Edge Model

Edge models are optimized deep learning models deployed on resource-constrained devices, aiming to balance computational efficiency with accuracy for real-time applications. Current research focuses on developing secure and efficient edge model architectures (e.g., CNN-Transformer hybrids, Spiking Neural Networks), robust training and adaptation methods (including federated learning and continuous learning), and techniques to mitigate discrepancies between edge and cloud-based models. This field is crucial for advancing AI's deployment in diverse applications, particularly those requiring low latency and privacy preservation, such as mobile robotics, IoT devices, and personalized healthcare.