The paper "Explainable Lightweight Federeted Learning for Load Forecasting with Smart Meter Data" - Prof. Phuc et al. was presented at the recent DSBFI 2025 conference and was selected as the best paper by the peer review panel.

The paper focuses on explaining the output of a federated learning model in the context of electricity load forecasting. The combination of elements such as federated learning helps protect user data privacy, a "lightweight" model helps meet limited resource conditions, the XAI element helps understand how the model works and creates trust in the application.

On behalf of research team, Eng. Hoang-Duc Le Vu received the award.

Explainability for AI models is necessary, because for most narrow AI tasks, an AI model with an advanced architecture will in many cases provide better performance. However, as this complexity increases, the outputs become more difficult to interpret and understand. XAI (Explainable Artificial Intelligence) was created to explain the outputs of an AI model and clarify how it works. This provides trust, transparency and sustainability to the AI ​​model.
 

Congratulations again to the authors!