Skip to content
#

threat-prediction

Here are 3 public repositories matching this topic...

Language: All
Filter by language

Hybrid LSTM-Markov attack chain forecasting for MITRE ATT&CK. Learns from 4,849 campaign chains + 8,437 real intrusion traces. Generates 26,051 risk-ranked multi-step attack futures via constrained beam search. 86% next-step accuracy, 0.76 Pearson correlation with NCISS severity. SECRYPT 2026 submission.

  • Updated Jun 23, 2026
  • Python

Add this topic to your repo

To associate your repository with the threat-prediction topic, visit your repo's landing page and select "manage topics."

Learn more