Research map / Probabilistic and scientific modeling / Molecules and proteins
SAFE: A Neural Survival Analysis Model for Fraud Early Detection
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 |
|---|---|---|---|
| Use SAFE instead of SVM | fraud early detection · Twitter, first 5 timestamps | F1: 0.3875 -> 0.6537 | Helped |
| Use SAFE instead of CPH | fraud early detection · Twitter, first 5 timestamps | Accuracy: 0.5453 -> 0.7180 | Helped |
| Use SAFE instead of M-LSTM | fraud early detection · Wiki, first 5 timestamps | F1: 0.6556 -> 0.7866 | Helped |
| Use SAFE instead of SVM | fraud early detection · Wiki, first 5 timestamps | Accuracy: 0.6754 -> 0.7640 | Helped |
| Add the early-detection-oriented loss instead of the regular survival loss (ablation) | SAFE · Twitter, T5 | Accuracy: 0.60 -> over 0.75 | Helped |
| Avoid assuming a Rayleigh survival-time distribution (ablation) | SAFE · Twitter, first 5 timestamps | F1: 0.0025 -> 0.6537 | Helped |
| Avoid assuming a Poisson survival-time distribution (ablation) | SAFE · Twitter, first 5 timestamps | F1: 0.0024 -> 0.6537 | Helped |
| Avoid assuming an Exponential survival-time distribution (ablation) | SAFE · Twitter, first 5 timestamps | Accuracy: 0.5267 -> 0.7180 | Helped |
| Avoid assuming a Weibull survival-time distribution (ablation) | SAFE · Twitter, first 5 timestamps | F1: 0.3850 -> 0.6537 | Helped |
| Use SAFE rather than M-LSTM | Twitter fraud early detection · Twitter | percentage of early-detected fraudsters: 24% -> 82% | Helped |
| Use SAFE rather than M-LSTM | Twitter fraud early detection · Twitter | early-detected timestamps: 9.6 -> 11.1 | Helped |
Closest papers by meaning
- Online Fraud Detection via Test-Time Retrieval-Based Representation Enrichment (AAAI 2025 · similarity 0.83)
- Pre-trained Online Contrastive Learning for Insurance Fraud Detection (AAAI 2024 · similarity 0.83)
- ScamNet: Toward Explainable Large Language Model-Based Fraudulent Shopping Website Detection (AAAI 2025 · similarity 0.82)
- Dynamic Neighborhood Modeling via Node-Subgraph Contrastive Learning for Graph-Based Fraud Detection (AAAI 2025 · similarity 0.81)
- Unveiling the Threat of Fraud Gangs to Graph Neural Networks: Multi-Target Graph Injection Attacks Against GNN-Based Fraud Detectors (AAAI 2025 · similarity 0.81)
- Defense Against Model Stealing Based on Account-Aware Distribution Discrepancy (AAAI 2025 · similarity 0.81)
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.