Data-Driven Anomaly Detectors for Time Series Data and Big Data (DD-ANDET)

The DD-AnDet project addresses the critical challenge of anomaly detection in time series and Big Data, a persistent issue in fields such as industrial maintenance, cybersecurity, and gesture recognition. Given the limitations of traditional methods when dealing with unlabeled data or dynamic environments, this research proposes the use of advanced deep learning techniques, including contrastive learning, multimodal learning, domain adaptation, and label shift. This international and multidisciplinary initiative integrates the expertise of researchers from Ecuador, Argentina, Brazil, and France. The primary focus is on creating robust models capable of identifying unusual behaviors in industrial machinery and dynamic gestures, such as those in sign language, thereby improving the accuracy and efficiency of monitoring systems. As a result, the project aims to generate high-quality experimental databases and publish scientific findings in indexed journals. Beyond its academic impact, the project has a clear social mission by promoting inclusion through the improvement of communication tools for the hearing impaired and by optimizing operational efficiency in Industry 4.0, reducing maintenance costs and downtime.<br/><br/><b>Goal</b>: <br/>To develop innovative anomaly detection models leveraging cutting-edge deep learning techniques for applications in predictive maintenance and dynamic gesture recognition. The project aims to advance the field through multidisciplinary approaches addressing challenges such as concept drift and domain adaptation.<br/><br/><b>Research lines</b>: <br/>Control engineering and automation technologies