Description: | In terms of the taxonomy we describe in this paper, the only classifier we are aware of that explicitly ensembles over different representations is the Collective of Transformation Ensembles (COTE). Bagnall $et\:al.$ propose the meta ensemble COTE, a combination of classifiers in the time, autocorrelation, power spectrum and shapelet domain. The components of EE and ST are pooled with classifiers built on a version of autocorrelation transform (ACF) and power spectrum (PS) transform. EE uses the 11 classifiers described above. ACF and PS employ the same 8 classifiers used in conjunction with the shapelet transform. We use the classifier structure called flat-COTE. This involves pooling all 35 classifiers into a single ensemble with votes weighted by train set cross validation accuracy. There is however one difference: we use the Shapelet Transform described in Binary shapelet transform for multiclass time series classification rather than the version in Classification of time series by shapelet transformation. |