Beginning Anomaly Detection Using

lr: Some float value where the learning rate lr >= 0. The learning rate 

is a hyperparameter that determines how big of a step to take when 

optimizing the loss function. The paper describes good results with a 

value of 0.001 (the paper refers to the learning rate as 

alpha).

• 

beta_1: Some float value where 0 < beta_1 < 1. This is usually some 

value close to 1, but the paper describes good results with a value of 0.9.

• 

beta_2: Some float value where 0 < beta_2 < 1. This is usually some 

value close to 1, but the paper describes good results with a value  

of 0.999.

• 

epsilon: Some float value where epsilon e >= 0. If None, then it 

defaults to K.epsilon(). Epsilon is some small number, described as 

10E-8 in the paper, to help prevent division by 0.

•