D2H-AD: New Hyperdimensional Computing Model for Anomaly Detection

A new research paper introduces D2H-AD, a hybrid model for anomaly detection that utilizes Hyperdimensional Computing (HDC). This framework distinguishes itself by combining distance-based similarity with density-aware encoding, leading to improved anomaly representation and detection capabilities. The D2H-AD model has demonstrated superior performance compared to several established anomaly detection baselines across various datasets. The framework is notable for its lightweight nature, interpretability, and computational efficiency, making it particularly well-suited for environments with limited resources, such as TinyML and edge AI deployments. Its design allows for low-latency operations and a small memory footprint, addressing common challenges in real-time anomaly detection. The findings suggest that D2H-AD offers a robust, accurate, and energy-efficient solution for identifying anomalies in dynamic systems.