XTRACTIS for Safety & Security

Passive Magnetic Identification of Land Mines

Benchmark vs. Logistic Regression, Random Forests, Boosted Trees & Neural Networks

Design an AI-based decision system that accurately detects land mines and identifies the type of mine only from a few variables, to deliver the appropriate decision rationally and instantly.

Goals & benefits

Find the cause-and-effect relationships between the relevant predictors among the 3 parameters of this study, and the actual presence of a mine and its type.

Enhance demining technicians and military experts’ knowledge by understanding the strategies to identify the type of mine.

Help engineers to design enhanced mine detectors, manual or autonomous, embedding a classifier making explainable and accurate automated decisions.

Assist the military profession in making a more reliable decision, thanks to rapid, systematic, explainable, and safer detection process with a passive magnetic sensor.


We get a Predictive Model that is:


A Decision System composed of 29 unchained gradual rules aggregated into 3 disjonctive fuzzy rules; each rule uses 1 to 3 variables that XTRACTIS confirmed as significant.


Good Real Performance on External Test.

Efficient & Operational.

Running in real-time up to 70,000 predictions per second (i7 @2.5GHz with 8 physical cores), offline or online (API).

UC16 graphe benchmark

LoR=Logistic Regression | RFo=Random Forests | BT=Boosted Trees | NN=Neural Networks

Detailed results and explanations in full document.

Use Case 2023/10 (v2.0)

Powered by XTRACTIS® REVEAL 13.0.44978 (2023/02)


  1. Problem Definition
  2. XTRACTIS-induced Decision System
  3. XTRACTIS Process
  4. Top-Model Induction
  5. Explained Predictions for 3 unkown cases
  6. Top-Models Benchmark
  7. Appendices