Research map / Probabilistic and scientific modeling / Variational and generative sampling
Variational Inference for Sequential Data with Future Likelihood Estimates
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use future likelihood estimates as baselines and replace high-variance term (VIFLE estimator) vs. VIMCO/REINFORCE | IWAE gradient estimator on discrete-latent state-space models (synthetic Bernoulli dynamics) · Synthetic 20-dim Bernoulli dynamics, sequence length 10-200 | ELBO per time step (log scale) | Helped |
| Use future likelihood estimates only as baselines (VIFLE_U) instead of full VIFLE (ablation) | IWAE score-function gradient estimator · Synthetic dynamics | ELBO per time step | No clear effect |
| Replace high-variance term with future-likelihood-based term (full VIFLE) on top of baseline (ablation) | IWAE gradient estimator · Synthetic dynamics | ELBO per time step | Helped |
| Full replacement (FR) of all particle coefficients using future likelihoods | IWAE gradient estimator · Synthetic 20-dim Gaussian dynamics, 200-step sequences | ELBO per time step | Hurt |
| Add VIFLE gradient estimator to IWAE training of continuous-latent SSM (N=4) | continuous latent SSM on polyphonic music · JSB / Nottingham / MuseData / Piano-midi.de | test marginal log-likelihood bound | Mixed |
| Add VIFLE gradient estimator to IWAE training of discrete-latent SSM (N=4) | discrete latent SSM on polyphonic music · JSB / Nottingham / MuseData / Piano-midi.de | test marginal log-likelihood bound | Helped |
Closest papers by meaning
- Variational Inference for Monte Carlo Objectives (ICML 2016 · similarity 0.92)
- Marginalized Stochastic Natural Gradients for Black-Box Variational Inference (ICML 2021 · similarity 0.92)
- Robust, Accurate Stochastic Optimization for Variational Inference (NeurIPS 2020 · similarity 0.91)
- Bayesian Online Natural Gradient (BONG) (NeurIPS 2024 · similarity 0.90)
- Variational Inference with Mixtures of Isotropic Gaussians (NeurIPS 2025 · similarity 0.90)
- SoftCVI: Contrastive variational inference with self-generated soft labels (ICLR 2025 · similarity 0.89)
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.