Signal Processing for Data Analysis

4 credits

This course will focus on ideas from classical and modern signal processing, with the main themes of sampling continuous data and building informative representations using orthonormal bases, frames, and data dependent operators. Topics include sampling theory, Fourier analysis, lossy transformations and compression, time and spatial filters, and random Fourier features and connections to kernel methods. Sources of data include time series and streaming signals and various imaging modalities. Prerequisites: (MATH 18 or MATH 31AH) and (MATH 20C or MATH 31BH) and DSC 80/R. Restricted to students with upper-division standing. Restricted to students within the data science or business data science major. All other students will be allowed as space permits.