
Unsupervised approximate-semantic vocabulary learning for human action and video classification
Keywords: Contextual spectral embedding; Pearson product moment correlation; Visual vocabulary