Felipe Maia Galvao Franca
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Felipe Maia Galvao Franca has a total of 309 co-authors. Here are the top co-authors by number of shared publications:
- Priscila Machado Vieira Lima
Federal University of Rio de Janeiro
25 shared publications - Nadia Nedjah
State University of Rio de Janeiro
20 shared publications Alexandre S. Nery
Department of Electrical Engineering, Universidade de Brasilia (UnB), Brasilia, Brazil
19 shared publications- Leandro Augusto Justen Marzulo
Universidade do Estado do Rio de Janeiro
12 shared publications Tiago A.O. Alves
State University of Rio de Janeiro
12 shared publications- Leandro Santiago de Araujo
Universidade Federal Fluminense (UFF)
11 shared publications
Latest Publications
- Open Access
- 3 July 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
- Vol. 23 (3) , 1-26
ET: An Energy Efficient Edge Transformer Architecture
Transformers are increasingly being deployed at the edge to support agentic AI, real-time processing, and privacy-preserving applications. However, their substantial computational and memory requirements often exceed the energy and latency constraints of edge devices. Through detailed characterization, we observe that the majority of computations and parameters in transformer models are concentrated in the feed-forward Multi-Layer Perceptron (MLP) blocks, identifying them as a key target for optimization. In this work, we introduce ET, a novel transformer architecture tailored for energy-efficient edge inference. By integrating shared Look Up Table (LUT) based feed-forward blocks with traditional self-attention modules, ET significantly reduces computational load, memory footprint, and energy usage, aligning closely with the operational and resource constraints of edge devices. We present the design, implementation, and evaluation of ET, and demonstrate that it delivers comparable accuracy to traditional transformers while achieving 2.1 × improvement over a baseline ASIC accelerator. It also achieves a 1.12 × efficiency on GPU acceleration, and about 1.34 × improvement over a prior BinaryBERT FPGA accelerator. We further report 1.2 × energy efficiency over recent multiplication-free models, such as Bitnet, showing that ET offers a compelling and practical solution for deploying transformers effectively at the edge, whether on FPGAs, ASICs or edge GPUs.
- 3 July 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
- Vol. 23 (3) , 1-26
Open AccessET: An Energy Efficient Edge Transformer Architecture
Transformers are increasingly being deployed at the edge to support agentic AI, real-time processing, and privacy-preserving applications. However, their substantial computational and memory requirements often exceed the energy and latency constraints of edge devices. Through detailed characterization, we observe that the majority of computations and parameters in transformer models are concentrated in the feed-forward Multi-Layer Perceptron (MLP) blocks, identifying them as a key target for optimization. In this work, we introduce ET, a novel transformer architecture tailored for energy-efficient edge inference. By integrating shared Look Up Table (LUT) based feed-forward blocks with traditional self-attention modules, ET significantly reduces computational load, memory footprint, and energy usage, aligning closely with the operational and resource constraints of edge devices. We present the design, implementation, and evaluation of ET, and demonstrate that it delivers comparable accuracy to traditional transformers while achieving 2.1 × improvement over a baseline ASIC accelerator. It also achieves a 1.12 × efficiency on GPU acceleration, and about 1.34 × improvement over a prior BinaryBERT FPGA accelerator. We further report 1.2 × energy efficiency over recent multiplication-free models, such as Bitnet, showing that ET offers a compelling and practical solution for deploying transformers effectively at the edge, whether on FPGAs, ASICs or edge GPUs.
