The Embedded AI Research Lab is an initiative from the Department of Computer Systems in close cooperation with the Departments of Software Science, Electronics and Health Technologies at the Tallinn University of Technology. We are promoting IoT ecosystem, which is vibrant and at the same time innovative. We develop new technologies and apply them in the industry to give you a competitive edge. Our aim is to create innovative applications and domain capability across verticals for industry needs (Smart City, Smart Health, Smart Industry, Smart Agriculture, etc). We have been a part of projects such as Iseauto and Smart Workwear and currently we are developing Smart Cardeck and Smart Elevator, among other things.
No open positions
PhD position in deep learning algorithm optimization for embedded systems
Machine learning model attack by inserting data with hidden noise or malicious data.
Result is a confidence reduction, misclassification or targeted classification. Possible application could also be a malfunctional sensor detection.
To analyze the effect of those anomalies there is a need for development of a machine learning model debugger.
This tool should help to analyze the validity of the results of each layer and optimize them, if possible.
Postdoc position in machine learning attack methods, analysis and detection
Deep learning algorithm optimization for low cost embedded hardware. Goal is to reduce the computational requirements, lower memory footprint or
decrease power consumption, depending on the particual application. Optimization could be done on different hardware, like microcontrollers,
FPGAs, neural processing units and GPU based hardware. As a result there should be a set of optimized models and best practices that could be
integrated into one embedded ML model optimization tool that speeds up the development of smart sensor.