end-users, supporting low latency, mobility, and location awareness to delay-sensitive applications. An increasing number of solutions in EC have employed machine learning (ML) methods to perform data classification and other information processing tasks on continuous and evolving data streams. Usually, such solutions have to cope with vast amounts of data that come as data streams while balancing energy consumption, latency, and the predictive performance of the algorithms. Ensemble methods achieve remarkable predictive performance when applied to evolving data streams due to several models and the possibility of selective resets. This work investigates a strategy that introduces short intervals to defer the processing of mini-batches. Well balanced, our strategy can improve the performance (i.e., delay, throughput) and reduce the energy consumption of bagging ensembles to classify data streams. The experimental evaluation involved six state-of-art ensemble algorithms (OzaBag, OzaBag Adaptive Size Hoeffding Tree, Online Bagging ADWIN, Leveraging Bagging, Adaptive RandomForest, and Streaming Random Patches) applying five widely used machine learning benchmark datasets with varied characteristics on three computer platforms. As a result, our strategy can significantly reduce energy consumption in 96% of the experimental scenarios evaluated. Despite the trade-offs, it is possible to balance them to avoid significant loss in predictive performance.
- 5 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network and Service Management
- Vol. 20 (3) , 3038-3054
Open AccessBalancing Performance and Energy Consumption of Bagging Ensembles for the Classification of Data Streams in Edge Computing
In recent years, the Edge Computing (EC) paradigm has emerged as an enabling factor for developing technologies like the Internet of Things (IoT) and 5G networks, bridging the gap between Cloud Computing services and end-users, supporting low latency, mobility, and location awareness to delay-sensitive applications. An increasing number of solutions in EC have employed machine learning (ML) methods to perform data classification and other information processing tasks on continuous and evolving data streams. Usually, such solutions have to cope with vast amounts of data that come as data streams while balancing energy consumption, latency, and the predictive performance of the algorithms. Ensemble methods achieve remarkable predictive performance when applied to evolving data streams due to several models and the possibility of selective resets. This work investigates a strategy that introduces short intervals to defer the processing of mini-batches. Well balanced, our strategy can improve the performance (i.e., delay, throughput) and reduce the energy consumption of bagging ensembles to classify data streams. The experimental evaluation involved six state-of-art ensemble algorithms (OzaBag, OzaBag Adaptive Size Hoeffding Tree, Online Bagging ADWIN, Leveraging Bagging, Adaptive RandomForest, and Streaming Random Patches) applying five widely used machine learning benchmark datasets with varied characteristics on three computer platforms. As a result, our strategy can significantly reduce energy consumption in 96% of the experimental scenarios evaluated. Despite the trade-offs, it is possible to balance them to avoid significant loss in predictive performance.
- 7 April 2022
- Published by Sociedade Brasileira de Computação
- p. 21-24
Seismic Wave Stencil Computation Performance Comparison
This work aims to provide comparative results between OpenCL and oneAPI performance on FPGAs, GPUs and CPUs regarding the execution of stencils in the simulation of seismic acoustic wave propagation. Our results show that the naive stencil code performs better on GPU and CPU than FPGA without strong knowledge on how to implement optimizations. Another result is that oneAPI code is easier and faster to implement than OpenCL as it automatically applies some FPGA-specific optimizations at compile time.
- 7 April 2022
- Published by Sociedade Brasileira de Computação
- p. 21-24
Seismic Wave Stencil Computation Performance Comparison
This work aims to provide comparative results between OpenCL and oneAPI performance on FPGAs, GPUs and CPUs regarding the execution of stencils in the simulation of seismic acoustic wave propagation. Our results show that the naive stencil code performs better on GPU and CPU than FPGA without strong knowledge on how to implement optimizations. Another result is that oneAPI code is easier and faster to implement than OpenCL as it automatically applies some FPGA-specific optimizations at compile time.
- 11 September 2021
- Published by Springer Nature
- p. 201-217
Non-cooperative Vehicular Density Prediction in VANETs
- 11 September 2021
- Published by Springer Nature
- p. 201-217
Non-cooperative Vehicular Density Prediction in VANETs
- 11 September 2021
- Published by Springer Nature
- p. 471-486
Distributed Novelty Detection at the Edge for IoT Network Security
- 11 September 2021
- Published by Springer Nature
- p. 471-486
Distributed Novelty Detection at the Edge for IoT Network Security
- Open Access
- 6 August 2020
- Published by Springer Nature in Scientific Reports
- Vol. 10 (1) , 1-15
Characterization of a SPM-1 metallo-beta-lactamase-producing Pseudomonas aeruginosa by comparative genomics and phenotypic analysis
Pseudomonas aeruginosa is one of the most common pathogens related to healthcare-associated infections. The Brazilian isolate, named CCBH4851, is a multidrug-resistant clone belonging to the sequence type 277. The antimicrobial resistance mechanisms of the CCBH4851 strain are associated with the presence of the bla\(_\text {SPM-1}\) gene, encoding a metallo-beta-lactamase, in combination with other exogenously acquired genes. Whole-genome sequencing studies focusing on emerging pathogens are essential to identify key features of their physiology that may lead to the identification of new targets for therapy. Using both Illumina and PacBio sequencing data, we obtained a single contig representing the CCBH4851 genome with annotated features that were consistent with data reported for the species. However, comparative analysis with other Pseudomonas aeruginosa strains revealed genomic differences regarding virulence factors and regulatory proteins. In addition, we performed phenotypic assays that revealed CCBH4851 is impaired in bacterial motilities and biofilm formation. On the other hand, CCBH4851 genome contained acquired genomic islands that carry transcriptional factors, virulence and antimicrobial resistance-related genes. Presence of single nucleotide polymorphisms in the core genome, mainly those located in resistance-associated genes, suggests that these mutations may also influence the multidrug-resistant behavior of CCBH4851. Overall, characterization of Pseudomonas aeruginosa CCBH4851 complete genome revealed the presence of features that strongly relates to the virulence and antibiotic resistance profile of this important infectious agent.- 6 August 2020
- Published by Springer Nature in Scientific Reports
- Vol. 10 (1) , 1-15
Open AccessCharacterization of a SPM-1 metallo-beta-lactamase-producing Pseudomonas aeruginosa by comparative genomics and phenotypic analysis
Pseudomonas aeruginosa is one of the most common pathogens related to healthcare-associated infections. The Brazilian isolate, named CCBH4851, is a multidrug-resistant clone belonging to the sequence type 277. The antimicrobial resistance mechanisms of the CCBH4851 strain are associated with the presence of the bla\(_\text {SPM-1}\) gene, encoding a metallo-beta-lactamase, in combination with other exogenously acquired genes. Whole-genome sequencing studies focusing on emerging pathogens are essential to identify key features of their physiology that may lead to the identification of new targets for therapy. Using both Illumina and PacBio sequencing data, we obtained a single contig representing the CCBH4851 genome with annotated features that were consistent with data reported for the species. However, comparative analysis with other Pseudomonas aeruginosa strains revealed genomic differences regarding virulence factors and regulatory proteins. In addition, we performed phenotypic assays that revealed CCBH4851 is impaired in bacterial motilities and biofilm formation. On the other hand, CCBH4851 genome contained acquired genomic islands that carry transcriptional factors, virulence and antimicrobial resistance-related genes. Presence of single nucleotide polymorphisms in the core genome, mainly those located in resistance-associated genes, suggests that these mutations may also influence the multidrug-resistant behavior of CCBH4851. Overall, characterization of Pseudomonas aeruginosa CCBH4851 complete genome revealed the presence of features that strongly relates to the virulence and antibiotic resistance profile of this important infectious agent.