DP-CSGP: Differentially Private Stochastic Gradient Push with Compressed Communication
This addresses the challenge of efficient and private decentralized learning for applications requiring data privacy and reduced communication overhead, representing an incremental improvement over existing methods.
The paper tackles the problem of decentralized learning over directed graphs by proposing DP-CSGP, an algorithm that maintains high model utility while ensuring differential privacy and compressed communication, achieving a tight utility bound of O(√(d log(1/δ))/(√n J ε)) and comparable accuracy with significantly lower communication cost in experiments.
In this paper, we propose a Differentially Private Stochastic Gradient Push with Compressed communication (termed DP-CSGP) for decentralized learning over directed graphs. Different from existing works, the proposed algorithm is designed to maintain high model utility while ensuring both rigorous differential privacy (DP) guarantees and efficient communication. For general non-convex and smooth objective functions, we show that the proposed algorithm achieves a tight utility bound of $\mathcal{O}\left( \sqrt{d\log \left( \frac{1}δ \right)}/(\sqrt{n}Jε) \right)$ ($J$ and $d$ are the number of local samples and the dimension of decision variables, respectively) with $\left(ε, δ\right)$-DP guarantee for each node, matching that of decentralized counterparts with exact communication. Extensive experiments on benchmark tasks show that, under the same privacy budget, DP-CSGP achieves comparable model accuracy with significantly lower communication cost than existing decentralized counterparts with exact communication.