Researchers have developed a new framework called the Group Attention Neural Hawkes Process (GAttNHP) to improve forecasting for temporal knowledge graphs (TKGs). This model addresses challenges in encoding long-range temporal dependencies, modeling mutual excitation between event chains, and handling heavy-tailed inter-arrival times. GAttNHP utilizes a self-attention encoder, a semantic soft-grouping module, and a Non-Crossing Quantile regression head to achieve better performance on entity and time prediction tasks across six benchmark datasets. AI
IMPACT This new model could improve predictive accuracy for time-series data in knowledge graphs, potentially impacting fields that rely on forecasting future events.
RANK_REASON The cluster contains a research paper published on arXiv detailing a new model.
- alphaXiv
- arXiv
- CatalyzeX
- DagsHub
- GAttNHP
- Gotit.pub
- Group Attention Neural Hawkes Process
- Hawkes process
- Hugging Face
- Non-Crossing Quantile Regression for Distributional Reinforcement Learning
- ScienceCast
- temporal knowledge graph
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