- 30 June 2026
- Published by Springer Nature
- p. 201-216
Arrhythmia Classification at the Edge Using a Weightless Neural Network Hardware Accelerator
- 30 June 2026
- Published by Springer Nature
- p. 201-216
Arrhythmia Classification at the Edge Using a Weightless Neural Network Hardware Accelerator
- 25 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 34 (2) , 444-452
Weightless Neural Networks on Flexible Substrates: A Novel Approach to Wearable Machine Learning
In this article, we present a novel approach that seamlessly integrates machine learning (ML) algorithms into wearable technology through the use of weightless neural networks (WNNs) and flexible integrated circuits (FlexICs). Our methodology employs combinational intelligent networks (COIN) for edge inference on resource-constrained devices, highlighting the advantages of WNNs in terms of power efficiency and minimal hardware requirements. We propose an automated design flow for implementing COIN as FlexICs aimed at developing scalable, cost-effective, and environmentally sustainable wearable monitoring solutions. As a proof-of-concept demonstrator, an arrhythmia detection FlexIC was fabricated using COIN to meet the stringent requirements of medium-complexity wearable applications, offering a promising path toward personalized and accessible healthcare solutions.
- 25 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 34 (2) , 444-452
Weightless Neural Networks on Flexible Substrates: A Novel Approach to Wearable Machine Learning
In this article, we present a novel approach that seamlessly integrates machine learning (ML) algorithms into wearable technology through the use of weightless neural networks (WNNs) and flexible integrated circuits (FlexICs). Our methodology employs combinational intelligent networks (COIN) for edge inference on resource-constrained devices, highlighting the advantages of WNNs in terms of power efficiency and minimal hardware requirements. We propose an automated design flow for implementing COIN as FlexICs aimed at developing scalable, cost-effective, and environmentally sustainable wearable monitoring solutions. As a proof-of-concept demonstrator, an arrhythmia detection FlexIC was fabricated using COIN to meet the stringent requirements of medium-complexity wearable applications, offering a promising path toward personalized and accessible healthcare solutions.
- Open Access
- 30 June 2025
- Published by TU Delft OPEN Publishing
- Vol. 1
Performance Indicators Development for Public Services using an AI-based clustering approach
Despite the hype for AI-based generative applications, the original pitfalls in implementing AI-based technologies remain: data quality, readiness, and availability. AI results are tied to the quality of the data provided to the underlying algorithms. Inevitably, bad data produces bad results. The same happens for data readiness and availability: without correct preparation and access, even high-quality data cannot be used in AI-based solutions. To help decision-makers in the public sector, performance indicators applied to processes and, therefore, to public services can be used as tools to guide public project selection, allocation of resources, monitoring of results, etc. In this way, performance indicators can be used to standardize the evaluation of publicpolicies, helping citizens and city administrators make more informed decisions regarding the environment they live in. In the present work, we propose a performance indicator development process (PIDP) based on AI clustering techniques, which can be used to gather key performance indicators (KPI) among available data from open data sources and consumed and/or produced by the core processes that implement a public service.
- 30 June 2025
- Published by TU Delft OPEN Publishing
- Vol. 1
Open AccessPerformance Indicators Development for Public Services using an AI-based clustering approach
Despite the hype for AI-based generative applications, the original pitfalls in implementing AI-based technologies remain: data quality, readiness, and availability. AI results are tied to the quality of the data provided to the underlying algorithms. Inevitably, bad data produces bad results. The same happens for data readiness and availability: without correct preparation and access, even high-quality data cannot be used in AI-based solutions. To help decision-makers in the public sector, performance indicators applied to processes and, therefore, to public services can be used as tools to guide public project selection, allocation of resources, monitoring of results, etc. In this way, performance indicators can be used to standardize the evaluation of publicpolicies, helping citizens and city administrators make more informed decisions regarding the environment they live in. In the present work, we propose a performance indicator development process (PIDP) based on AI clustering techniques, which can be used to gather key performance indicators (KPI) among available data from open data sources and consumed and/or produced by the core processes that implement a public service.
