An online-learning classifier (adaptive random forest) with a Sketch-Min counter reportedly identifies hot and cold data with 90% accuracy on sampled memory traces, though the evidence is not reproducible.
Hdfs optimization strategy based on hierarchical storage of hot and cold data
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Hammer: Towards Efficient Hot-Cold Data Identification via Online Learning
An online-learning classifier (adaptive random forest) with a Sketch-Min counter reportedly identifies hot and cold data with 90% accuracy on sampled memory traces, though the evidence is not reproducible.