#### Contact:

Instructor: Rina Foygel Barber (rina@uchicago.edu, office: Eckhart 113)

#### Hours/locations:

Class:
• Tue/Thu 10:30-11:50am, Eckhart 117

Office hours:
• Tue 12-2pm or by appointment

#### Topics covered:

The material covered in this course will depend on student interest, & will include:
• Sparse signals & applications
• Conditions on the measurement matrix for compressed sensing; random measurement matrix
• Greedy selection & orthogonal matching pursuit
• Combinatorial group testing
• L1 minimization for sparse signal recovery
• Algorithms for L1 minimization
• Other types of sparsity: block-wise sparsity, fused Lasso, etc.
• Low-rank matrices & matrix completion
• Demixing structured signals: Robust PCA & other examples
• Applications for each topic will be presented & included in HW.

#### Course information:

• Prerequisites: familar with linear algebra & probability theory
• Homework will be assigned every 1-2 weeks, including both theory and programming. HW assignments will include small projects using publicly available data sets. There will be no exams.
• We will use Matlab in this course (contact instructor if you don't currently have access to Matlab).

#### Resources:

Links to many theory / algorithm / application papers: http://dsp.rice.edu/cs
Boyd & Vandenberghe, Convex Optimization. See Appendix A for a reference on norms, gradients, linear algebra, etc.

What does compressive sensing mean for X-ray CT and comparisons with its MRI application (lecture by Emil Sidky).

#### Assignments:

HW1. A matlab function for HW1: OGA.m.

HW2. Some matlab code for HW2: HW2_one_bit_CS_code.m. And, an optional additional problem for HW2.

HW3. Matlab data for HW3: HW3_HIVdata.mat. A matlab function you'll need: ADMM group lasso.

HW4. Link to video data: videos. Matlab code for HW4: get_video_data.m, admm_video.m.