- 24 May 2025
- Published by Elsevier in Neurocomputing
- Vol. 646
Updating KernelCanvas for weightless graph classification
- 24 May 2025
- Published by Elsevier in Neurocomputing
- Vol. 646
Updating KernelCanvas for weightless graph classification
- 12 March 2024
- Published by Springer Nature in Neural Computing and Applications
- Vol. 36 (17) , 10257-10278
Object modeling through weightless tracking
- 12 March 2024
- Published by Springer Nature in Neural Computing and Applications
- Vol. 36 (17) , 10257-10278
Object modeling through weightless tracking
- Open Access
- 1 March 2024
- Published by EDP Sciences in RAIRO - Operations Research
- Vol. 58 (2) , 1401-1428
Emissions Reporting Maturity Model: supporting cities to leverage emissions-related processes through performance indicators and artificial intelligence
Climate change and global warming have been trending topics worldwide since the Eco-92 conference. However, little progress has been made in reducing greenhouse gases (GHGs). The problems and challenges related to emissions are complex and require a concerted and comprehensive effort to address them. Emissions reporting is a critical component of GHG reduction policy and is therefore the focus of this work. It is crucial to improve the process efficiency of emissions reporting in order to achieve better emissions reduction results, as there is a direct link between effective emissions policies implemented by cities and emissions reduction (or increase) due to the effectiveness of these policies. Hence, to achieve this goal, this work proposes a series of steps to investigate, search and develop performance indicators (PIs) for emissions reporting. These performance indicators are based on the data provided by cities on the processes they go through to address emission problems. PIs can be used to guide and optimize the policies responsible for implementing emission reduction measures at the city level. Therefore, the main goal of this work is two-fold: (i) to propose an emission reporting evaluation model to leverage emissions reporting overall quality and (ii) to use artificial intelligence (AI) to support the initiatives that improve emissions reporting. Thus, this work presents an Emissions Reporting Maturity Model (ERMM) for examining, clustering, and analysing data from emissions reporting initiatives to help the cities to deal with climate change and global warming challenges. The model is built using Capability Maturity Model (CMM) concepts and uses artificial intelligence clustering technologies, performance indicator candidates and a qualitative analysis approach to find the data flow along the emissions-related processes implemented by cities. The Performance Indicator Development Process (PIDP) proposed in this work provides ways to leverage the quality of the available data necessary for the execution of the evaluations identified by the ERMM. Hence, the PIDP supports the preparation of the data from emissions-related databases, the classification of the data according to similarities highlighted by different clustering techniques, and the identification of performance indicator candidates, which are strengthened by a qualitative analysis of selected data samples. Thus, the main goal of ERRM is to evaluate and classify the cities regarding the emission reporting processes, pointing out the drawbacks and challenges faced by other cities from different contexts, and at the end to help them to leverage the underlying emissions-related processes and emissions mitigation initiatives.
- 1 March 2024
- Published by EDP Sciences in RAIRO - Operations Research
- Vol. 58 (2) , 1401-1428
Open AccessEmissions Reporting Maturity Model: supporting cities to leverage emissions-related processes through performance indicators and artificial intelligence
Climate change and global warming have been trending topics worldwide since the Eco-92 conference. However, little progress has been made in reducing greenhouse gases (GHGs). The problems and challenges related to emissions are complex and require a concerted and comprehensive effort to address them. Emissions reporting is a critical component of GHG reduction policy and is therefore the focus of this work. It is crucial to improve the process efficiency of emissions reporting in order to achieve better emissions reduction results, as there is a direct link between effective emissions policies implemented by cities and emissions reduction (or increase) due to the effectiveness of these policies. Hence, to achieve this goal, this work proposes a series of steps to investigate, search and develop performance indicators (PIs) for emissions reporting. These performance indicators are based on the data provided by cities on the processes they go through to address emission problems. PIs can be used to guide and optimize the policies responsible for implementing emission reduction measures at the city level. Therefore, the main goal of this work is two-fold: (i) to propose an emission reporting evaluation model to leverage emissions reporting overall quality and (ii) to use artificial intelligence (AI) to support the initiatives that improve emissions reporting. Thus, this work presents an Emissions Reporting Maturity Model (ERMM) for examining, clustering, and analysing data from emissions reporting initiatives to help the cities to deal with climate change and global warming challenges. The model is built using Capability Maturity Model (CMM) concepts and uses artificial intelligence clustering technologies, performance indicator candidates and a qualitative analysis approach to find the data flow along the emissions-related processes implemented by cities. The Performance Indicator Development Process (PIDP) proposed in this work provides ways to leverage the quality of the available data necessary for the execution of the evaluations identified by the ERMM. Hence, the PIDP supports the preparation of the data from emissions-related databases, the classification of the data according to similarities highlighted by different clustering techniques, and the identification of performance indicator candidates, which are strengthened by a qualitative analysis of selected data samples. Thus, the main goal of ERRM is to evaluate and classify the cities regarding the emission reporting processes, pointing out the drawbacks and challenges faced by other cities from different contexts, and at the end to help them to leverage the underlying emissions-related processes and emissions mitigation initiatives.
