Abstract: Split learning (SL) is widely regarded as a promising distributed machine learning framework with superior privacy-preserving properties, lower communication and computation costs. However, ...