Using Machine Learning to Help with Daily Workflows
summary
A Drupal 8 custom module labels web server logs (HTTP status, full user agent) from several Drupal sites, uses a human labeling interface with three tags (good/bad/unsure) to build a training set, and trains a classifier that emits a risk score per IP so admins set thresholds to block. Dataset strips PII; commercial alternatives like PerimeterX may misclassify API callers. Prototype runs locally on CSVs and reaches about 89% accuracy; plan adds behavioral-history features.
AI-written summary · about
chapters (4)
- 0:00Start
- 13:17Humans write software. Humans have biases Software carries these biases. Assumptions made will affect decisions
- 34:35Practice Run
- 37:01Conclusion
related talks
21mMachine Learning and Cyber Security - Detecting malicious URLs in the haystack
01h 01mD1T2 - Applying Machine Learning to User Behavior Anomaly Analysis
44mApplied Machine Learning for Data Exfil and Other Fun Topics
39mDetecting Website Intrusion and Account Compromise with Machine Learning
48mCatch me if you can: Building a Web Malware Analyzer using Machine Learning
30mData Driven Security
23mBot or human? Detecting malicious bots with machine...
42mTesting Machine Learning Modules in Production