REVIEW 3 major objections 5 minor 2 references
Likelihood Matching for Diffusion Models
T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash
Pith's one-line read Training diffusion models by directly maximizing an approximate data likelihood, using matched Gaussian reverse transitions, is consistent and yields explicit sampling error bounds.
desk verdict Genuinely new quasi-likelihood objective with Hessian-aware sampling, but Theorem 1's sampler bound omits a non-vanishing Gaussian approximation error. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The central machinery is the reverse-path likelihood identity of Proposition 1 combined with the Gaussian quasi-likelihood of equation (9). The identity rewrites the expected log-likelihood of the initial data density as an expected sum of log-transition densities of the reverse process, so maximizing data likelihood is equivalent to maximizing reverse path likelihood up to constants. The quasi-likelihood replaces each intractable reverse transition density $p_{t_{k-1}|t_k}$ by a Gaussian $\varphi_d(y;\mu_{t_{k-1}|t_k},\Sigma_{t_{k-1}|t_k})$ whose conditional mean and covariance are the true ones from Proposition 2: $\mu_{s|t} = (x_t + \sigma^2_{t|s}\nabla\log q_t(x_t;\theta))/m_{t|s}$ and $\Sigma_{s|t} = (\sigma^2_{t|s}/m^2_{t|s})(I_d + \sigma^2_{t|s}\nabla^2\log q_t(x_t;\theta))$. This machinery turns maximum likelihood estimation into a tractable objective depending on the score and Hessian, and the same matched moments define the stochastic sampler (13)-(14).
What would settle it
Direct calculation on a distribution with known non-Gaussian reverse transitions, such as a well-separated two-component Gaussian mixture in dimension 2 or 10, would settle the matter: compute the exact negative log-likelihood $-E[\log q_0(X_0;\theta)]$, the LM objective (15), and the total-variation distance of samples from the sampler (13) as $T$ grows with oracle score and Hessian. If the gap between the LM objective and the true negative log-likelihood persists at a fixed positive level as $T \to \infty$, the Gaussian quasi-likelihood approximation is not harmless.
Extended reading notes
Core claim
The paper's central claim is that minimizing the Likelihood Matching objective in equation (15) is an operational way to maximize the log-likelihood of the target data distribution. Proposition 1 gives the identity $-E[\log q_0(X_0;\theta)] \approx -E[\sum_{k=1}^N \log p_{t_{k-1}|t_k}(X_{t_{k-1}}|X_{t_k};\theta)] + C_T$, expressing the data negative log-likelihood as an expected sum of reverse transition log-densities along the forward sample paths. Since the true reverse transition densities are non-Gaussian and unavailable, Proposition 2 derives their conditional mean and covariance, showing they depend on the score $\nabla \log q_t$ and the Hessian $\nabla^2 \log q_t$, and equation (9) replaces each transition density with a Gaussian having those matched moments. The paper parameterizes the score and Hessian by a common parameter $\phi$, minimizes the resulting quasi-likelihood (15), and samples by (13)-(14) using both learned moments. The result is a training objective that contains score matching as a component but adds covariance matching and likelihood weighting, with consistency of the quasi-MLE (Theorem 2) and the non-asymptotic total-variation bound (Theorem 1) as the theoretical payoffs.
Load-bearing premise
The load-bearing premise is that each non-Gaussian reverse transition density of the true diffusion process can be replaced by a Gaussian with the same conditional mean and covariance without losing information that matters for the final sample distribution, and that the resulting approximation error stays small enough that the Likelihood Matching objective remains a faithful surrogate for the data log-likelihood.
Editorial extensions
If this is right
- The LM objective (15) is an approximation of the data log-likelihood rather than an upper bound, so training on it aims directly at the quantity statistics prefers to maximize.
- With exact score and Hessian, the sampler's total variation error decays as $d^3\log^{4.5} T/T$ plus score and Hessian error terms, so the discretization error can be driven down by increasing diffusion steps.
- In the oracle-score setting, the quasi-MLE is consistent: as $n$ and $T$ grow, the estimated parameters converge to the true parameters of the data distribution.
- Because the sampler (13) uses the learned covariance as well as the score, it can produce coherent samples with fewer reverse steps than a score-only DDPM sampler in the MNIST experiments.
- The low-rank Hessian parameterization with the Sherman-Morrison-Woodbury formula keeps the per-step cost of evaluating the matched covariance manageable, making the likelihood objective computationally feasible.
Reading between the lines
- The paper does not quantify the Gaussian approximation error; a direct extension would be to bound the gap between the LM objective and the true negative log-likelihood for non-Gaussian data, which would tell practitioners when the method is safe.
- A testable prediction of Theorem 1 is that lowering Hessian error by increasing the Hessian rank $r$ pays off less than lowering score error, because the Hessian error term enters at rate $\log T/\sqrt{T}$ while the score error term enters at $\sqrt{\log T}$.
