How to Make Sure Your Data Science Isn't Vulnerable to Attack

Explore data science security principles and best practices in this 57-minute conference talk from BSidesLV 2016. Delve into data exploration preparation, algorithm application, and effective communication techniques. Learn how to establish strong foundations, implement proper naming conventions, and conduct thorough analysis. Discover methods for detecting new vulnerabilities, handling metadata, and addressing data quirks. Gain insights on balancing caveats with usability, understanding different data perspectives, and extracting actionable insights from vulnerability assets. Examine the importance of basic statistics, common pitfalls to avoid, and lessons learned in measuring risk. Enhance your ability to secure data science projects and communicate findings effectively across various infosec contexts.

Intro
What is data science
Data exploration preparation principles
Applying the algorithms
Communication
Machine Learning
Strong Foundations
Measure
Naming conventions
Detect
Analysis
New detection
New vulnerabilities
Metadata
Data quirks
Check new detection
How to communicate
Data flow in infosec
Balance between caveats and usability
Different perspectives on the data
Vulnerability assets
Actual insight
Actionable insight
Beyond your data set
Other data sets
Get more data
The takeaway
Importance of basic statistics
Common mistakes
Lessons learned
Measuring risk