A sequence-to-point knowledge distillation approach trains compact student models to match the short-term forecasting performance of larger teacher models while cutting model size and memory by over 10x on AI data center load data.
Challenges and approaches to time-series forecasting in data center telemetry: A survey
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.SY 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Deployment-Efficient Short-Term Load Forecasting in AI Data Centers via Sequence-to-Point Knowledge Distillation
A sequence-to-point knowledge distillation approach trains compact student models to match the short-term forecasting performance of larger teacher models while cutting model size and memory by over 10x on AI data center load data.