Ghent, Flemish Region, Belgium
Check out some of the things I've built at https://tomasfiers.net
Topic: From Voltage to Wiring – Synaptic connectivity inference from neural voltage recordings
I work in the lab of Fabian Kloosterman (Neuro-Electronics Research Flanders, at imec), who researches how the brain represents information and how memories guide behaviour. I am developing an enhanced algorithm that detects, in real-time, when the brain is replaying a memory, using data from voltage sensors in the hippocampus.
Work out business plan for joint venture with Belgian renewable energy firm.
Byteflies makes wearables – small, sensing devices that can be worn anywhere on the body. One of the things they can measure is motion (via accelerometers, gyroscopes and magnetometers). My job was to analyse this motion data. To quantify how it holds up against a golden baseline (3D motion capture), and to research how to best use the data. This entailed your typical signal processing / data science work: parsing, filtering and transforming the raw time series data, performing exploratory data analysis, aligning different data sets, making lots of visualisations, and performing some basic machine learning (labelling data, calculating features, reducing feature space dimensionality, applying a supervised learning algorithm, and reporting on its performance). I helped in the design of new experiments. I also did research on representing 3D orientations mathematically, on the physical properties of the different sensors, and on how to combine their strengths for robust motion tracking. My main output were a set of thematically organised and richly documented Jupyter Notebooks (using NumPy, scikit-learn, Matplotlib, etc, and running in a Docker container). I communicated my results both to a technical and a non-technical audience, in the form of a report, a presentation, and a blog post. The results of my work were also presented at a congress in the US.