Research map / Time series / Causal and generative time series
Learning Conditional Generative Models for Temporal Point Processes
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 |
|---|---|---|---|
| add conditional Wasserstein loss to MLE loss (ablation) | Seq2Seq · MIMIC | Pr.Dev.: 0.17 -> 0.10 | Helped |
| add conditional Wasserstein loss to MLE loss (ablation) | Seq2Seq · LinkedIn | Pr.Dev.: 0.14 -> 0.11 | Helped |
| add conditional Wasserstein loss to MLE loss (ablation) | Seq2Seq · IPTV | Pr.Dev.: 1.19 -> 0.95 | Helped |
| add conditional Wasserstein loss to MLE loss (ablation) | Seq2Seq · NYSE | Pr.Dev.: 1.55 -> 1.23 | Helped |
| add conditional Wasserstein loss to MLE loss (ablation) | Seq2Seq · Synthetic IP, SE, SC | Pa.Dev.: 0.31 -> 0.09 (IP); 0.19 -> 0.02 (SE); 0.12 -> 0.07 (SC) | Helped |
| add conditional Wasserstein loss to MLE loss (ablation) | Seq2Seq · Synthetic mixtures | Pr.Dev.: 0.93 -> 0.76 (IP+SE); 0.48 -> 0.31 (IP+SC); 0.92 -> 0.63 (SC+SE) | Helped |
| use neural seq2seq CWE instead of correctly specified parametric MLE model | CWE · Synthetic IP | Pa.Dev.: 0.03 -> 0.09 | Hurt |
| use CWE instead of misspecified parametric MLE models | CWE · Real-world datasets | Pr.Dev.: 0.15 -> 0.10 (MIMIC vs MLE-SE) | Helped |
| vary Wasserstein-loss weight trade-off sigma (too small or >10) (ablation) | CWE · Synthetic SE | Pa.Dev. | Mixed |
| vary Lipschitz regularization weight lambda (ablation) | CWE · Synthetic SE | Pa.Dev. | No clear effect |
| compare scheduled sampling to Seq2Seq | SS · Real-world datasets | Pr.Dev.: 1.19 -> 1.13 (IPTV) | Helped |
Closest papers by meaning
- Interacting Diffusion Processes for Event Sequence Forecasting (ICML 2024 · similarity 0.86)
- Deep Continuous-Time State-Space Models for Marked Event Sequences (NeurIPS 2025 · similarity 0.85)
- EasyTPP: Towards Open Benchmarking Temporal Point Processes (ICLR 2024 · similarity 0.84)
- Distance Marching for Generative Modeling (NeurIPS 2026 · similarity 0.84)
- Realism VS Accuracy: Event Sequence Forecasting from a Generative Modeling Perspective (NeurIPS 2026 · similarity 0.84)
- Time Series Generation Under Data Scarcity: A Unified Generative Modeling Approach (NeurIPS 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.