Detecting fraud and improving monitoring

Using predictive and pattern recognition algorithms, we can detect normal and abnormal behavior, identify anomalies and optimize regulatory activities.

Detection of irregular land use by satellite images

The Departmental Directorate of Territories and Sea (DDTM) of Herault is in charge of detecting illegal land occupations and has called upon our teams to develop a tool to assist in protecting the territory. We did this by seeing irregular land occupations using satellite images.

Detection of irregular land use through image recognition.
Construction of an image annotation tool to train the model.
Development of a scoring model, taking into account the propensity of a parcel of land in Herault to contain an illegal construction.
Development of a web interface for the restitution of the algorithm's detection results.

Detecting cyber attacks in government from event log analysis using AI

Our client is responsible for securing digital development and detecting computer attacks within public administrations. Our teams propose new approaches and exploit AI models to automate time-consuming manual detection tasks.

Detect abnormal behavior missed by traditional approaches
Prioritize alerts to reduce the number of false positives.
Provide analysts with interpretability elements to explain the reasons for alerts
Provide a library of generic methods for detecting atypical behavior to be integrated into the customer's ecosystem.

Our publications

- Government

Labeling text clusters with keywords

We propose to explore several keyword extraction techniques to label text clusters obtained after a Text Clustering or a Topic Modeling pipeline. This work is following our previous articles about Topic Modeling and Text Clustering (here and here).

2024