no code implementations • 29 Aug 2017 • Pelumi Oluwasanya
The approach is based on the principle that the entropy of the data is increased when an anomalous data point is measured.
no code implementations • 27 Aug 2017 • Pelumi Oluwasanya
The main work can be described as this; given a set of measurements from sensors that represent a normal situation, [1] proceeds by first estimating the probability density function (pdf) of the set using a data-split approach, then estimate the entropy of the set using the arithmetic mean as an approximation for the expectation.