1/13
Looks like no tags are added yet.
What is formula for True Error?

What is the problem with True Error (CD)?
The quantity is unknown as distribution is unknown
What is the formula for Observed error (ĉS)?

Is cD a deterministic quantity (a number) or a random variable?
If c is a fixed function then cD is deterministic (a number in [1,0]), if c is the classifier trained on sample S then cD is a random variable as a function of c
What is the distribution of n * ĉS ?

What is Binomial CDF (cumulative distribution function)?

What is the pivot of the cumulative?
The largest True error such that the probability observing k or fewer errors in n trial (test points) is at least the confidence interval (δ), where p is the True Error.

What is an upper bound approximation (pivot of cumulative) that is simpler to compute but makes the bound less tight (Corollary)?

What does the Test Set Bound tell us?

What is S?

What would happen if S was a Training sample?

What is Occam’s Razor Bound Theorem?

What is Corollary for Occam’s Razor Bound?

Conclusions of this topic:
