Talk:Error tolerance (PAC learning)

What is n in the definition of learnable?

"in a number of calls to the oracle bounded by $p\left({\frac {1}{\varepsilon }},{\frac {1}{\delta }},n,{\text{size}}(f)\right)$" What is $n$ here? And how do you measure ${\text{size}}(f)$? 65.112.8.17 (talk) 18:32, 9 September 2024 (UTC)

Inconsistent number of arguments to p

In the "Definition" subsection of the "Classification Noise" section: "a polynomial $p(\cdot ,\cdot ,\cdot ,\cdot )$ such that ... bounded by $p\left({\frac {1}{1-2\eta _{B}}},{\frac {1}{\varepsilon }},{\frac {1}{\delta }},n,size(f)\right)$". The first <math>p<\math> has four arguments, the second has five. What is correct here? 65.112.8.17 (talk) 18:36, 9 September 2024 (UTC)