TimeMCL uses Winner-Takes-All training of multiple heads to quantize the conditional distribution of future time series, producing diverse forecasts at low inference cost.
Here, TimeMCL follows the same experimental setup as in the previous benchmark, except that we used Z-Score normalization (instead of mean scaling) during training
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Winner-takes-all for Multivariate Probabilistic Time Series Forecasting
TimeMCL uses Winner-Takes-All training of multiple heads to quantize the conditional distribution of future time series, producing diverse forecasts at low inference cost.