Research map / Probabilistic and scientific modeling / Dynamical systems
Prediction-Centric Learning of Independent Cascade Dynamics from Partial Observations
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 |
|---|---|---|---|
| Learn model parameters by maximizing the DMP-based marginal likelihood with SLICER rather than using true parameters for DMP inference | Independent Cascade on a square lattice · Marginal probability prediction, N=100, T=20 | relative ℓ1 marginal prediction error | Helped |
| Add replicas, initialized by perturbing the preceding replica solution (ablation) | SLICER effective model · Square lattice, full and partial observability (ξ=15%), N=100, M=10^6, T=20 | marginal prediction error | Helped |
| Initialize the multi-replica model by perturbing the previous solution instead of random initialization (ablation) | Multi-replica SLICER · Square lattice, N=100, M=10^6, T=20 | marginal prediction error | Helped |
| Hide high-degree rather than randomly selected nodes (ablation) | SLICER · Erdős–Rényi, N=100, T=10, M=10^5, ξ=15% | average ℓ1 parameter error: 0.039 -> 0.032 | Helped |
| Hide low-degree rather than randomly selected nodes (ablation) | SLICER · Erdős–Rényi, N=100, T=10, M=10^5, ξ=15% | average ℓ1 parameter error: 0.039 -> 0.047 | Hurt |
| Use SLICER instead of DMPrec for gradient evaluation | Independent Cascade parameter learning · 3-regular tree, runtime per gradient step | computational complexity: O(|E|^2 T M) -> O(|E| T M) | Helped |
| Use SLICER rather than DMPrec | Independent Cascade parameter recovery · 3-regular tree, N=20, T=5 | parameter recovery accuracy | No clear effect |
| Use SLICER rather than maximum likelihood | Independent Cascade parameter learning · Fully observed tree graphs | parameter recovery accuracy | No clear effect |
| Increase the observation horizon from T=5 to T=10 on loopy networks (ablation) | SLICER parameter estimation · Fully observed square lattice | parameter recovery error | Hurt |
| Increase the observation horizon in tree-like settings with partial observations (ablation) | SLICER parameter estimation · Regular trees and regular random graphs, ξ>0 | parameter reconstruction quality | Helped |
| Hide 25% of node activation times (ablation) | SLICER parameter estimation · Tree-like synthetic networks | parameter reconstruction quality | No clear effect |
| Use a multi-replica solution rather than a single-replica solution (ablation) | SLICER effective model · Square lattice, full and partial observability | marginal prediction error | Helped |
Closest papers by meaning
- Enhancing Multi-Scale Diffusion Prediction via Sequential Hypergraphs and Adversarial Learning (AAAI 2024 · similarity 0.88)
- Ghidorah: Towards Robust Multi-Scale Information Diffusion Prediction via Test-Time Training (AAAI 2025 · similarity 0.86)
- Training Data Attribution in Diffusion Models via Mirrored Unlearning and Noise-Consistent Skew (NeurIPS 2026 · similarity 0.85)
- Dynamical Diffusion: Learning Temporal Dynamics with Diffusion Models (ICLR 2025 · similarity 0.85)
- Learning Diffusions under Uncertainty (AAAI 2024 · similarity 0.85)
- Scalable Discrete Diffusion Samplers: Combinatorial Optimization and Statistical Physics (ICLR 2025 · similarity 0.85)
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.