filter - Anomaly Detection Algorithms -
i tasked detecting anomalies (known or unknown) using machine-learning algorithms data in various formats - e.g. emails, ims etc.
what favorite , effective anomaly detection algorithms?
what limitations , sweet-spots?
how recommend limitations addressed?
all suggestions appreciated.
statistical filters bayesian filters or bastardised version employed spam filters easy implement. plus there lots of online documentation it.
the big downside cannot detect unknown things. train large sample of known data can categorize new incoming data. can turn traditional spam filter upside down: train recognize legitimate data instead of illegitimate data doesn't recognize anomaly.
Comments
Post a Comment