- Because the reverse-path identity (7) holds for arbitrary time grids, the same quasi-likelihood construction could be used with non-Gaussian transition approximations, such as mixtures, without changing the overall maximum-likelihood framing.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proposes Likelihood Matching (LM), a training objective for diffusion models that approximates the reverse-path log-likelihood by replacing the true reverse transition densities with Gaussian quasi-transition densities whose conditional means and covariances are matched via estimated score and Hessian functions. The authors derive a stochastic sampler that uses the estimated score and Hessian, prove consistency of the quasi-maximum-likelihood estimator in an oracle setting, and state a non-asymptotic total-variation bound for the sampler. Experiments on synthetic mixture models and low-resolution image datasets compare LM with score matching and study the effect of the number of transition points N and the Hessian rank r.
Significance. If the theoretical guarantees were correct, the paper would make a useful contribution by connecting diffusion-model training to quasi-likelihood estimation, by explicitly using Hessian information in both training and sampling, and by providing finite-sample TV bounds. The derivation of the conditional mean and covariance formulas in Proposition 2 and the use of external reverse-time diffusion theory are clear and standard. However, the central convergence theorem, as stated, omits a non-vanishing approximation error introduced by the Gaussian quasi-likelihood, and the consistency theorem is limited to an oracle score model. The paper also honestly acknowledges the lack of large-scale experimental validation; the experiments on MNIST and low-resolution image data are encouraging but not definitive. Overall, the methodological idea is interesting, but the main theoretical claim needs substantial revision before the paper can be accepted.
major comments (3)
- [Theorem 1, Eq. (17), sampler (13)-(14)] The bound in (17) does not include any term for the error caused by replacing the true reverse transition density p_{t_{k-1}|t_k} by the Gaussian quasi-density in (9). This error is not vanishing in T. For a fixed step, say from time 1 to time 0, the true reverse transition is non-Gaussian whenever q0 is non-Gaussian, and the Gaussian quasi-kernel with matched conditional mean and covariance is not equal to it, even with exact score and Hessian. Concretely, take q0 to be a well-separated two-component Gaussian mixture, set T=1 with the paper's convention t_k=k and β_t=1, and use the exact score and Hessian. The sampler (13)-(14) then applies one Gaussian quasi-kernel to Y_1 ~ N(0, I_d), producing a distribution that is not q0, so TV(q0||p̃0) is strictly positive, while the right-hand side of (17) is d^3 log^{4.5}(1)/1 = 0 when the score and Hessian errors vanish. More generally, as T grows, the final reverse step from time 1 to time 0 is independent of the horizon and its Gaussian approximation error persists, so the first term in (17) cannot account for it. The theorem as stated is therefore internally inconsistent for non-Gaussian q0. The authors need to add an explicit quasi-Gaussian approximation error term, or restrict the theorem to cases where the reverse transitions are exactly Gaussian, or otherwise explain why the missing term is covered by the existing terms.
- [Theorem 2 and Section 3.1] The consistency result is proved only for the oracle case where the score is known up to the true parameter θ, and it relies on the Gaussian quasi-likelihood being a valid inference objective. For non-Gaussian q0, the quasi-likelihood (10) is misspecified at the level of the full transition density, even when the conditional means and covariances are correct. Consistency of the quasi-MLE then requires additional identification and regularity conditions that are not stated in the main text and are referenced only to an Appendix B.4. The statement 'θ* is the parameter of the original data distribution q0' needs support showing that the population quasi-likelihood L(ϕ) is uniquely minimized at the true θ, and that the empirical objective (15) converges uniformly over the parameter space. Without these details, the consistency claim is not established for the actual setting of the paper.
- [Equation (8) and the finite-T approximation] The text replaces E[log p_{t_N}(X_{t_N}; θ)] by a θ-free constant on the grounds that the marginal converges to N_d(0, I_d). For a finite diffusion horizon T, which is the horizon used in Theorem 1 and in the experiments, this term is not exactly θ-free; it depends on the initial distribution through q_T(·; θ). The approximation in (8) therefore introduces an additional error of order depending on how close q_T is to stationarity. This error should be made explicit and accounted for in the objective (15), or at least bounded in the theory.
minor comments (5)
- [Section 3.1] There are several typographical issues in the notation for transition densities, e.g., 'pt−1|t(Ytt−1|Yt; θ)' should be 'p_{t_{k-1}|t_k}(Y_{t_{k-1}}|Y_{t_k}; θ)', and 'Yt d = Xt' is typeset inconsistently.
- [Figure 3 caption] The caption contains the typo 'Socre Matching' instead of 'Score Matching'.
- [Equation (7) and Proposition 1] The underbrace text 'converge to Nd(0,Id)' is grammatically incomplete; it should state that p_{t_N} converges to the standard normal density, and the convergence should be qualified with respect to T.
