Conference Agenda

We propose a tutorial around the usage and applicability of ML in EMC. The aim would be to highlight the best practices when using ML in the context of EMC and related topics.

As ML models are data-driven, particular attention must be paid to data capture and processing. The data which is typically used in EMC applications (from measurement or simulation) is treated from the point of view of the ML. Therefore, we will insist on the expertise needed to build adequate experimental designs in order to obtain reliable data, and to carry out the appropriate data processing.

After presenting strategies to collect the data correctly, we will talk about the methodology for assessing the data's potential to address our problem. Then, we will focus on how to select the most suitable ML models. In particular, we will present the pros and cons of the most common ML models. In addition, we will talk about the different steps of the process, including training, validation, and testing, and emphasizing the importance of using appropriate evaluation metrics.

Finally, we will show several examples related to EMC challenges where everything we spoke about was put into practice, taking into account industrial constraints. These examples underline the need for expertise in both EMC and ML to make them compatible.