Research map / Probabilistic and scientific modeling / Variational and generative sampling
Beyond black box densities: Parameter learning for the deviated components
Dat Do, Nhat Ho, XuanLong Nguyen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Estimate the deviated proportion and mixture components with MLE under distinguishability | Deviating mixture model with a pre-trained Normalizing Flow h0 · Distinguishable synthetic setting | error rate for λ: $(\log(n)/n)^{1/2}$ | Improved |
| Fit the known number of deviated components | Deviating mixture model · Distinguishable synthetic setting | W1 error rate for G: $(\log(n)/n)^{1/2}$ | Improved |
| Allow more fitted components than the true number | Deviating mixture model · Distinguishable synthetic setting | W2 error rate for G: $(\log(n)/n)^{1/4}$ | Hurt |
| Fit in the regime λ̂n < λ* | Partially distinguishable deviating mixture model · Partially distinguishable setting | convergence of λ̂n to λ*: parametric rate | Improved |
| Fit in the regime λ̂n > λ* | Partially distinguishable deviating mixture model · Partially distinguishable setting | convergence of λ̂n to λ* | Hurt |
| Use the fitted mixture G to represent the equivalent deviation when λ > λ* | Partially distinguishable deviating mixture model · Partially distinguishable setting, λ̂n > λ* | convergence of W4(Ĝn, Ḡ*(λ̂n)): $(\log(n)/n)^{2\overline{r}(2)}=(\log(n)/n)^{1/8}$ | Improved |
| Use over-fitted rather than exact-fitted component count | MLE for the deviating mixture model · Distinguishable setting | mixture-measure error rate: $(\log(n)/n)^{1/2}$ versus $(\log(n)/n)^{1/4}$ | Hurt |
Closest papers by meaning
- Efficient Mixture Learning in Black-Box Variational Inference (ICML 2024 · similarity 0.88)
- Refined Convergence Rates for Maximum Likelihood Estimation under Finite Mixture Models (ICML 2022 · similarity 0.87)
- Subtractive Mixture Models via Squaring: Representation and Learning (ICLR 2024 · similarity 0.85)
- End-to-end Learning of Gaussian Mixture Priors for Diffusion Sampler (ICLR 2025 · similarity 0.85)
- Nonparametric Estimation of a Factorizable Density using Diffusion Models (NeurIPS 2026 · similarity 0.84)
- CoInD: Enabling Logical Compositions in Diffusion Models (ICLR 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
