Description:

Capsule networks (CapsNets) are a class of neural networks that address one of the main issues and deficiencies of CNNs. In CNNs, extracted features in initial layers pass through pooling layers where critical information, including spatial relationships, could be missed, resulting in incorrect classifications. CapsNets can better model hierarchical relationships and have shown great performance in many fields. CapsNets have a different architecture in their layers and hence, different computation complexity. This project investigates hardware accelerations for different layers of this class of neural networks for real-time applications.

Type:

Undergraduate

Outcome:

• Investigating state-of-the-art Capsule Networks and comparing their layer architectures in different versions.
• Investigating the computational complexities of Capsule Networks and their computational intensity in different layers.
• Developing a hardware implementation of CapsNet layers on the FPGA platform. 

Prerequisites

15 points from COMPSYS 305 

15 points from COMPSYS 304 or enrolment in COMPSYS 701 

Students should have a very strong background in digital systems design and have a good understanding of computer system architectures. Competency with Electronic Design Automation (EDA) tools is required. They should be willing to learn to work with new design tools from AMD.

Specialisations

• Computer Systems Engineering

Categories

• AI & Machine learning
• Embedded Systems

Supervisor

• Maryam Hemmati

Co-supervisor

Team

• Jerome Jose
• Nicholas Wolf

Lab

Embedded Systems (405.760, Lab)