Bin Model vs. Learning
- unknown orange prob μ ↔ fixed hypothesis h(x)=? target f(x)
- marbel ∈ bin ↔ x∈
- orange
↔
h
is wrong:
h(x)≠f(x) - green
↔
h
is right:
h(x)=f(x) - size-N sample form bin ↔ check on ={(xn,yn)}
- i.i.d. marbles ↔ i.i.d. xn
If large N & i.i.d.
xn , can probably infer unknown