Pesquisa da UNIFOR analisa a robustez de detectores de fraude baseados em Graph Neural Networks (GNNs) frente ao drift de dados em knowledge graphs financeiros. Estudo demonstra que arquiteturas mais simples como R-GCN são mais resilientes à degradação tempora…
This paper presents a validated IoT monitoring system for photovoltaic plants using LoRa communication, demonstrating technical feasibility with 95.49% packet delivery rates and measurement accuracy confirmed by official meteorological data. The system success…
This paper from UFC evaluates emerging aerobic biotechnologies (MBBR, IFAS, AGS, AGS-BF) for sewage treatment and resource recovery. The research demonstrates that AGS and AGS-BF systems significantly outperform traditional methods in nitrogen removal (>70%) a…