PhD Position on Optimising Model Inference in Edge and IoT Environments with Transprecise Computing

Organisation/Company
Queen's University Belfast
Research Field
Computer science » Other
Researcher Profile
First Stage Researcher (R1)
Country
United Kingdom
Application Deadline
31 Jan 2025 - 00:00 (UTC)
Type of Contract
To be defined
Job Status
Negotiable
Is the job funded through the EU Research Framework Programme?
Not funded by an EU programme
Is the Job related to staff position within a Research Infrastructure?
No
Offer DescriptionReal-time remote monitoring of physiological indicators and early intervention can save lives. Providing these critical services requires wearable technologies with strong predictive abilities, fast networks, and fast servers to extract insights from the collected data. Unfortunately, these technology components are inaccessible to hundreds of millions of people, specifically, people who live in areas with limited broadband connectivity and limited means to invest in local computing and communication infrastructure.The aim of this PhD project is to investigate techniques that help address this gap by investigating novel techniques for serving and deployment of machine learning (ML) models, specifically in the context of health monitoring devices. This studentship is part of the project SWEET on Sustainable Wearable Edge Intelligence (SWEET), a collaboration between researchers from Queens University Belfast (UK), University College Dublin (Ireland) and Virginia Tech (USA).Deploying ML models on wearable or portable devices is challenging due to limitations on size, weight and power of the devices. Such constraints create a tension between accuracy of the model on one hand, and its size, inference delay and power consumption on the other hand. Alternatively, communicating captured data from wearable devices to the cloud requires fast and reliable communication channels, which again add significant power consumption demands for transmission. This problem reoccurs throughout numerous applications of edge-based intelligence, including traffic control, video surveillance, cattle monitoring, and in this instance health monitoring. The goal of this PhD project is to address the tension between computational demands and model accuracy by adapting the location(s) where inference is performed and the precision of computation and communication.The project will take a view on the stated problem through the lens of transprecise computing, which considers that not all computations need to be exact and that a disciplined and quantified trade-off between accuracy, speed, energy consumption and other performance and quality-of-experience metrics enables superior system designs. To this end, this project will: (i) build performance and utility models for health analytics across wearable and portable devices, edge computing and cloud computing infrastructure; (ii) design methods for deriving appropriate configurations of ML model inference across devices with varying design points for inference latency, power consumption, inference accuracy, etc; (iii) design online methods for inferring and enacting the most beneficial configurations at that time instance due to varying circumstances, such as signal quality, power availability, and computational workload.Fees and stipend supported only for UK students.
#J-18808-Ljbffr