PTMT is a lightweight framework that automates parameter tuning for memory tiering via hybrid offline database building and online customized reinforcement learning, delivering 14-30% gains over defaults and 32% over prior art on four systems.
Parameters tuning of multi-model database based on deep reinforcement learning.Jour- nal of Intelligent Information Systems, 61(1):167–190
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.OS 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Hybrid Adaptive Tuning for Tiered Memory Systems
PTMT is a lightweight framework that automates parameter tuning for memory tiering via hybrid offline database building and online customized reinforcement learning, delivering 14-30% gains over defaults and 32% over prior art on four systems.