Abstract

Deep learning models for network intrusion detection struggle with rare attack types due to class imbalance. FAA-Net introduces Focal-Aware Attention (FAIIA), which embeds imbalance awareness directly into attention computation using learnable attack-class prototypes. Achieves 94.95% recall with only 128K parameters — suitable for edge deployment.

Abstract

Current AI models for diagnosing Parkinson's from speech discard the very vocal irregularities that carry diagnostic value. We propose a signal-first approach using Burg-based restoration to preserve these biomarkers, then classify via Vision Transformers with Grad-CAM explainability for clinical trust.

Abstract

TritonZ is a remotely operated underwater rover with a manipulator arm for exploration and rescue. Integrates real-time video streaming, advanced navigation, and precise control for challenging aquatic environments.