SAGE: Training-Free Semantic Evidence Composition for Edge-Cloud Inference under Hard Uplink Budgets
SAGE:在严格上行预算下无训练语义证据组合用于边缘-云计算推断
机构 * School of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST)(电气工程学院,韩国科学技术院(KAIST))
AI总结 本文提出SAGE方法,通过结合重要性过滤与嵌入多样性采样,在边缘-云计算推断中实现高精度传输,优于传统重要性方法。
Comments 11pages, 9 figures