Skip to main content
Artificial Intelligence

Return to Answer

added 15 characters in body
Source Link
Reza_va
Reza_va
  • 89
  • 7

I think what he means is that, while distribution P(y=∣x,s)$P(y \mid x,s)$ has three variables, P(y∣x)$P(y \mid x)$ has two variables. The number of parameters required to describe a distribution (or samples to approximate it) grows exponentially with the number of variables (for more info see for instance Ian goodfellow, .. deep learning").

I think what he means is that, while distribution P(y=∣x,s) has three variables, P(y∣x) has two variables. The number of parameters required to describe a distribution (or samples to approximate it) grows exponentially with the number of variables (for more info see for instance Ian goodfellow, .. deep learning").

I think what he means is that, while distribution $P(y \mid x,s)$ has three variables, $P(y \mid x)$ has two variables. The number of parameters required to describe a distribution (or samples to approximate it) grows exponentially with the number of variables (for more info see for instance Ian goodfellow, .. deep learning").

Source Link
Reza_va
Reza_va
  • 89
  • 7

I think what he means is that, while distribution P(y=∣x,s) has three variables, P(y∣x) has two variables. The number of parameters required to describe a distribution (or samples to approximate it) grows exponentially with the number of variables (for more info see for instance Ian goodfellow, .. deep learning").