ELEC 517 - ARCHITECTING ALGORITHMS
Long Title: ARCHITECTING MODERN LEARNING ALGORITHMS
Department: Electrical & Computer Eng.
Grade Mode: Standard Letter
Language of Instruction: Taught in English
Course Type: Lecture
Credit Hours: 3
Restrictions: Must be enrolled in one of the following Level(s):
Graduate
Description: This course focuses on architecture development and hardware realization of contemporary learning algorithms. A multitude of new learning algorithms have been recently developed, in particular in the sparse approximation domain. Thus far, the basic functionality of the new algorithms have been mostly verified and evaluated in simulation packages such as Matlab and software implementation. Application-specific customization and hardware implementation would bring orders-of-magnitude energy-performance efficiency improvement to important learning methods. The course will include FPGA reconfigurable fabric architecture and design flow, high analysis of multimedia processing VLSI architectures, and prototyping on FPGA. The focus of the project will be implementation of the state-of-the-art signal processing and learning algorithms on FPGA. Recommended Prerequisite(s): A digital logic design course and hands-on experience such as ELEC 326/ELEC 327, Background in VLSI, computer architecture, and signal processing/learning is also very useful, but the course is designed to be self-contained.