ECE64700 - Convex and Stochastic Optimization and Applications
Convex optimization and dynamic programming with application to various disciplines in ECE, including wireless and wireline communications, networks, electrical systems, and machine learning.
Credit Hours: 3
Instructor:
Learning Objective:
Be equipped with the basic theory of convex optimization, Markov decision processes, (approximate) dynamic algorithms, and be able to apply these tools to research problems.
Description:
Convex optimization and dynamic programming with application to various disciplines in ECE, including wireless and wireline communications, networks, electrical systems, and machine learning. The goal is to model a research problem as an optimization problem, and use the insights from the optimization algorithms to design implementable protocols and solutions that are tailored to the needs of the applications (e.g., operating in decentralized settings and under stochastic uncertainty). The instructor may tailor the content to reflect current trends and goals of research. Examples of relevant topics are: convex sets and functions, convex program, Lagrange duality, primal and dual decomposition, proximal optimizations, stochastic approximation, Markov decision processes, dynamic programming and approximate dynamic programming. The exposition of the theory will be connected to various ECE applications, such as congestion control, wireless power control and opportunistic scheduling, cross-layer design, demand response, renewable integration, spectrum sharing and cognitive radio, machine learning, optimization of UAV (unmanned aerial vehicles) aided communications.
Prerequisite: ECE 60000