Research map / Time series / Causal and generative time series
Factorized Inference in Deep Markov Models for Incomplete Multimodal Time Series
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 BFVI instead of forward-RNN inference with zero-masking | MDMM · Noisy Spirals — reconstruction | MSE: 0.02 (0.01) vs 0.02 (0.01) | No clear effect |
| Use BFVI instead of backward-RNN inference with zero-masking | MDMM · Noisy Spirals — drop half | MSE: 0.04 (0.01) vs 0.04 (0.01) | No clear effect |
| Use BFVI instead of forward-RNN inference with zero-masking | MDMM · Noisy Spirals — forward extrapolation | MSE: 0.12 (0.10) vs 0.10 (0.08) | Hurt |
| Use BFVI instead of backward-RNN inference with zero-masking | MDMM · Noisy Spirals — backward extrapolation | MSE: 0.07 (0.03) vs 0.04 (0.01) | Hurt |
| Use BFVI instead of forward-RNN inference with zero-masking | MDMM · Noisy Spirals — conditional generation | MSE: 0.26 (0.26) vs 1.37 (1.39) | Helped |
| Use BFVI instead of backward-RNN inference with update skipping | MDMM · Weizmann Human Actions — reconstruction | SSIM: .85 (.03) vs .80 (.04) | Helped |
| Use BFVI instead of backward-RNN inference with zero-masking | MDMM · Weizmann Human Actions — drop half | SSIM: .84 (.04) vs .79 (.04) | Helped |
| Use BFVI instead of backward-RNN inference with update skipping | MDMM · Weizmann Human Actions — forward extrapolation | SSIM: .84 (.04) vs .80 (.04) | Helped |
| Use BFVI instead of backward-RNN inference with zero-masking | MDMM · Weizmann Human Actions — backward extrapolation | SSIM: .83 (.05) vs .79 (.04) | Helped |
| Use BFVI instead of backward-RNN inference with zero-masking | MDMM · Weizmann Human Actions — conditional generation | SSIM: .85 (.03) vs .76 (.06) | Helped |
| Use BFVI instead of backward-RNN inference with zero-masking | MDMM · Weizmann Human Actions — label prediction | Accuracy: .69 (.33)* vs .46 (.34)* | Helped |
Closest papers by meaning
- ChatTime: A Unified Multimodal Time Series Foundation Model Bridging Numerical and Textual Data (AAAI 2025 · similarity 0.86)
- Integrating Multimodal Data for Joint Generative Modeling of Complex Dynamics (ICML 2024 · similarity 0.85)
- Empowering Time Series Analysis with Large-Scale Multimodal Pretraining (NeurIPS 2026 · similarity 0.85)
- Deep Functional Factor Models: Forecasting High-Dimensional Functional Time Series via Bayesian Nonparametric Factorization (ICML 2024 · similarity 0.85)
- Rethinking Post-Training Recipes for Multimodal Time-Series Forecasting (NeurIPS 2026 · similarity 0.85)
- HDT: Hierarchical Discrete Transformer for Multivariate Time Series Forecasting (AAAI 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.