Abstract:The powerful deep learning models in remote sensing are inseparable from the support of massive data. However, the privacy and sensitivity of remote sensing data (RSD) restrict the possibility of each party to collaboratively train and share a large general model. Although multiparty learning (MPL) is a feasible solution, it is difficult for the existing MPL methods to uniformly process different remote sensing tasks (RSTs), and the data held by each party is nonindependent and identically distributed (non-IID), heterogeneous, and multisources. Therefore, it is urgent to explore a solution for the personalized processing of different RSTs. In this article, we formulate a novel multiparty personalized collaborative learning (MPCL) framework in terms of models and tasks. Specifically, in each iteration of the communication round, we aim to decouple personalized model optimization from global model learning. Different participants are allowed to explore their personalized local models at a certain distance from the global aggregation models according to the characteristics of their local data. In terms of task personalization, MPCL provides different personalized global models to handle the corresponding RSTs. For participants with different RSTs, it can be implemented in the multitask collaborative training strategy to explore the connection between different tasks. To demonstrate the feasibility of MPCL, we take remote sensing image (RSI) classification as a case study and provide a detailed feasibility scheme. We constructed four benchmark datasets compliant with MPL and personalized MPL, including single-source and multisource about SAR, hyperspectral, and optical RSD. The experimental results demonstrate that our MPCL is superior in these four RSDs, which ranked first in the competition with the classic or state-of-the-art (SOTA) MPL and personalized MPL algorithms. In addition, the scalability of MPCL is also verified on image segmentation RSTs of building and road extraction.

Ruyi: A Configurable and Efficient Secure Multi-Party Learning Framework with Privileged Parties

pMPL: A Robust Multi-Party Learning Framework with a Privileged Party

Multi-Party Dual Learning

MPCLeague: Robust MPC Platform for Privacy-Preserving Machine Learning

Multiparty Dual Learning

Multiparty Secure Broad Learning System for Privacy Preserving

Toward Multiparty Personalized Collaborative Learning in Remote Sensing

Private, Yet Practical, Multiparty Deep Learning.

Multi-party Secure Broad Learning System for Privacy Preserving

MPC-Pipe: an Efficient Pipeline Scheme for Secure Multi-party Machine Learning Inference

The Communication-Friendly Privacy-Preserving Machine Learning against Malicious Adversaries

Iprivjoin: an ID-Private Data Join Framework for Privacy-Preserving Machine Learning.

Private, Efficient, and Accurate: Protecting Models Trained by Multi-party Learning with Differential Privacy

SHAPER: A General Architecture for Privacy-Preserving Primitives in Secure Machine Learning.

Device-Performance-Driven Heterogeneous Multiparty Learning for Arbitrary Images

VERTICES: Efficient Two-Party Vertical Federated Linear Model with TTP-aided Secret Sharing

SSNet: A Lightweight Multi-Party Computation Scheme for Practical Privacy-Preserving Machine Learning Service in the Cloud

Low-Latency Privacy-Preserving Deep Learning Design via Secure MPC

A Mechanism Design Approach for Multi-Party Machine Learning

EFMVFL: An Efficient and Flexible Multi-party Vertical Federated Learning without a Third Party

Privacy-Preserving Federated Learning Framework Based on Chained Secure Multiparty Computing