Publications & Presentations

Analysis of a Variable Region Size in Regional Voting

September 2025 · Oral presentation

IEEE UIC 2025Hirt, J. and Chen, L.Honours thesis research

Consensus methods can look stable in theory and still fail when noise is concentrated in space, the kind of corruption that shows up in real sensors and imaging pipelines. This paper studies how changing the region size in regional voting affects stability, and shows that multi-scale ensembles are a practical way to stay reliable under those conditions.

Oral presentation: deriving the multi-region stability formula
At the podium during the IEEE UIC oral session

Key takeaways

  • Independently derived mathematical theorems and proofs for election stability under concentrated noise.
  • Designed multi-scale voting ensembles to mitigate spatially correlated data corruption and filter corrupt data.
  • Connected the theory to practical failure modes (sensor artifacts / spatially correlated noise) and identified an “edge of stability” across noise configurations.
  • Delivered an oral presentation at the IEEE 22nd International Conference on Ubiquitous Intelligence and Computing (UIC).
Paper PDF (self-hosted)IEEE XploreRelated project write-up