Tutorial 02
From RSS to XYZ: Building Machine Learning Pipelines for Indoor Localization Using Python
Abstract
Artificial intelligence has become ever-present in everyday life, as well as in research communities across disciplines. Understanding the inner mechanics of machine learning (ML) is nowadays not necessary to deploy such models, yet expert knowledge of the domain, as well as the hyperparameters, may strongly improve the model performance. This tutorial’s main goal is to demonstrate the applicability of various ML models on different levels of expertise, enabling researchers to effortlessly realize and validate their ideas.
Tutors
Roman Klus is a Doctoral Researcher at Tampere University, Finland, preparing to defend his PhD thesis in the upcoming days -- on 5th September. His academic journey began with an M.Sc. in Electronics and Communications from Brno University of Technology (2019), and has led him into cutting-edge research on machine learning for 5G and beyond. In the past years, work focused on neural networks for mobility management and wireless positioning, but recently, he is involved in the projects dealing with neural transmitters for improved signal characteristics and leveraging generative models and LLMs for RF fingerprint identification to enhance robustness in wireless communication. He have had the privilege of enriching his research through international stays at Jaume I University in Castellón, Spain, where he collaborated on indoor localization topics leading to several papers as well as measuring and publishing a fingerprinting dataset, and UCLA in Los Angeles, USA, where his research revolved around localization in the near field region of mmWave system with large antenna arrays. Both experiences greatly broadened his perspective, expertise in numerous fields, and collaboration network.