Research map / Probabilistic and scientific modeling / Causal discovery
Temporally Disentangled Representation Learning under Unknown Nonstationarity
Xiangchen Song, Weiran Yao, Yewen Fan, Xinshuai Dong, Guangyi Chen, Juan Carlos Niebles, Eric Xing, Kun Zhang
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 |
|---|---|---|---|
| use NCTRL instead of TDRL | NCTRL · Synthetic Dataset A | MCC: 83.99 -> 98.85 | Improved |
| use NCTRL instead of TDRL | NCTRL · Synthetic Dataset B | MCC: 72.02 -> 99.01 | Improved |
| model unknown nonstationarity with Markov domain indices plus time-delayed causal links instead of HMNLICA (no time-delayed relations) | NCTRL · Synthetic average | MCC: 58.54 -> 98.93 | Improved |
| use NCTRL instead of TDRL on CartPole | NCTRL · Modified CartPole | MCC: 85.26 -> 96.06 | Improved |
| use NCTRL instead of SKD (disentanglement without identifiability) on CartPole | NCTRL · Modified CartPole | MCC: 73.24 -> 96.06 | Improved |
| i-VAE moves from simple to complex nonlinear mixing (A to B) | i-VAE · Synthetic datasets | MCC: 89.74 -> 44.50 | Hurt |
| TDRL moves from simple to complex nonlinear mixing (A to B) | TDRL · Synthetic datasets | MCC: 83.99 -> 72.02 | Hurt |
| NCTRL moves from simple to complex nonlinear mixing (A to B) | NCTRL · Synthetic datasets | MCC: 98.85 -> 99.01 | No clear effect |
| use NCTRL on CartPole instead of synthetic data for c_t inference | NCTRL · c_t accuracy | accuracy: 89.96 -> 79.23 | Hurt |
| use NCTRL on CartPole instead of synthetic data for A estimation | NCTRL · MSE estimating A | MSE: 1.01 $\times 10^{-3}$ -> 5.01 $\times 10^{-2}$ | Hurt |
Closest papers by meaning
- Causal Temporal Representation Learning with Nonstationary Sparse Transition (NeurIPS 2024 · similarity 0.92)
- On the Identification of Temporal Causal Representation with Instantaneous Dependence (ICLR 2025 · similarity 0.92)
- Disentangled Representation Learning in Non-Markovian Causal Systems (NeurIPS 2024 · similarity 0.91)
- Causal Representation Learning Made Identifiable by Grouping of Observational Variables (ICML 2024 · similarity 0.91)
- Identifiable Latent Polynomial Causal Models through the Lens of Change (ICLR 2024 · similarity 0.91)
- Towards Identifiability of Hierarchical Temporal Causal Representation Learning (NeurIPS 2025 · similarity 0.91)
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.
