Research map / Probabilistic and scientific modeling / Molecules and proteins
HYPRO: A Hybridly Normalized Probabilistic Model for Long-Horizon Prediction of Event Sequences
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 AttNHP by hybrid HYPRO when first error at event 1 | AttNHP · Taobao | subsequent event type error rate (%): 71.28\% -> 66.99\% | Helped |
| replace AttNHP by hybrid HYPRO when first error at event 5 | AttNHP · Taobao | subsequent event type error rate (%): 58.88\% -> 51.89\% | Helped |
| replace AttNHP by hybrid HYPRO when first error at event 10 | AttNHP · Taobao | subsequent event type error rate (%): 44.48\% -> 35.58\% | Helped |
| scale up NHP and AttNHP to HYPRO parameter count (ablation) | NHP/AttNHP · Taobao, Taxi, StackOverflow | RMSE and OTD | No clear effect |
| use Multi-NCE instead of Binary-NCE (ablation) | HYPRO · three datasets | RMSE and OTD | Helped |
| add distance-based regularization Omega, Binary-NCE (ablation) | HYPRO · Taobao, Taxi | RMSE and OTD: p-value $<0.05$ | Helped |
| add distance-based regularization Omega, Multi-NCE (ablation) | HYPRO · Taobao, Taxi | RMSE and OTD: p-value $\approx 0.1$ | No clear effect |
| use continuous-time LSTM as energy function instead of Transformer (ablation) | HYPRO · Taobao, Taxi | RMSE and OTD | Hurt |
| increase training negative samples from 1 to 5 (ablation) | HYPRO hypro-a-b · Taobao | RMSE and OTD | Helped |
| increase training negative samples from 5 to 10 (ablation) | HYPRO hypro-a-b · Taobao | RMSE and OTD | No clear effect |
| increase inference samples from 5 to 20 (ablation) | HYPRO hypro-a-b · Taobao | RMSE and OTD | Helped |
| increase inference samples beyond 20 (ablation) | HYPRO hypro-a-b · Taobao | RMSE and OTD | No clear effect |
Closest papers by meaning
- Realism VS Accuracy: Event Sequence Forecasting from a Generative Modeling Perspective (NeurIPS 2026 · similarity 0.86)
- Interacting Diffusion Processes for Event Sequence Forecasting (ICML 2024 · similarity 0.85)
- Adaptive Meta-Learning Probabilistic Inference Framework for Long Sequence Prediction (AAAI 2024 · similarity 0.85)
- EasyTPP: Towards Open Benchmarking Temporal Point Processes (ICLR 2024 · similarity 0.84)
- Autoregressive Learning in Joint KL: Sharp Oracle Bounds and Lower Bounds (NeurIPS 2026 · similarity 0.83)
- BARNN: A Bayesian Autoregressive and Recurrent Neural Network (ICML 2025 · similarity 0.83)
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.