The development and application of federated machine learning are facing the critical challenges of balancing the tradeoff among privacy, security, performance, and efficiency, how to realize supervision covering the whole life cycle, and how to get explainable results. Thus, trustworthy federated machine learning is proposed to solve the above problem. In this standard, a general view of framework for trustworthy federated machine learning is provided in four parts: a principle in trustworthy federated machine learning, requirements from the perspective of different principles and different federated machine learning participants, and methods to realize trustworthy federated machine learning. Also provided is guidance on how trustworthy federated machine learning is used in various scenarios.

This document provides a reference framework for trustworthy federated machine learning, including the principles of trustworthy federated machine learning, requirements for different roles and principles of trustworthy federated machine learning, and several technologies to realize trustworthy federated machine learning. It also lists some scenarios where trustworthy federated machine learning can be applied.

The purpose of this guide is to provide a credible, practical, and controllable solution guidance for trustworthy federated machine learning and other privacy computing applications.