- Rating
- 2(+1)
- Confidence
- 4
- Informativeness
- 1
- Presentation
- 1
- Soundness
- 1
Read review
Paper summary
The paper considers a problem of sampling from a tilted distribution , where $ p_0$ can be sampled using a diffusion model.
The authors propose an approximate sampling scheme to sample from by alternating between two steps
Starting from the algorithm alternates between two steps
-
Given current map it to clean estimate by applying map , where is the approximation of the flow map sending to . Denote by the distribution of .
-
Noise back using kernel K to get that approximates
subject to
In practice they use
- take as the posterior mean approximation .
- Take as , where
They show that the algorithm performs strongly on FFHQ and ImageNet.
Critical strengths
Algorithm presents stong empirical performance
Critical weaknesses
The paper is very hard to follow, and I am still not sure if I understood it correctly.
-
If I understand correctly, the algorithm closely resembles the TMPD algorithm the authors cite; the posterior mean approximation formula is inherited from it, and the only difference is the further approximation of the kernel . However, I do not see the comparison between them in Tables 2 and 3.
-
The choice of the kernel does look arbitrary; the approximation by while standard breaks the premise that the kernel satisfies the condition on the algorithm.
Questions for authors
What is the purpose of the rematching step? When we have a posterior mean, with Tweedie we have access to the posterior score and can simulate the backward process (I believe this is what TMPD does). The ablation to see whether this helps is required.