Adversarial Machine Learning for Radar

The focus of this presentation will be "AI/Counter AI" (also known as "Adversarial Machine Learning or AML") for Radar object recognition. It has been noted that artificial intelligence / machine learning (AI/ML) model developed using a particular dataset may not yield desired classification accuracy when data are altered/manipulated/perturbed intentionally or unintentionally. Hence, a robust AI/ML model development based on clear understanding of AI/ML theory is a mandatory. Many AI/ML developers for radar object recognition lack deep understanding of radar signature phenomenology and thus their models are not robust and reliable. This talk will highlight various technical issues involving Radar AI/ML model development.