HydroLens is a FL-based underwater LiDAR image classification system that ensures adversarial robustness and privacy for lightweight navigation systems with limited computational resources. The system enables on-site training, reducing data transfer and bandwidth requirements, while an on-land base station aggregates results. Privacy-preserving mechanisms protect sensitive data, and adversarial simulations, such as data poisoning attacks, validate the system’s resilience against noisy or corrupted underwater data.
Accessibility Controls