Research map / Time series / Foundation models for time series
Olivia: Harmonizing Time Series Foundation Models with Power Spectral Density
Jingru Fei, Kun Yi, Alex Xing Wang, Qingsong Wen, Xiangxiang Zhu, Wei Fan
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 Harmonizer (ablation) | Olivia · ETTh2 horizon 96 zero-shot | MSE: 0.294 -> 0.274 | Improved |
| add Harmonizer (ablation) | Olivia · ETTm2 horizon 96 zero-shot | MSE: 0.215 -> 0.193 | Improved |
| add Harmonizer to SEMPO (ablation) | SEMPO · ETTh1 horizon 96 zero-shot | MSE: 0.389 -> 0.382 | Improved |
| add Harmonizer to SEMPO (ablation) | SEMPO · Electricity horizon 96 zero-shot | MSE: 0.169 -> 0.161 | Improved |
| use HarmonicAttention instead of Full Attention (ablation) | Olivia · ETTh1 zero-shot | MSE: 0.472 -> 0.399 | Improved |
| use HarmonicAttention instead of Linear Attention (ablation) | Olivia · ETTh1 zero-shot | MSE: 0.412 -> 0.399 | Improved |
| use HarmonicAttention instead of Nystrom Attention (ablation) | Olivia · Electricity zero-shot | MSE: 0.216 -> 0.188 | Improved |
| use HarmonicAttention instead of Full Attention (ablation) | Olivia · Weather zero-shot | MSE: 0.268 -> 0.247 | Improved |
| Olivia vs SEMPO on ETTm2 | Olivia · TSLib ETTm2 zero-shot | MSE: 0.289 -> 0.291 | No clear effect |
| Olivia vs SEMPO inference cost | Olivia · ETTh1 zero-shot | inference time: 8.162 -> 43.051 | Hurt |
| Olivia vs SEMPO on Weather | Olivia · TSLib Weather zero-shot | MSE: 0.248 -> 0.247 | No clear effect |
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.
- SEMPO — SEMPO: Lightweight Foundation Models for Time Series Forecasting (NeurIPS 2025) · better on 8 of 9 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
- TimeCHEAT: A Channel Harmony Strategy for Irregularly Sampled Multivariate Time Series Analysis (AAAI 2025 · similarity 0.85)
- Time Series Generation Under Data Scarcity: A Unified Generative Modeling Approach (NeurIPS 2025 · similarity 0.84)
- SimPSI: A Simple Strategy to Preserve Spectral Information in Time Series Data Augmentation (AAAI 2024 · similarity 0.84)
- Introducing Spectral Attention for Long-Range Dependency in Time Series Forecasting (NeurIPS 2024 · similarity 0.84)
- Revisiting the Generic Transformer: Deconstructing a Strong Baseline for Time Series Foundation Models (NeurIPS 2026 · similarity 0.84)
- Towards a General Time Series Forecasting Model with Unified Representation and Adaptive Transfer (ICML 2025 · similarity 0.84)
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.
