Overview
What this challenge is about.
You receive the classifier (a PyTorch ResNet variant) and a 4,000-image labeled validation slice. Apply randomized smoothing (Cohen et al.) at sigma in {0.25, 0.5, 1.0}. Report certified accuracy at L2 radii 0.25, 0.5, 1.0. Estimate inference cost and sample complexity for the certification. Discuss limitations honestly. Deliver the certification notebook, results table, and a 4-page regulatory-affairs memo with a clear claim about what the bound does and does not guarantee.
The Brief
What you'll do, and what you'll demonstrate.
Produce certified-robustness numbers for a medical-imaging classifier via randomized smoothing and document the regulatory claim honestly.
Earning criteria — what you'll demonstrate
- Apply randomized smoothing to a real classifier
- Quantify certified accuracy at multiple L2 radii
- Estimate the cost of certification at production scale
- Write regulatory-grade claims about what a robustness bound does and does not guarantee
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Research Scientist
Producing certified-robustness numbers for a regulated medical product is exactly the work that lands research scientists at safety-critical AI shops.
This challenge sharpens
- certified-robustness
- randomized-smoothing
- formal-verification
AI Safety Researcher
Translating provable bounds into honest regulatory claim language is the AI safety researcher's craft in healthcare AI.
This challenge sharpens
- certified-robustness
- regulatory-documentation
- randomized-smoothing
ML Researcher
Sample-complexity analysis and cost estimation for certification is core ML-researcher work at scale-deployed ML.
This challenge sharpens
- certified-robustness
- uncertainty-quantification
- pytorch