Inductive Logic Programming for a Fraud-Rule Discovery Pilot
Overview
What this challenge is about.
You receive a labeled fraud dataset (around 25,000 transactions, around 4% positive class), a feature schema (28 features including device, geo, behavioral history), and a baseline LightGBM model with held-out precision-recall numbers. Use an ILP system (Aleph or Popper) to learn a set of first-order rules; you may need to discretize continuous features into a Prolog-friendly form. Report rule-set precision, recall, and rule readability (subjective rating + average rule length). Compare against the LightGBM baseline on the same held-out set. Write a 2-page memo on whether ILP earns a seat in the production fraud stack.
The Brief
What you'll do, and what you'll demonstrate.
Quantify whether Inductive Logic Programming surfaces useful, auditable fraud rules that complement a gradient-boosted baseline.
Earning criteria — what you'll demonstrate
- Apply Inductive Logic Programming to a real labeled dataset
- Discretize continuous features for symbolic learners
- Evaluate rule-based ML on precision, recall, and readability
- Reason about combining symbolic and statistical models in production
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.
ML Researcher
Applying symbolic ML methods to a labeled dataset and writing the production-fit memo is the kind of methodology work ML researchers ship in industry research labs.
This challenge sharpens
- inductive-logic-programming
- symbolic-ai
- rule-learning
Data Scientist
Combining symbolic rule-learning with a GBM in a fraud-detection stack is exactly the day-one work of a data scientist at any fintech with regulator pressure.
This challenge sharpens
- fraud-detection
- evaluation
- rule-learning
AI Safety Researcher
Surfacing auditable rules to complement a black-box model is the AI safety craft of building accountable ML systems.
This challenge sharpens
- symbolic-ai
- rule-learning
- fraud-detection