Overfitting, curse of dimensionality, etc. As in many machine learning tasks, extracting good starting-point features is very very important, and that often req

Overfitting, curse of dimensionality, etc. As in many machine learning tasks, extracting good starting-point features is very very important, and that often requires domain-specific knowledge.