Bosch is marvelous. Mucha and Monet are good. Michelangelo tries to be more like his sculptures. Not sure about Walt Disney and Roy Lichtenstein, and Rembrandt

Bosch is marvelous. Mucha and Monet are good. Michelangelo tries to be more like his sculptures. Not sure about Walt Disney and Roy Lichtenstein, and Rembrandt version is especially ugly. Maybe it will worth to get rid of "van Rijn" in that sample. Overall: the impression is better with author's popularity. I think if we train the model only with well-tagged filtered dataset - results may be much better, but we effectively will get a 5-year old Prizm app.