Research map / Probabilistic and scientific modeling / Variational and generative sampling
Scaling Factorial Hidden Markov Models: Stochastic Variational Inference without Messages
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 |
|---|---|---|---|
| Replace SMF message passing with recognition-network-parameterised bivariate Gaussian copula chains trained by SGD | FHMM variational inference · Bach Chorales | test LL | Mixed |
| Replace SMF with proposed message-free recognition-network copula inference | FHMM on long simulated sequences with fixed 2hr budget · Simulated data, sequence length scaling | test LL | Helped |
| Replace SMF-EM with proposed algorithm trained on full 10^6-point household power data | 10-chain FHMM on household power consumption · Household power consumption | test MSE (all 6 dimensions): lower MSE in all six dimensions | Helped |
| Shared recognition networks and subchain SGD on simulated 2-dim data | Proposed algorithm vs SMF-EM on 2-chain FHMM · Simulated data validation | test LL: -2.332 vs -2.338 | No clear effect |
Closest papers by meaning
- Variational BOLT: Approximate Learning in Factorial Hidden Markov Models With Application to Energy Disaggregation (AAAI 2018 · similarity 0.89)
- Factorized Inference in Deep Markov Models for Incomplete Multimodal Time Series (AAAI 2020 · similarity 0.88)
- Markov Latent Feature Models (ICML 2016 · similarity 0.88)
- The Segmented iHMM: A Simple, Efficient Hierarchical Infinite HMM (ICML 2016 · similarity 0.87)
- FlowHMM: Flow-based continuous hidden Markov models (NeurIPS 2022 · similarity 0.87)
- Pre-trained Large Language Models Learn to Predict Hidden Markov Models In-context (NeurIPS 2025 · similarity 0.86)
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.