Scholarlink

FL can reduce data-transfer volumes and mitigate privacy risks by pushing computation to the edge; however, it may also increase energy use and carbon emissions if naively orchestrated across heterogeneous devices and networks. This chapter formalizes sustainability objectives for FL, connects them to the United Nations Sustainable Development Goals (UN SDGs), and develops concrete, measurable strategies for energy- and carbon-aware FL pipelines. We conclude with a robustness module based on Representational Similarity Analysis (RSA) that prevents wasteful training under data poisoning, thereby improving both accuracy and sustainability.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Sustainability in Federated Learning

  • 章节

  • 发表日期：

    2026年

查看全文
  • Kai Li,
  • Xin Yuan,
  • Wei Ni

摘要

FL can reduce data-transfer volumes and mitigate privacy risks by pushing computation to the edge; however, it may also increase energy use and carbon emissions if naively orchestrated across heterogeneous devices and networks. This chapter formalizes sustainability objectives for FL, connects them to the United Nations Sustainable Development Goals (UN SDGs), and develops concrete, measurable strategies for energy- and carbon-aware FL pipelines. We conclude with a robustness module based on Representational Similarity Analysis (RSA) that prevents wasteful training under data poisoning, thereby improving both accuracy and sustainability.