Edge Environment

Edge environments, encompassing resource-constrained devices like smartphones and IoT sensors, are increasingly central to machine learning, aiming to perform computation and training locally to improve privacy, reduce latency, and conserve bandwidth. Current research focuses on optimizing distributed training across heterogeneous edge devices using techniques like hybrid pipeline parallelism and federated learning, often addressing challenges such as data imbalance and resource limitations through efficient algorithms and model quantization. This work is significant for enabling privacy-preserving AI applications and improving the efficiency and robustness of machine learning in resource-limited settings.