STA 209 - Optimization for Big Data Analytics

Subject: STA 209
Title: Optimization for Big Data Analytics
Units: 4.0
School: College of Letters and Science LS
Department: Statistics STA
Effective Term: 2018 Spring Quarter

Learning Activities

  • Lecture - 3.0 hours
  • Discussion - 1.0 hours

Description

Optimization algorithms for solving problems in statistics, machine learning, data analytics. Review computational tools for implementing optimization algorithms (gradient descent, stochastic gradient descent, coordinate descent, Newton’s method.).

Prerequisites

STA 200A; STA 208

Expanded Course Description

Summary of Course Content:

This course explores aspects of optimization for big data analytics. After providing a quick review of some fundamental optimization algorithms, the first part of the course aims to connect these optimization algorithms to statistics/machine learning problems. Then in the second part we will teach modern techniques and tools for handling large-scale problems, including MapReduce, Tensorflow, etc. The course will have a final project, where students can apply the techniques they learned in this class to solve a real world problem.

Illustrative Reading:

Convex Optimization (Boyd)

Potential Course Overlap:

MAT 160, MAT 168 and MAT 258A covers traditional optimization and their mathematical background, but does not cover the applications in statistics and big data analytics. MAT 258B covers discrete optimization, which is totally different from this course. STA 141C covers some statistical computing techniques but not optimization.

Final Exam:
No Final Exam