↓Skip to main content
Jonas Emrich

Jonas Emrich

Master’s Student Bridging Engineering and AI/ML

I’m a Master’s student in Electrical Engineering & Information Technology, currently conducting my thesis at the Technical University of Munich (TUM) in collaboration with the Technical University of Darmstadt (TUDa). My specialization lies at the intersection of Signal Processing and Machine Learning / AI, covering a wide range of methods from model-based to data-driven approaches to solve various (interdisciplinary) problems. I am passionate about leveraging research and innovation to address real-world challenges and drive meaningful impact.

Previously, I worked as a research intern at BMW, where I contributed to research on generative AI for 3D object synthesis. Prior to that, I was part of the Robust Data Science Group as a research assistant developing novel methods for biomarker extraction in medical data. As a graduate exchange student, I was able to spend time at the Aalto University in Finland. During my studies, I worked over several years as a teaching assistant for several courses and participated actively in the student council as an elected member.

Furthermore, I was honored as a scholarship holder of the Studienstiftung des Deutschen Volkes and Thomas Weiland Foundation supporting young scientists with outstanding academic talents, and twice supported by the Deutschlandstipendium. For my bachelor graduation, I received the Best Student Award.

Publications

Physiology-Informed ECG Delineation Based on Peak Prominence
Jonas Emrich
, 
Andrea Gargano
, 
Taulant Koka
, 
Michael Muma
32st European Signal Processing Conference (EUSIPCO), 2024
Download Code Project Page
Accelerated Sample-Accurate R-Peak Detectors Based on Visibility Graphs
Jonas Emrich
, 
Taulant Koka
, 
Sebatsian Wirth
, 
Michael Muma
31st European Signal Processing Conference (EUSIPCO), 1090-1094, 2023
Video Code Project Page Cite
× 
              @inproceedings{Emrich_Accelerated_Sample-Accurate_R-Peak_2023,
    author = {Emrich, Jonas and Koka, Taulant and Wirth, Sebastian and Muma, Michael},
    booktitle = {2023 31st European Signal Processing Conference (EUSIPCO)},
    doi = {10.23919/EUSIPCO58844.2023.10290007},
    month = sep,
    pages = {1090--1094},
    series = {2023 31st European Signal Processing Conference (EUSIPCO)},
    title = {{Accelerated Sample-Accurate R-Peak Detectors Based on Visibility Graphs}},
    year = {2023}
}

Projects

Wild Bootstrap Tests
2023/24·seminar project
The classical bootstrap approach encounters challenges when the i.i.d. assumption is not fulfilled. In applications like defect detection in surface inspection, where the data is not identically distributed, and the noise distribution may depend on observational data, the Wild Bootstrap arises as a possible solution. In this seminarproject, we present the fundamental principles of the wild bootstrap and explain its procedure, illustrated by the application of (wild) bootstrap testing. The wild bootstrap test will be evaluated on simulated data and in the context of surface inspection by using real images.
Finalist Engineering Competition
2023·Hackathon·Rohde & Schwarz Engineering Competition
As a team of four students, we qualified for the finals in Munich by optimizing real-time spectrum analysis through the implementation of highly efficient algorithms on hardware simulators, including vector hardware & stream pipeline, using C++.
Mitosis Detection in Breast Cancer Histology Images using Deep Cascaded CNN
2022·course project
In this project, a deep learning-based approach for mitosis detection in breast cancer histology images was examined. The approach consists of two stages. In the first stage, a deep convolutional neural network (CNN) is utilized to detect mitosis candidates in the input image. In the second stage, a deep cascaded CNN discriminates between mitosis and non-mitosis. Transfer learning was employed to train both models due to the limited availability of large datasets.

Private Projects

Better TUCaN - Firefox Add-On
Browser Add-On
Firefox browser Add-On adding charts and advanced grade statistics as well as a dark theme to the campus management website www.tucan.tu-darmstadt.de
Download
Notenkurve.de
since 2016·Website / Webapplication
Notenkurve was originally developed as a web application in which students can enter their school grades and then get a visualization of their performance using a curve and further statistics.
However, the original web application is not being further developed and was taken offline due to concerns regarding the GDPR (DSGVO). Currently, several converters for students, such as the Punkte-in-Noten-Umrechner, are among the website’s most popular services, used by around 1 million people a year.