Machine Learning Spacecraft Designing for Cybersecurity
summary
Claims near perfect detection on validation sets using scikit-learn SVM and MLP, trained on about 660000 generated samples collected from pen test labs and generators, with lexical bigrams and tf-idf features. Sequential models (LSTM/GRU) slightly improve fit but a simple MLP runs ~200× faster, so product uses the MLP to meet real‑time constraints on ~1.5M packets/sec and reduce latency. Deployment focuses on a pipeline for parsing, feature extraction, online scoring and continuous monitoring to lower false positives.
AI-written summary · about
chapters (8)
- 0:00Introduction to cybersecurity with machine learning
- 1:36ML challenges in cybersecurity
- 7:20Detecting attacks via anomalies
- 12:05SQL injection attacks
- 15:22ML models for SQL injection
- 19:50Model complexity vs accuracy
- 20:31Deploying ML in production
- 22:18Summary & Q&A
related talks
32mAI/Machine Learning: What is Actually Working in Cybersecurity?
21mMachine Learning and Cyber Security - Detecting malicious URLs in the haystack
30mData Driven Security
16mProduction MLOps for Spam Detection: Real-Time Multi-Modal AI
01h 17mMachine Learning with Scikit-Learn
22mMachine Learning Based Attack Vector Modeling for CyberSecurity
48mMethod of detecting vulnerability in WebApps using Machine Learning
31mBuild threat detection models using ML