Explainable AI: A 33× Leap in Fraud Detection
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisis ton rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
In the field of explainable artificial intelligence, a neuro-symbolic model has recently demonstrated significant advancements in fraud detection. Traditionally, the SHAP tool is used to explain fraud predictions, requiring 30 milliseconds to provide a stochastic explanation. This process occurs after the decision-making and necessitates the maintenance of a reference dataset at the time of inference.
In contrast, the neuro-symbolic model evaluated in this study offers a deterministic and human-readable explanation in just 0.9 milliseconds. This explanation is generated as a byproduct of the forward pass itself, representing an impressive speed gain of 33 times compared to SHAP.
This model was tested on the Kaggle credit card fraud dataset, and it was found that the fraud recall remains the same, despite the significant increase in speed.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.