Research map / Time series / Causal and generative time series
Integration-free Training for Spatio-temporal Multimodal Covariate Deep Kernel Point Processes
YIXUAN ZHANG, Quyu Kong, Feng Zhou
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 deep kernel DMPP with deep-kernel DKMPP trained by score matching (RQ kernel) (ablation) | DMPP · Crimes in Vancouver | ACC (%): 51.52 -> 80.16 | Improved |
| add multimodal covariates and score-based estimator versus DeepSTPP | DeepSTPP · NYC Vehicle Collisions | TLL: 396.60 -> 411.03 | Improved |
| compare against HomoPoisson | Score-DKMPP (RQ) · NYC Vehicle Collisions | ACC (%): 66.811 -> 79.42 | Improved |
| use Score-DKMPP+ (denoising) instead of Score-DKMPP with OU kernel (ablation) | Score-DKMPP · NYC Complaint Data | TLL: 3.06 -> 4.28 | Improved |
| use Score-DKMPP+ (RBF) instead of Score-DKMPP (RBF) (ablation) | Score-DKMPP · NYC Vehicle Collisions | TLL: 409.02 -> 402.54 | Hurt |
| increase representative points from 64 to 125 (ablation) | Score-DKMPP · synthetic | RMSE: 3.86 -> 3.66 | Improved |
| increase representative points from 125 to 216 (ablation) | Score-DKMPP · synthetic | RMSE: 3.66 -> 3.66 | No clear effect |
| increase layers from 1 to 4 (ablation) | Score-DKMPP · synthetic | RMSE: 3.84 -> 4.13 | Hurt |
| increase layers from 1 to 4 (ablation) | Score-DKMPP+ · synthetic | RMSE: 3.78 -> 4.08 | Hurt |
| batch size 100 versus 200 (ablation) | Score-DKMPP · synthetic | RMSE: 4.07 -> 3.47 | Improved |
| batch size 50 versus 100 (ablation) | Score-DKMPP+ · synthetic | RMSE: 3.67 -> 3.61 | No clear effect |
| increase noise variance from 0.01 to 10 (ablation) | Score-DKMPP+ · synthetic | intensity recovery | Hurt |
Closest papers by meaning
- Spatio-temporal point processes with deep non-stationary kernels (ICLR 2023 · similarity 0.88)
- State-of-art minibatches via novel DPP kernels: discretization, wavelets, and rough objectives (NeurIPS 2026 · similarity 0.86)
- Beyond Point Prediction: Score Matching-based Pseudolikelihood Estimation of Neural Marked Spatio-Temporal Point Process (ICML 2024 · similarity 0.84)
- Inducing Clusters Deep Kernel Gaussian Process for Longitudinal Data (AAAI 2024 · similarity 0.83)
- Sparse Inducing Points in Deep Gaussian Processes: Enhancing Modeling with Denoising Diffusion Variational Inference (ICML 2024 · similarity 0.83)
- Seahorse: A Unified Benchmarking Framework for Spatiotemporal Event Modeling (NeurIPS 2026 · 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.
