Research map / Time series / Long-term forecasting
Memory in Memory: A Predictive Neural Network for Learning Higher-Order Non-Stationarity From Spatiotemporal Dynamics
Check 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 the temporal forget gate with cascaded MIM-N and MIM-S modules | MIM / ST-LSTM · Moving MNIST test | MSE: 56.8 -> 52.0 | Helped |
| Integrate MIM-N and MIM-S into Causal LSTM memory cells | Causal LSTM · Moving MNIST test | MSE: 46.5 -> 44.2 | Helped |
| Use cascaded MIM-N and MIM-S memory modules | Spatiotemporal prediction models · TaxiBJ | MSE: 0.318 -> 0.309; 0.427 -> 0.390; 0.516 -> 0.475; 0.595 -> 0.542 | Helped |
| Use cascaded MIM-N and MIM-S memory modules | Spatiotemporal prediction models · Radar Echo | MSE: 31.8 -> 27.8 | Helped |
| Use cascaded MIM-N and MIM-S memory modules | Spatiotemporal prediction models · Human3.6M Walking | MSE: 484.1 -> 429.9 | Helped |
| Add MIM-N to the MIM block (ablation) | MIM · Moving MNIST test | MSE: 54.4 -> 52.0 | Helped |
| Add MIM-S to the MIM block (ablation) | MIM · Moving MNIST test | MSE: 55.7 -> 52.0 | Helped |
| Add MIM-N to the MIM block (ablation) | MIM · Moving MNIST test | SSIM: 0.858 -> 0.874 | Helped |
| Add MIM-S to the MIM block (ablation) | MIM · Moving MNIST test | SSIM: 0.853 -> 0.874 | Helped |
| Use 3 MIM blocks instead of 2 (ablation) | MIM · Moving MNIST test | MSE: 54.6 -> 52.0 | Helped |
| Use 4 MIM blocks instead of 3 (ablation) | MIM · Moving MNIST test | MSE: 52.0 -> 53.7 | Hurt |
Closest papers by meaning
- Learning Spatiotemporal Dynamical Systems from Point Process Observations (ICLR 2025 · similarity 0.87)
- Graph-based Forecasting with Missing Data through Spatiotemporal Downsampling (ICML 2024 · similarity 0.86)
- Wavelet-Driven Spatiotemporal Predictive Learning: Bridging Frequency and Time Variations (AAAI 2024 · similarity 0.86)
- Learning with Calibration: Exploring Test-Time Computing of Spatio-Temporal Forecasting (NeurIPS 2025 · similarity 0.86)
- ST-ReP: Learning Predictive Representations Efficiently for Spatial-Temporal Forecasting (AAAI 2025 · similarity 0.86)
- Attractor Memory for Long-Term Time Series Forecasting: A Chaos Perspective (NeurIPS 2024 · similarity 0.85)
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.