Why time-series models fail despite low training error

Original: Flow map learning in nonlinear vector autoregressive models: influence of the feature-library structure on the training error

Source: arxiv.org ↗