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.
The Nystr\"om method for functional quantization with an application to the fractional Brownian motion
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
In this article, the so-called "Nystr\"om method" is tested to compute optimal quantizers of Gaussian processes. In particular, we derive the optimal quantization of the fractional Brownian motion by approximating the first terms of its Karhunen-Lo\`eve decomposition. A numerical test of the "functional stratification" variance reduction algorithm is performed with the fractional Brownian motion.
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.