Machine Learning

数字数据的广泛出现创造了新的机会，既可以打破机器学习预测系统的核心瓶颈，也可以解决与信息共享相关的道德和安全问题。这项研究旨在开发新的信息传输方法，促进在不同和不断变化的现实世界场景中有效的信息共享，旨在减少对监督学习的人类注释的依赖，同时解决隐私和公平性问题。关键成果将是新一代信息传输技术，使机器学习能够应用于各种社会有益和环境适应性服务。

The widespread emergence of digital data has created new opportunities both to break the central bottleneck of machine learning for prediction systems caused by the reliance on human-annotated data and to address ethical and security concerns associated with information sharing. This research seeks to develop new information transfer methods that facilitate effective information sharing across diverse and changing real-world scenarios, aiming to reduce dependence on human annotation for supervised learning, while addressing privacy and fairness considerations. The key outcome will be a new generation of information transfer technologies that allow machine learning to be applied in diverse, socially beneficial and environmentally adaptive services.