Research map / Time series / Foundation models for time series
Geometry-Aware Visual Representation for Remaining Useful Life Prediction
Hieu Vu, Thanh Nguyen, Eyad Elyan
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 |
|---|---|---|---|
| add global healthy-reference anchoring to phase-space reconstruction (ablation) | PSDI + KD · Case 1 (bearing RUL) | RMSE: 0.128 -> 0.099 | Improved |
| add global anchoring (ablation) | PSDI + KD · Case 3 (bearing RUL) | RMSE: 0.116 -> 0.071 | Improved |
| add global anchoring (ablation) | PSDI + KD · Case 2 (bearing RUL) | RMSE: 0.084 -> 0.107 | Hurt |
| add global anchoring (ablation) | PSDI + KD · Case 5 (bearing RUL) | RMSE: 0.062 -> 0.060 | No clear effect |
| add global anchoring (ablation) | PSDI + KD · Feature space clustering | Silhouette: 0.101 -> 0.516 | Improved |
| use PSDI instead of wavelet image (ablation) | KD framework · Case 1 (bearing RUL) | RMSE: 0.119 -> 0.099 | Improved |
| use PSDI instead of wavelet image (ablation) | KD framework · Case 2 (bearing RUL) | RMSE: 0.093 -> 0.107 | Hurt |
| use PSDI instead of STFT image (ablation) | VGG · Case 1 (bearing RUL) | RMSE: 0.354 -> 0.154 | Improved |
| use PSDI instead of wavelet image (ablation) | VGG · Case 5 (bearing RUL) | RMSE: 0.062 -> 0.134 | Hurt |
| use KD framework instead of standalone backbone (ablation) | PSDI · Case 1 (bearing RUL) | RMSE: 0.154 -> 0.099 | Improved |
| distill from MAE-Large rather than MAE-Base teacher (ablation) | EfficientNet-B0 student · Case 1 (bearing RUL) | RMSE: 0.108 -> 0.141 | Hurt |
| add compact student with sliding-window aggregation over teacher alone (ablation) | CLIP teacher · Case 1 (bearing RUL) | RMSE: 0.170 -> 0.099 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TimeMixer++ — TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting (ICLR 2024) · better on 4 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ODEWorld: A Continuous Predictive Architecture via Physical-Time Flow (NeurIPS 2026 · similarity 0.82)
- Generative Image Dynamics (CVPR 2024 · similarity 0.81)
- Hierarchical State Space Models for Continuous Sequence-to-Sequence Modeling (ICML 2024 · similarity 0.81)
- Geometry-Centered 3D Latent World Models for Growing Surfaces (NeurIPS 2026 · similarity 0.81)
- Identifying latent state transitions in non-linear dynamical systems (ICLR 2025 · similarity 0.81)
- Uncertainty Visualization via Low-Dimensional Posterior Projections (CVPR 2024 · similarity 0.81)
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.
