Research map / Time series / Causal and generative time series
PSA-GAN: Progressive Self Attention GANs for Synthetic Time Series
Read the paperCheck 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 |
|---|---|---|---|
| Use PSA-GAN to impute missing lag observations during far-forecasting | DeepAR · Far-forecasting | NRMSE | Helped |
| Use PSA-GAN to impute missing-value stretches during training | DeepAR · Missing-value stretches | NRMSE | Helped |
| Use PSA-GAN to impute lags and context for cold-start series | DeepAR · Cold starts | NRMSE: 0.70 | Helped |
| Use PSA-GAN samples for data augmentation | DeepAR · Data augmentation | NRMSE: 0.49 -> 0.64; 0.26 -> 0.3; 1.07 -> 1.07; 0.45 -> 0.50 | Hurt |
| Add 64 previous values as context | PSA-GAN · Direct forecasting across four datasets | NRMSE: 1.53 -> 1.18; 0.61 -> 0.33; 1.35 -> 1.32; 0.66 -> 0.57 | Helped |
| Remove moment loss (ablation) | PSA-GAN · Context-FID ablation | Context-FID | Hurt |
| Remove self-attention (ablation) | PSA-GAN · Context-FID ablation | Context-FID | Hurt |
| Remove progressive fade-in of new layers (ablation) | PSA-GAN · Context-FID ablation | Context-FID | Hurt |
| Select models using Context-FID | Time-series GAN development · Forecasting tasks | NRMSE: 0.71 Pearson; 0.67 Spearman | Helped |
Closest papers by meaning
- Generative Modeling of Regular and Irregular Time Series Data via Koopman VAEs (ICLR 2024 · similarity 0.87)
- Time Series Generation Under Data Scarcity: A Unified Generative Modeling Approach (NeurIPS 2025 · similarity 0.86)
- TEMPO: Prompt-based Generative Pre-trained Transformer for Time Series Forecasting (ICLR 2024 · similarity 0.85)
- Timer: Generative Pre-trained Transformers Are Large Time Series Models (ICML 2024 · similarity 0.85)
- Market-GAN: Adding Control to Financial Market Data Generation with Semantic Context (AAAI 2024 · similarity 0.85)
- Time Weaver: A Conditional Time Series Generation Model (ICML 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.