thesis1/prep/10Normalising_Flows/q

6 lines
230 B
Plaintext

Deep Learning Method, in which the output is normalised
$\int f(x) dx=1 \; \forall f(x)$
Can be used to estimate probability density functions
->Thus useful for AD
$\int f(h(x)) \|\frac{\delta h}{\delta x}\| dx=1 \; \forall h(x)$