The Arctic Green Computing group is looking for Masters students to work on Capstone and Masters projects in energy informatics (e.g., energy-efficient computing). Please contact us if you are interested in the projects and are a Masters student at the Department of Computer Science, UiT
Below are some of the available topics:
Optimizing the trade-off between energy efficiency and accuracy in approximate computing
Energy consumption is becoming a major concern across the full spectrum of computing systems from mobile devices (i.e., battery life) to data centers (i.e. energy cost). Energy consumption can be reduced using approximate computing that allows to trade accuracy for energy efficiency. Approximate computing is applicable to several application domains such as big data applications.
This project will investigate the trade-off between energy efficiency and accuracy in approximate computing and subsequently optimize it. The trade-off will be investigated and optimized at different levels (e.g., application and system) and with different objectives (i.e. minimizing energy consumption or providing guarantees on energy consumption).
Low-power continuous 3D face authentication for mobile devices
Face authentication, which leverages built-in cameras on mobile devices, provides a memory-less alternative to legacy passwords. The 3D face authentication is one of the liveness detection techniques for face authentication to defend against the media-based facial forgery (MFF) where the attacker forges a photo/video containing the victim’s face. The 3D face authentication detects the depth characteristics of a real face using optical flow analysis and changes of face views. Due to its optical flow analysis, the continuous 3D face authentication consumes a lot of energy for mobile devices, limiting its deployment in mobile applications.
This project will investigate and develop low-power continuous 3D face authentication, making the authentication more viable for mobile applications. The new low-power continuous 3D face authentication will utilize the ultra low-power machine vision technology developed by Movidius for mobile devices. The new authentication will be developed on Movidius Myriad2 platform.