- [Appendix] The main text refers to Appendix B for the proofs of Proposition 2, Theorem 1, and Theorem 2, and to Appendix C for experimental details, but the appendices are not included in the submitted version. The proofs and implementation details need to be available for verification.
- [Notation in Eq. (13)-(14)] The sampler equations use both 'Y_{t-1}' and 't' in a way that is ambiguous with the continuous-time index; it would be clearer to write the discrete index explicitly, e.g., 'Y_{k-1}' for k = T, ..., 1.
Circularity Check
No significant circularity: the likelihood-matching equivalence rests on external reverse-time SDE theory, and the Gaussian quasi-likelihood is an acknowledged approximation rather than an identity forced by construction.
full rationale
The paper's central derivation does not reduce to its own inputs by construction. Proposition 1 is imported from external reverse-time SDE theory (Anderson, 1982; Haussmann and Pardoux, 1986), and Proposition 2 supplies conditional mean and covariance formulas derived from the forward Ornstein-Uhlenbeck transition, not from the Likelihood Matching objective. Equation (9) explicitly defines the quasi-likelihood as a Gaussian replacement for the true reverse transition density, and the paper consistently describes this as a quasi-likelihood approximation rather than an exact identity. The consistency theorem (Theorem 2) is stated only under an oracle score model and is a separate statistical claim, not a restatement of the objective's definition. There are no self-citations carrying load-bearing arguments, and no fitted parameters are relabeled as predictions. The main theoretical risk is the correctness of the unquantified Gaussian approximation in Theorem 1, but that is a validity concern, not a circularity. Therefore, no circular steps are identified.
Assumptions & free parameters
free parameters (3)
- Hessian rank r =
10, 20, 30, 40 in MNIST experiments
- Number of transition points N =
2, 3, 8 for mixtures; 2 for image experiments
- Diffusion coefficient schedule beta_t =
No explicit formula in main text; follows Li et al. (2023)
assumptions (5)
- standard math Anderson and Haussmann-Pardoux time-reversal equivalence: reverse process marginals equal forward marginals
- standard math Regularity conditions on q0 and beta_t for the path-likelihood identity
- domain assumption Assumption 4: lambda_min((1-alpha_t) grad^2 log q_t(x)) >= epsilon_0 > -1 uniformly
- domain assumption Assumptions 2 and 3: score and Hessian estimators exist with L2 and Frobenius errors epsilon_s and epsilon_H
- ad hoc to paper Gaussian quasi-likelihood: reverse transition p_{t_{k-1}|t_k} is approximated by N(mu, Sigma) with matched moments
Cite this review
Pith. "Pith review of Likelihood Matching for Diffusion Models." pith.science (2026). https://pith.science/paper/ZMTC7JVN
@misc{pith2026250803636,
author = {Pith},
title = {Pith review of: Likelihood Matching for Diffusion Models},
year = {2026},
howpublished = {\url{https://pith.science/paper/ZMTC7JVN}},
note = {Machine review of arXiv:2508.03636}
}
read the original abstract
We propose a Likelihood Matching approach for training diffusion models by first establishing an equivalence between the likelihood of the target data distribution and a likelihood along the sample path of the reverse diffusion. To efficiently compute the reverse sample likelihood, a quasi-likelihood is considered to approximate each reverse transition density by a Gaussian distribution with matched conditional mean and covariance, respectively. The score and Hessian functions for the diffusion generation are estimated by maximizing the quasi-likelihood, ensuring a consistent matching of both the first two transitional moments between every two time points. A stochastic sampler is introduced to facilitate computation that leverages both the estimated score and Hessian information. We establish consistency of the quasi-maximum likelihood estimation, and provide non-asymptotic convergence guarantees for the proposed sampler, quantifying the rates of the approximation errors due to the score and Hessian estimation, dimensionality, and the number of diffusion steps. Empirical and simulation evaluations demonstrate the effectiveness of the proposed Likelihood Matching and validate the theoretical results.
Figures
Figures from the paper (1 more)
Reference graph
Works this paper leans on
-
[1]
Anderson, B. D. (1982). Reverse-time diffusion equation models. Stochastic Processes and their Applications, 12(3):313–326. Benton, J., De Bortoli, V ., Doucet, A., and Deligiannidis, G. (2023). Nearlyd-linear convergence bounds for diffusion models via stochastic localization. International Conference on Learning Representations. Betker, J., Goh, G., Jin...
arXiv 1982
-
[27]
Gretton, A., Borgwardt, K. M., Rasch, M. J., Schölkopf, B., and Smola, A. (2012). A kernel two-sample test. The Journal of Machine Learning Research, 13(1):723–773. Haussmann, U. G. and Pardoux, E. (1986). Time reversal of diffusions. The Annals of Probability, pages 1188–1205. Ho, J., Jain, A., and Abbeel, P. (2020). Denoising diffusion probabilistic mod...
arXiv 2012
Reviewed August 6, 2026 · model on record in the stance chip above.
Discussion (0). Sign in to comment.