- Rating
- 2
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 1
Read review
Paper summary
This paper tries to make a neural posterior estimator trained on small datasets work on larger datasets without retraining. Its method, center-estimated NPE (CE-NPE), combines a statistical center with an explicit contraction factor, leaving the network to learn a local center correction and residual density. For a larger dataset, the center is recomputed and the same network is reused with the prescribed scaling.
Critical strengths
-
The paper studies a useful problem. A network that accepts datasets of different sizes does not necessarily produce an accurate posterior when the sample size exceeds the training range. The paper makes this distinction clear.
-
The method has a clear statistical motivation. Using a statistical center and an explicit contraction factor reduces what the network needs to learn. The analysis also explains why getting the contraction rate right is not enough when the center is inaccurate.
-
The paper includes some useful experiments. The comparisons between free conditioning, structural scaling, and CE-NPE, together with the fixed-center controls, provide some evidence about the roles of centering and scaling within the tested settings.
Critical weaknesses
-
The additional value of the network is still unclear. The authors have already provided a way to learn a good center, and dedicated the contraction rate. This raises the question of whether a classical method (like Gaussian approximation) using the same center and model information could achieve comparable accuracy. The existing fixed-center controls do not answer this question, since they still use a neural network to learn covariance. A direct comparison would help establish what the learned correction and residual density actually add.
-
The method relies on model-specific choices of the statistical center and contraction rate. Although the paper discusses these choices and their limitations, it remains unclear how to select and validate them. Even with a suitable statistical center and contraction rate, there is not theoretical guarqntee that the correction and residual density learned at small n remain accurate at larger n. Keeping center error bounded in posterior-standard-deviation units is not enough to guarantee an accurate posterior.
-
The benefit of the additional computation is unclear. As the authors acknowledge, CE-NPE is partially amortized when center estimation requires numerical optimization. That is, numerical optimization must be redone for each new dataset to estimate the statistical center. In fact, existing literature already has fully amortized posterior sampling for high-dimensional observations (Kim et al. 2025). Note that their work also used a simultaneous summary statistics learning. Another example is Wildberger et al. (2023) which is an amortized flow matching-based method. Schmitt et al. (2024) proposed consistency model posterior estimation (CMPE), which enables fully amortized posterior sampling with a small number of network evaluations and no dataset-specific optimization. A comparison of posterior accuracy and total computational cost would help clarify whether CE-NPE’s additional optimization is worthwhile.
-
Kim et al. (2025), Deep Generative Quantile Bayes.
-
Wildberger et al. (2023), Flow Matching for Scalable Simulation-Based Inference.
-
Schmitt et al. (2024), Consistency Models for Scalable and Fast Simulation-Based Inference.
Questions for authors
-
On a suitable benchmark, could you compare CE-NPE with a Gaussian approximation using the same statistical center and a covariance estimated from local curvature?
-
For SIR or Lotka–Volterra, how do posterior accuracy and inference time change with the optimization budget for the statistical center, keeping the trained network fixed?
3 comments
W3. How does partial amortization compare with fully amortized alternatives in accuracy and cost? CE-NPE is partially amortized only through the center: for a new dataset, the only computation besides the network is , which needs no simulator calls and no retraining and takes milliseconds on four tasks and about one second on SIR and LV (general comment, G3). DQB, FMPE and CMPE are fully amortized over datasets, but none of them addresses sample sizes beyond those seen in training. To apply them here, a set encoder and a size input must be added, which makes them free size-conditioned estimators. We will discuss them in related work. Within the response period we implemented FMPE in this way (same encoder, input and dataset budget as our baselines, three seeds); we did not run DQB or CMPE.
Accuracy. FMPE fits the training range only on the two Gaussian-linear tasks (0.564 and 0.687 at ), and there it degrades to 0.940 and 0.942 at , against 0.558 and 0.562 for CE-NSF. On the other four tasks it is above 0.92 already at , so we draw no conclusion from them.
Cost. Sampling is cheaper than with FMPE, which integrates an ODE for every sample: for a batch of 400 queries CE-G is faster than FMPE on every benchmark.
Is the additional optimization worthwhile? Among learned estimators, yes: every variant with free size conditioning that we tried (other size encodings, best-effort tuning, the center as an input, 15 times the training, FMPE) is at 0.83–1.00 at , against 0.51–0.77 for CE-NPE. The comparison with a same-center Gaussian is under W1.
