Research map / Time series / Causal and generative time series
UNIPoint: Universally Approximating Point Processes Intensities
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 a flexible UNIPoint intensity representation instead of restricted Hawkes or RMTPP forms | UNIPoint · MOOC and Reddit | holdout log-likelihood | Helped |
| Use a mixture of power-law and ReLU basis functions | UNIPoint · Across evaluated datasets | holdout log-likelihood | Helped |
| Increase the number of basis functions (ablation) | PLSum · MOOC | holdout log-likelihood | Helped |
| Replace the standard RNN with an LSTM (ablation) | PLSum · MOOC | holdout log-likelihood: 4.939 ± .085 → 4.934 ± .086 | Hurt |
| Replace the standard RNN with an LSTM (ablation) | PLSum · Reddit | holdout log-likelihood: 0.162 ± .046 → 0.212 ± .048 | Helped |
| Replace the standard RNN with an LSTM (ablation) | PLSum · StackOverflow | holdout log-likelihood: -0.814 ± .023 → -0.791 ± .023 | Helped |
| Replace softplus with max(sigmoid(x), x) (ablation) | UNIPoint · Real-world datasets, across basis variants | holdout log-likelihood | Mixed |
| Replace softplus with max(sigmoid(x), x) (ablation) | CosSum · MOOC, Reddit, and StackOverflow | holdout log-likelihood: 4.471 ± .075 → 3.353 ± .153; 0.139 ± .044 → 0.151 ± .045; -0.814 ± .023 → -0.814 ± .022 | Hurt |
| Use one Monte Carlo integration sample instead of 256 (ablation) | ReLUSum · MOOC | holdout log-likelihood difference: 0.005 ± 0.019 vs - | No clear effect |
Closest papers by meaning
- Decomposable Transformer Point Processes (NeurIPS 2024 · similarity 0.86)
- Exact, Fast and Expressive Poisson Point Processes via Squared Neural Families (AAAI 2024 · similarity 0.86)
- Unlocking Point Processes through Point Set Diffusion (ICLR 2025 · similarity 0.86)
- Neural Jump-Diffusion Temporal Point Processes (ICML 2024 · similarity 0.86)
- EasyTPP: Towards Open Benchmarking Temporal Point Processes (ICLR 2024 · similarity 0.86)
- Hyper Hawkes Processes: Interpretable Models of Marked Temporal Point Processes (NeurIPS 2026 · similarity 0.86)
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.