.png){kind=logo}
Confiance.ai
Desensitize industrial data from sensors
Challenges
- Use data collected by sensors, made available by a member of the Confiance.ai program, fortraining anomaly detection models
- Anonymizing complex data: time series, a data type that is omnipresent in industrial environments
Maintaining statistical quality and utility
Anonymization of signals with peak detection: anonymization alters the most unique curves and manages to maintain the different modes present in the data.
A model trained on avatar data makes it possible to detect anomalies on real data as well as a model trained on original data.
ROI
- Development of an AI model on data maintaining strategic information such as manufacturing secrets
- Maintaining the utility data to ensure the reproducibility of models
- Acceleration technological innovation through the analysis of secure data
- Details of the ROIs in a white paper [link here] on the use case, paving the way for more secure and effective management of sensitive information
“The work carried out by Octopize and Sopra Steria makes it possible to remove obstacles to the use of Machine Learning in areas where data privacy is essential, while guaranteeing their security thanks to concrete metrics, necessary to build trust.” - Yves Nicolas, AI group Program Director, Sopra Steria
“The avatar of data opens the way to statistical exploitation while minimizing the risks of exposure of sensitive data, which allows a paradigm shift in data management.” - Alexis Rouet, Chief Data Officer HR, Renault Group
“Together, we are exploring the application of Octopize's avatar anonymization method, in particular for training Machine Learning algorithms, and plan to share the results in a future white paper on AI and Defense.” - Marine Martinez, Program Lead Cyber @StationF, Thales
Our customers
All our customers