Q2. How do SIR and LV accuracy and runtime change with center-optimization budget, keeping the network fixed? With the trained networks fixed (three seeds), we recomputed the center at with increasing budgets. Trajectories are deterministic ODE solves and the time is per query; the last column is the median posterior-mean error of CE-G in reference SDs. This sweep used a separate training run, so its values differ slightly from those elsewhere.
| Task | Grid, GN steps, starts (LV only) | Trajectories | Time | C2ST, CE-G / CE-NSF | Mean error (SDs) |
|---|---|---|---|---|---|
| SIR | , 0 | 122 | 0.18 s | 0.999 / 1.000 | 60.0 |
| , 2 | 138 | 0.53 s | 0.887 / 0.874 | 3.77 | |
| , 2 | 458 | 0.53 s | 0.811 / 0.801 | 0.84 | |
| , 2 | 1,698 | 0.53 s | 0.785 / 0.757 | 0.47 | |
| , 4 (default) | 1,714 | 0.89 s | 0.783 / 0.754 | 0.46 | |
| , 6 | 6,610 | 1.24 s | 0.783 / 0.754 | 0.45 | |
| LV | , 0, 1 | 2,402 | 0.31 s | 1.000 / 1.000 | 82.8 |
| , 2, 1 | 2,426 | 0.93 s | 0.944 / 0.955 | 1.95 | |
| , 2, 4 | 6,661 | 0.93 s | 0.788 / 0.817 | 0.28 | |
| , 2, 8 | 28,761 | 0.92 s | 0.727 / 0.756 | 0.18 | |
| , 4, 8 (default) | 28,953 | 1.53 s | 0.718 / 0.750 | 0.17 | |
| , 6, 16 | 84,689 | 2.15 s | 0.718 / 0.749 | 0.17 |
Accuracy saturates at the default budgets: larger budgets change C2ST by at most 0.001. Smaller budgets trade accuracy for time; on SIR, two Gauss–Newton steps instead of four save 40% of the time at a cost of 0.002–0.003. Runtime is set by the number of Gauss–Newton steps, since the grid is evaluated in one batched solve. The appendix states three refinement steps for SIR; the code and all reported results use four, and we will correct the text.
I appreciate the authors for the response and additional experiments. The center optimization budget experiment largely addresses my question on this point.
That being said, the benefit over classical approximations part and the scope of the theoretical claims remain insufficient to me. I therefore maintain my rating for now.
W1 / Q1. What does the network add relative to a same-center classical Gaussian? We ran this comparison on all six benchmarks, with the covariance from the posterior Hessian at the center, from the Gauss–Newton (GN) curvature of the moment objective, or from a parametric bootstrap of the center. C2ST at n=20/100, the largest training size and five times it (classical controls: three evaluation seeds):
The classical Gaussians have four limitations that CE-NPE does not share:
Where no likelihood is available, the classical option is the method-of-moments Gaussian at the same center, with covariance (J⊤nΣ−1J)−1 from the center's moment curves and the sample covariance Σ of the observations. We ran it on the held-out simulated population-genetics data of Appendix G (400 datasets per size, generating parameters known, no bootstrap). It is less accurate than CE-NPE at every size (training sizes 5–20):
Their strength is the large-n regime of regular models. At five times the training maximum they are close to the ideal, as the Bernstein–von Mises theorem predicts: CE-NSF is within 0.02 of the best of them on GL, GL uniform and the mixture, and behind on the GLM, SIR and LV (0.65–0.76 against 0.51–0.61). We will add them to the main comparison as this large-n reference and state when CE-NPE is the better choice
W2. How are the center and rate selected and validated, and what guarantees remain at larger n?
Selection. See the general comment G1, G2. Our response to Reviewer TF4Z, W5.
Guarantees. Appendix B states what holds once the network is fixed, and its results order the Gaussian estimators by KL divergence to the true posterior as n grows. Proposition 3 splits KL(Pn∥Qn) exactly into a Gaussian-shape term and mean and covariance terms, each bounded above and below by the errors of the fitted mean and covariance in the local coordinates. By Proposition 1, a learned correction that stays bounded keeps these errors bounded when it acts on the posterior scale, as in CE-G, so the KL of CE-G beyond the shape term stays bounded. A learned center whose error in parameter units does not shrink at rate n−α, which nothing in Free-G or Scale-G enforces, makes the mean term grow as n2α, and a fitted width that does not contract at the true rate makes the covariance term grow; their KL diverges. This is a partial ordering: CE-G below Free-G and Scale-G, with Free-G and Scale-G not ordered against each other. Proposition 2 gives how center and rate errors grow with the extrapolation factor. We will state the ordering as a corollary in Appendix B.
Validation. Accuracy at a new size reduces to how small the network's local errors remain there, which can be checked at the deployment size with no reference posterior and no retraining: simulate a few hundred datasets of that size and compute the coverage of the generating parameters (or SBC) together with their held-out log density. We will add this procedure to the paper.