Downsampling recommender training data can cut runtime and estimated carbon emissions by roughly 18 to 52 percent, with performance losses that vary strongly by algorithm, dataset, and downsampling design.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Optimal Dataset Size for Recommender Systems: Evaluating Algorithms' Performance via Downsampling
Downsampling recommender training data can cut runtime and estimated carbon emissions by roughly 18 to 52 percent, with performance losses that vary strongly by algorithm, dataset, and downsampling design.