TrimCaching introduces parameter-sharing edge caching for AI models, formulates it as a submodular maximization problem with submodular constraints, provides approximation algorithms for special and general cases, and shows improved cache hit ratios in simulations.
PartialLoading: User scheduling and bandwidth allocation for parameter-sharing edge inference
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.NI 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
TrimCaching: Parameter-sharing Edge Caching for AI Model Downloading
TrimCaching introduces parameter-sharing edge caching for AI models, formulates it as a submodular maximization problem with submodular constraints, provides approximation algorithms for special and general cases, and shows improved cache hit ratios in simulations.