- Open Access
- 14 December 2023
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
- Vol. 20 (4) , 1-24
ULEEN: A Novel Architecture for Ultra-low-energy Edge Neural Networks
‘‘Extreme edge” 1 devices, such as smart sensors, are a uniquely challenging environment for the deployment of machine learning. The tiny energy budgets of these devices lie beyond what is feasible for conventional deep neural networks, particularly in high-throughput scenarios, requiring us to rethink how we approach edge inference. In this work, we propose ULEEN, a model and FPGA-based accelerator architecture based on weightless neural networks (WNNs). WNNs eliminate energy-intensive arithmetic operations, instead using table lookups to perform computation, which makes them theoretically well-suited for edge inference. However, WNNs have historically suffered from poor accuracy and excessive memory usage. ULEEN incorporates algorithmic improvements and a novel training strategy inspired by binary neural networks (BNNs) to make significant strides in addressing these issues. We compare ULEEN against BNNs in software and hardware using the four MLPerf Tiny datasets and MNIST. Our FPGA implementations of ULEEN accomplish classification at 4.0–14.3 million inferences per second, improving area-normalized throughput by an average of 3.6× and steady-state energy efficiency by an average of 7.1× compared to the FPGA-based Xilinx FINN BNN inference platform. While ULEEN is not a universally applicable machine learning model, we demonstrate that it can be an excellent choice for certain applications in energy- and latency-critical edge environments.
- 14 December 2023
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
- Vol. 20 (4) , 1-24
Open AccessULEEN: A Novel Architecture for Ultra-low-energy Edge Neural Networks
‘‘Extreme edge” 1 devices, such as smart sensors, are a uniquely challenging environment for the deployment of machine learning. The tiny energy budgets of these devices lie beyond what is feasible for conventional deep neural networks, particularly in high-throughput scenarios, requiring us to rethink how we approach edge inference. In this work, we propose ULEEN, a model and FPGA-based accelerator architecture based on weightless neural networks (WNNs). WNNs eliminate energy-intensive arithmetic operations, instead using table lookups to perform computation, which makes them theoretically well-suited for edge inference. However, WNNs have historically suffered from poor accuracy and excessive memory usage. ULEEN incorporates algorithmic improvements and a novel training strategy inspired by binary neural networks (BNNs) to make significant strides in addressing these issues. We compare ULEEN against BNNs in software and hardware using the four MLPerf Tiny datasets and MNIST. Our FPGA implementations of ULEEN accomplish classification at 4.0–14.3 million inferences per second, improving area-normalized throughput by an average of 3.6× and steady-state energy efficiency by an average of 7.1× compared to the FPGA-based Xilinx FINN BNN inference platform. While ULEEN is not a universally applicable machine learning model, we demonstrate that it can be an excellent choice for certain applications in energy- and latency-critical edge environments.
- Open Access
- 12 February 2023
- Published by Association for Computing Machinery (ACM)
- p. 232
An FPGA-Based Weightless Neural Network for Edge Network Intrusion Detection
- 12 February 2023
- Published by Association for Computing Machinery (ACM)
- p. 232
Open AccessAn FPGA-Based Weightless Neural Network for Edge Network Intrusion Detection
- Open Access
- 1 January 2023
- Published by Universite Catholique de Louvain
- p. 387-392
WiSARD-based Ensemble Learning
- 1 January 2023
- Published by Universite Catholique de Louvain
- p. 387-392
Open AccessWiSARD-based Ensemble Learning