摘要: When given multiple models it is often useful to combine them for improved reliability or performance over the individual models. Over years many outlier metrics and detection methods have been developed purposed of finding data incongruous with rest data. Inspired by successes supervised ensemble machine learning, proposed combining anomaly together. We investigate usefulness building ensembles purpose detection. find that currently, best our knowledge, there no great advantage in using anything more complicated than simple average all available scores.