Leveraging Assumptions of Simplicity When Handling Outliers and Missing Data

Many modern engineering, signal processing, and statistical inference problems experience the blessing and curse of big data. Large datasets give us the promise of understanding complex worldly phenomena in order to help us make predictions and decisions. Unfortunately, big data inevitably turns out to be messy, with corruptions, missing data, and heterogeneous sensing modalities. In this talk I will discuss signal processing approaches to overcoming this challenge of massive low-quality data.