-
Gaussian/Normal Distribution
-
sigma = one standard deviation
-
Gaussian = Range of Doubtness
-
probability x is depend on the mew and sigma parameters
-
Fixed formula and no need to be remembered
-
Less the sigma, higher the probability, make a narrower and higher graph
-
more sigma, more doubtness, fatter gaussian distribution
-
we want to plot the Gaussian distribution of every examples
-
The problem is we don't know what to set mew and sigma squared parameter
-
Or maybe we can set the parameters?
-
We can set the parameters based on the distance of the parameters gathered
-
mew = average of the example
-
sigma squared = is the range of error. how much the variance of x from the average(mew) value. Summation of range error between x and amew
-
Maximum Likelihood Estimate is the name of formula above
-
ML tend to use m training set rather than m-1
SUMMARY
-
Given a training set, now we can set mew and sigma for Gaussian Distribution
-
Next how to implement GD in Anomaly Detection
