How to Approximate Inference with Subtractive Mixture Models
Not on the map. Why?
Abstract (arXiv)
Classical mixture models (MMs) are widely used tractable proposals for approximate inference settings such as variational inference (VI) and importance sampling (IS). Recently, mixture models with negative coefficients, called subtractive mixture models (SMMs), have been proposed as a potentially more expressive alternative. However, how to effectively use SMMs for VI and IS is still an open question as they do not provide latent variable semantics and therefore cannot use sampling schemes for classical MMs. In this work, we study how to circumvent this issue by designing several expectation estimators for IS and learning schemes for VI with SMMs, and we empirically evaluate them for distribution approximation. Finally, we discuss the additional challenges in estimation stability and learning efficiency that they carry and propose ways to overcome them. Code is available at: https://github.com/april-tools/delta-vi.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 36 of its 91 references that are in the map. First posted on arXiv 17 Apr 2026.
Builds on
36 of 91 references.
References · 36
- 2015Importance Weighted Autoencoderscited by 456
- 2019Monte Carlo Gradient Estimation in Machine Learningcited by 283
- 2016Automatic Differentiation Variational Inferencecited by 159
- 2018Neural Importance Samplingcited by 155
- 2022Variational inference via Wasserstein gradient flowscited by 85
- 2024Beyond ELBOs: A Large-Scale Evaluation of Variational Methods for Samplingcited by 56
- 2018Importance Weighting and Variational Inferencecited by 55
- 2018Yes, but Did It Work?: Evaluating Variational Inferencecited by 53
- 2020Non-parametric Models for Non-negative Functionscited by 41
- 2016On the Latent Variable Interpretation in Sum-Product Networkscited by 40
- 2023Subtractive Mixture Models via Squaring: Representation and Learningcited by 39
- 2021Variational Refinement for Importance Sampling Using the Forward Kullback-Leibler Divergencecited by 29
24 more
- 2024BlackJAX: Composable Bayesian inference in JAXcited by 26
- 2021PSD Representations for Effective Probability Modelscited by 23
- 2016Boosting Variational Inferencecited by 22
- 2024Batch and match: black-box variational inference with a score-based divergencecited by 22
- 2024What is the Relationship between Tensor Factorizations and Circuits (and How Can We Exploit it)?cited by 22
- 2024Sum of Squares Circuitscited by 21
- 2020Probabilistic Circuits for Variational Inference in Discrete Graphical Modelscited by 20
- 2023How to Turn Your Knowledge Graph Embeddings into Generative Modelscited by 20
- 2020Automatic Differentiation Variational Inference with Mixturescited by 19
- 2015Layered Adaptive Importance Samplingcited by 18
- 2024On the Relationship Between Monotone and Squared Probabilistic Circuitscited by 14
- 2019Variational approximations using Fisher divergencecited by 13
- 2021Adaptive Importance Sampling meets Mirror Descent: a Bias-variance tradeoffcited by 13
- 2022Multiple Importance Sampling ELBO and Deep Ensembles of Variational Approximationscited by 12
- 2015Leave Pima Indians alone: binary regression as a benchmark for Bayesian computationcited by 11
- 2024EigenVI: score-based variational inference with orthogonal function expansionscited by 11
- 2025A Probabilistic Neuro-symbolic Layer for Algebraic Constraint Satisfactioncited by 11
- 2016Langevin Incremental Mixture Importance Samplingcited by 5
- 2021Monotonic Alpha-divergence Minimisation for Variational Inferencecited by 5
- 2025How to Square Tensor Networks and Circuits Without Squaring Themcited by 5
- 2014Learning Negative Mixture Models by Tensor Decompositionscited by 3
- 2023Adaptive importance sampling for heavy-tailed distributions via α-divergence minimizationcited by 3
- 2024Simulating signed mixturescited by 2
- 2025Approximate Bayesian Inference via Bitstring Representationscited by 1
Built on by
Citing papers per year: 2026: 1.
1 paper.