OPELIS: Artificial Intelligence for the protection of sensitive data

In a context marked by the exponential growth of data and the proliferation of strict privacy regulations, companies require tools capable of identifying and protecting sensitive information in an automated way. OPELIS emerges as a response to this need, ensuring data sovereignty and digital security within the business ecosystem.

The project, as parto f the the Basque Government’s Hazitek programme, has focused its activity on optimizing the lifecycle of Deep Learning models through Active Learning and Online Learning techniques. This significantly reduces human effort in the creation and updating of models, shortens computation times, and enables a more efficient use of expert knowledge in the detection of critical data.

To achieve this, Vicomtech has developed an architecture that integrates active and continuous learning services, accessible via APIs. This solution makes it possible to efficiently identify the most relevant data within datasets, optimize model training, and perform incremental updates in production. To this end, algorithms such as Elastic Weight Consolidation (EWC) are used, avoiding the need for full retraining while ensuring continuous adaptation of models to new data.

The OPELIS project has been led by Serikat, demonstrating its capacity for innovation in artificial intelligence and cybersecurity. With this initiative, the company reaffirms its commitment to developing high-level proprietary technological solutions, meeting the standards required by the Hazitek programme and consolidating its leadership in advanced digital consulting services.

With OPELIS, Vicomtech continues to contribute to the optimization of industrial processes and the continuous improvement of security, applying its technology to deliver efficient and innovative solutions that drive productivity and operational efficiency.