REVIEW 1 cited by
Theory and Algorithms for Forecasting Time Series
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Theory and Algorithms for Forecasting Time Series
read the original abstract
We present data-dependent learning bounds for the general scenario of non-stationary non-mixing stochastic processes. Our learning guarantees are expressed in terms of a data-dependent measure of sequential complexity and a discrepancy measure that can be estimated from data under some mild assumptions. We also also provide novel analysis of stable time series forecasting algorithm using this new notion of discrepancy that we introduce. We use our learning bounds to devise new algorithms for non-stationary time series forecasting for which we report some preliminary experimental results.
Forward citations
Cited by 1 Pith paper
-
Learning Ergodic Dynamical Systems from a Finite Trajectory
For uniformly geometrically ergodic Markov systems, regularized least squares learns the optimal one-step predictor from a single finite trajectory with excess-risk bounds of order T^{-1/2} with respect to the invaria...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.