Research map / Time series / Causal and generative time series
Predicting User Activity Level In Point Processes With Mass Transport Equation
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 MC with HYBRID at equal 10^3 samples (ablation) | HYBRID · Twitter user activeness | MAPE: 4x error reduction | Helped |
| Use 10^6 MC samples instead of 10^3 for MC | MC · Twitter user activeness | MAPE: MC-1e3 0.13 -> MC-1e6 0.69 (rank corr.) | Helped |
| Account for all stochasticity (HYBRID) vs heuristic baselines SEISMIC and FPE | HYBRID · Twitter user activeness | MAPE: 2x more accurate | Helped |
| HYBRID with 60% vs baselines needing 80% training data | HYBRID · Twitter user activeness | MAPE vs train size | Helped |
| Increase test time horizon | all methods · Twitter user activeness | MAPE | Hurt |
| Increase training data proportion | all methods · Twitter user activeness | MAPE | Helped |
| HYBRID vs MC to reach error 0.1 | HYBRID · Twitter user activeness | computation time: 27.8 hours -> 14.4 minutes | Helped |
| HYBRID vs MC-1E6 to reach error 0.1 | HYBRID · IPTV item popularity | computation time: 7.5 hours -> 9.8 minutes | Helped |
| HYBRID vs FPE and SEISMIC | HYBRID · IPTV and Reddit item popularity | MAPE: 15% more accurate than FPE and 20% than SEISMIC | Helped |
| HYBRID vs FPE in rank correlation | HYBRID · IPTV and Reddit item popularity | rank correlation: 0.78 vs 0.51 (IPTV), 0.77 vs 0.58 (Reddit) | Helped |
| HYBRID vs MC on number of samples for f(x)=x | HYBRID · Synthetic Hawkes | samples to reach error 0.01: 10^5 -> 10^3 | Helped |
| HYBRID vs MC on number of samples for f(x)=x log(x) | HYBRID · Synthetic Hawkes | samples to reach error 0.01: 10^6 -> 10^4 | Helped |
Closest papers by meaning
- Learning Registered Point Processes from Idiosyncratic Observations (ICML 2018 · similarity 0.82)
- EasyTPP: Towards Open Benchmarking Temporal Point Processes (ICLR 2024 · similarity 0.82)
- Listwise Preference Diffusion Optimization for User Behavior Trajectories Prediction (NeurIPS 2025 · similarity 0.81)
- Unlocking Point Processes through Point Set Diffusion (ICLR 2025 · similarity 0.80)
- Deep Continuous-Time State-Space Models for Marked Event Sequences (NeurIPS 2025 · similarity 0.80)
- Decomposable Transformer Point Processes (NeurIPS 2024 · similarity 0.80)
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.