ACL: Adaptive Edge-Cloud Collaborative Learning for Heterogeneous Devices With Unlabeled Local Data

Edge-cloud collaborative learning emerges as a promising paradigm for adapting pre-trained deep neural network (DNN) models to the ever-changing edge data environments and specific downstream tasks. However, the heterogeneity of edge devices and unlabeled local data hinder the effectiveness of existing collaborative learning approaches. To address the above issues, we propose <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ACL</i>, a novel adaptive edge-cloud collaborative learning paradigm for heterogeneous devices with unlabeled local data. In <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ACL</i>, we first use <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">FedNAS</i>, a neural architecture search algorithm designed for collaborative learning to generate a customized model on each participating device, and then a lightweight semi-supervised collaborative learning framework <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">HSSCL</i> is used to fine-tune the pre-trained DNN model. Compared with the SOTA collaborative learning approaches, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ACL</i> achieves significant accuracy improvement, averaging 31.5% for image classification and 15.5% for object detection. Furthermore, it reduces time overhead by 3.1-5.1× and memory overhead by 6.3-12.5×. We will release our models and tools.