Research map / Architectures and efficiency / Attention and transformers
Transformer Hawkes Process
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 |
|---|---|---|---|
| Replace recurrent sequence modeling with THP self-attention | Point-process event sequences · Retweets | per-event log-likelihood: -4.56 -> -2.04 | Helped |
| Replace recurrent sequence modeling with THP self-attention | Point-process event sequences · MemeTrack | per-event log-likelihood: -6.04 -> 0.68 | Helped |
| Replace recurrent sequence modeling with THP self-attention | Point-process event sequences · Financial | per-event log-likelihood: -3.60 -> -1.11 | Helped |
| Replace recurrent sequence modeling with THP self-attention | Point-process event sequences · MIMIC-II | per-event log-likelihood: -0.52 -> 0.820 | Helped |
| Replace recurrent sequence modeling with THP self-attention | Point-process event sequences · StackOverflow | per-event log-likelihood: -1.86 -> 0.042 | Helped |
| Add temporal encoding to NHP (ablation) | NHP · Retweets | validation log-likelihood: -5.60 -> -2.50 | Helped |
| Add temporal encoding to NHP (ablation) | NHP · MemeTrack | validation log-likelihood: -6.23 -> -1.64 | Helped |
| Add positional encoding to the attention model (ablation) | Atten · Retweets | validation log-likelihood: -5.29 -> -5.25 | No clear effect |
| Add positional encoding to the attention model (ablation) | Atten · MemeTrack | validation log-likelihood: -5.09 -> -4.70 | Helped |
| Add temporal encoding to the attention model (ablation) | Atten · Retweets | validation log-likelihood: -5.29 -> -2.03 | Helped |
| Add temporal encoding to the attention model (ablation) | Atten · MemeTrack | validation log-likelihood: -5.09 -> 0.68 | Helped |
| Share information across all vertices using full attention | THP on 911-Calls and Earthquake · 911-Calls and Earthquake | event prediction | Mixed |
Closest papers by meaning
- Hyper Hawkes Processes: Interpretable Models of Marked Temporal Point Processes (NeurIPS 2026 · similarity 0.88)
- The Neural Hawkes Process: A Neurally Self-Modulating Multivariate Point Process (NeurIPS 2017 · similarity 0.88)
- EasyTPP: Towards Open Benchmarking Temporal Point Processes (ICLR 2024 · similarity 0.85)
- Decomposable Transformer Point Processes (NeurIPS 2024 · similarity 0.85)
- Interacting Diffusion Processes for Event Sequence Forecasting (ICML 2024 · similarity 0.84)
- ExPERT: Modeling Human Behavior Under External Stimuli Aware Personalized MTPP (AAAI 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.