Post by Erick Barboza

🌟 Exciting news! Our latest research paper "Robustness in TinyML: A Systematic Literature Review" has been accepted for publication in ACM Transactions on Embedded Computing Systems! Authored by Emanuel Pereira, Ícaro Araújo, Saulo Santos, Gabriel Andrade, Itallo Patrick, Allan Martins and myself. 👨‍🎓 This work is the first Emanuel Pereira PhD's the outcome, which I'm having the pleasure of co-advising alongside Allan Martins. 🔦 Our paper explores how TinyML—machine learning on low‑power devices—can remain reliable under real‑world constraints. 📍 Examines how robustness is assessed across TinyML studies, including dataset accessibility and evaluation metrics. 📍 Identifies major challenges such as energy limitations, noise sensitivity, and lack of standardized benchmarking frameworks. 📍 Maps current defense mechanisms and highlights opportunities for physics‑informed and secure learning approaches. 📍 Provides a consolidated view of robustness strategies to guide future research and practical deployment. 💡 This work bridges the gap between AI theory and embedded engineering, offering a foundation for scientists and engineers to design more resilient, efficient, and trustworthy TinyML systems. By clarifying how robustness is measured and achieved, it empowers innovation in edge computing, IoT, and sustainable AI—paving the way for smarter, safer, and more accessible machine learning at scale. 🔗Follow the link to access the article: https://lnkd.in/eedwmi_b