Tag
This paper shows that an observer at the split boundary in split learning can reconstruct most of a client's text from activations and gradients—94.20% of tokens from activations alone on GPT-2, rising to 97.38% with gradients—while document-level reconstruction jumps from 13.71% to 37.77%. It argues leakage should be reported both per token and per document, and that split-model traffic deserves the same sensitivity as the raw text.
Forbes article explains split learning, a technique that enables AI models to train on sensitive data without exposing raw data, improving privacy compliance and reducing communication costs.
QSplitFL proposes a DQN-based framework for optimal split point selection in split federated learning, using client hardware metrics to adapt to heterogeneous devices. Experiments show improved convergence and accuracy across multiple datasets and architectures.