Research map / Graph learning / Graph self-supervised learning
Sentient: Detecting APTs via Capturing Indirect Dependencies and Behavioral Logic
Wenhao Yan, Ning An, Wei Qiao, Weiheng Wu, Zhigang Lu, Bo Jiang, Baoxu Liu, Junrong Liu
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 |
|---|---|---|---|
| add graph-transformer pre-training (ablation) | Sentient · DARPA E3 Cadets | precision: 20.75% precision drop without it | Improved |
| add Bi-Mamba2 Intent Analysis Module (ablation) | Sentient · DARPA E3 Cadets | precision: 31.59% precision drop without it | Improved |
| add 1000 benign events as mimicry attack | Sentient · DARPA E3 Cadets | F-score: 97.83% -> 97.16% | No clear effect |
| add 2000 benign events as mimicry attack | Sentient · DARPA E3 Cadets | F-score: 97.83% -> 96.90% | No clear effect |
| add 3000 benign events as mimicry attack | Sentient · DARPA E3 Cadets | F-score: 97.83% -> 97.31% | No clear effect |
| Sentient vs Threatrace (ablation) | Sentient · Unicorn Wget | false positive rate: 7.4% -> 4.1% | Improved |
| Sentient vs Flash | Sentient · DARPA E3 Theia | false positive rate: 0.8% -> 0.4% | Improved |
| Sentient vs Slot | Sentient · DARPA E3 Cadets | false positive rate: 0.2% -> 0.2% | No clear effect |
| increase node embedding dimension beyond 64 (ablation) | Sentient · DARPA E3 Cadets | accuracy/precision/recall | Hurt |
| increase behavior sequence length beyond 30 (ablation) | Sentient · DARPA E3 Cadets | accuracy/precision/recall | No clear effect |
Closest papers by meaning
- DRIFT: Dynamic Rule-Based Defense with Injection Isolation for Securing LLM Agents (NeurIPS 2025 · similarity 0.82)
- Attractive Metadata Attack: Inducing LLM Agents to Invoke Malicious Tools (NeurIPS 2025 · similarity 0.82)
- IRIS: LLM-Assisted Static Analysis for Detecting Security Vulnerabilities (ICLR 2025 · similarity 0.82)
- The Case for Learned Provenance-based System Behavior Baseline (ICML 2025 · similarity 0.82)
- When Agents Go Rogue: Activation-Based Detection of Malicious Behaviors in Multi-Agent Systems (ICML 2026 · similarity 0.81)
- PragLocker: Protecting Agent Intellectual Property in Untrusted Deployments via Non-Portable Prompts (ICML 2026 · 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.
