Do Software Architecture Patterns Reduce Security Vulnerabilities? xB;Insight from Causal Learning
Abstract
Motivation for Causal Learning: Controlling costs requires knowing which independent factors actually cause item outcomes, so that we may change items in a predictable manner. Just as correlation may be fooled by spurious association, so can regression. We must move beyond correlation to causation, if we want to make use of cause and effect relationships. We can now evaluate causation without expensive and difficult experiments. Establishing causation with observational data remains a vital need and a key technical challenge, but is becoming more feasible and practical.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 26, 2019
- Accession Number
- AD1085200
Entities
People
- Michael D. Konrad
- Rick Kazman
- Robert W. Stoddard
- William R. Nichols
Organizations
- Carnegie Mellon University