Research map / Video / Video segmentation and tracking
Adaptive Skip Intervals: Temporal Abstraction for Recurrent Dynamical Models
Alexander Neitz, Giambattista Parascandolo, Stefan Bauer, Bernhard Schölkopf
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 adaptive skip intervals instead of fixed one-step or two-step prediction | Recurrent dynamics model · Funnel board | validation accuracy | Improved |
| Use adaptive skip intervals instead of fixed one-step or two-step prediction | Recurrent dynamics model · Room runner | validation accuracy | Improved |
| Add exploration of temporal matching (ablation) | ASI · Funnel board | validation accuracy | Improved |
| Add exploration of temporal matching (ablation) | ASI · Room runner | validation accuracy | Improved |
| Replace adaptive skip intervals with fixed one-step prediction (ablation) | Recurrent dynamics model · Funnel board and Room runner | validation accuracy | Hurt |
| Replace adaptive skip intervals with fixed two-step prediction (ablation) | Recurrent dynamics model · Funnel board and Room runner | validation accuracy | Hurt |
| Use a strided or dilated convolutional architecture variant (ablation) | ASI dynamics model · Room runner and Funnel board | validation accuracy | No clear effect |
| Switch training from zero-bounciness funnels to standard funnel physics | Fixed-frame-rate baselines · Funnel board with perturbed dynamics | validation accuracy | Hurt |
| Switch training from zero-bounciness funnels to standard funnel physics | ASI · Funnel board with perturbed dynamics | validation accuracy | Mixed |
Closest papers by meaning
- Recurrent Reinforcement Learning with Memoroids (NeurIPS 2024 · similarity 0.83)
- MaRK: Markov-adapted Recurrent Kernels for Dynamic Operator Conditioning in State Space Models (NeurIPS 2026 · similarity 0.82)
- Continuity Laws for Sequential Models (NeurIPS 2026 · similarity 0.82)
- Temporal-Difference Variational Continual Learning (NeurIPS 2025 · similarity 0.82)
- Exploring the Promise and Limits of Real-Time Recurrent Learning (ICLR 2024 · similarity 0.82)
- End-to-End Identifiable and Consistent Recurrent Switching Dynamical Systems (NeurIPS 2026 · similarity 0.82)
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.
