Chapter 19: EM-Based Methods and Hyperparameter Estimation

Learning Objectives

• Derive EM-GAMP for joint signal and hyperparameter estimation in RF imaging
• Implement EM updates for noise variance, sparsity rate, and signal variance
• Formulate generalized linear models for 1-bit CS, Poisson, and power-only measurements
• Understand the output function g_out that replaces the Gaussian likelihood in GAMP
• Analyze multi-layer VAMP for scenes drawn from deep generative priors