Timeline for Purpose of scaling mean by $\sqrt{1 - \beta_t}$ in forward diffusion process
Current License: CC BY-SA 4.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 10, 2023 at 20:59 | comment | added | Cedric Martens | > If you start with just a number $x_0 = 0$, after $T$ steps you will have a value $x_T \in [-T, T]$, Why is this true? I thought the normal distribution has a range from $-\infty$ to $\infty$ | |
| Apr 18, 2023 at 17:18 | comment | added | Einlar | @user25004 Exactly. This is because $\epsilon_1$ is independent of $x_0$, which implies that their covariance is zero. | |
| Feb 19, 2023 at 20:12 | vote | accept | Adrian Stoll | ||
| Feb 7, 2023 at 21:34 | comment | added | user25004 | From the way the variance is computed $Var(x_1) = a^2 +\beta_1$, the Gaussian $\epsilon_1$ and $x_0$ should have covariance zero. Why is that the case? | |
| Jan 29, 2023 at 17:25 | review | Late answers | |||
| Jan 29, 2023 at 17:33 | |||||
| S Jan 29, 2023 at 17:05 | review | First answers | |||
| Jan 29, 2023 at 17:58 | |||||
| S Jan 29, 2023 at 17:05 | history | answered | Einlar | CC BY-SA 4.0 |