On calibration error of randomized forecasting algorithms

Vladimir V. V'yugin

Research output: Contribution to journalArticlepeer-review

4 Citations (Scopus)

Abstract

It has been recently shown that calibration with an error less than Δ > 0 is almost surely guaranteed with a randomized forecasting algorithm, where forecasts are obtained by random rounding the deterministic forecasts up to Δ. We show that this error cannot be improved for a vast majority of sequences: we prove that, using a probabilistic algorithm, we can effectively generate with probability close to one a sequence "resistant" to any randomized rounding forecasting with an error much smaller than Δ. We also reformulate this result by means of a probabilistic game.

Original languageEnglish
Pages (from-to)1781-1795
Number of pages15
JournalTheoretical Computer Science
Volume410
Issue number19
DOIs
Publication statusPublished - 28 Apr 2009
Externally publishedYes

Keywords

  • Algorithmic prediction
  • Calibration
  • Machine learning
  • Randomized prediction
  • Randomized rounding
  • Universal prediction

Fingerprint

Dive into the research topics of 'On calibration error of randomized forecasting algorithms'. Together they form a unique fingerprint.

Cite this