Rafael Schild Reusch Receives Prestigious Award at SBESC 2023

We closed the year celebrating the achievements of our team members. DBServant Rafael Schild Reusch received the award for third-best paper at the Brazilian Symposium on Computer Systems Engineering (SBESC 2023). Rafael earned this recognition for his paper, “Application of Machine Learning in Energy- and Memory-Constrained Devices for Human Activity Recognition,” based on his master’s thesis submitted at PUCRS in November 2023.

Rafael is part of our team, working as a senior software developer for one of our clients in the avionics sector. His dedication and expertise highlight not only his technical capabilities but also his ability to apply innovation in real-world scenarios.

Overcoming Challenges: Artificial Intelligence in Resource-Constrained Devices

Rafael’s research addresses a critical challenge in artificial intelligence: how to efficiently deploy advanced models on devices with strict energy and memory constraints. As AI models become more accurate and complex, significant challenges emerge when implementing them in low-resource systems such as smartwatches and IoT devices.

In the context of Human Activity Recognition (HAR), Rafael highlights the prevalence of complex approaches, especially those based on LSTM networks. While effective for sensor data, these models often require more computational resources than low-power IoT devices can support, making them impractical in many scenarios.

Key Contribution: A New Perspective for Embedded Systems

Rafael proposes an innovative approach by developing a Convolutional Neural Network (CNN) optimized for embedded systems with limited resources. Using advanced optimization techniques and efficient neural architecture design, he was able to reduce the reference model size by 2.34x while improving accuracy from 74% to an impressive 85.2%.

Read the Full Paper

To fully understand the depth and impact of Rafael’s work, we invite readers to explore the full paper. This research not only advances the application of artificial intelligence in constrained environments but also showcases Rafael’s ability to tackle complex engineering challenges.

Click here to access the full paper

We thank DBServant Rafael for his commitment and achievement, and we continue to drive innovation together. This is just one example of what our team is capable of achieving, and we look forward to new challenges and new milestones ahead. Congratulations, Rafael Schild Reusch, on this well-deserved recognition!