M.Sc. David Neumann

PhD Candidate and Research Associate in Machine Learning
working at the Fraunhofer HHI, Berlin, Germany
in the Artificial Intelligence Department

I'm a research scientist at Fraunhofer HHI, where I focus on developing efficient machine learning solutions. My research interests involve federated learning, recommender systems, neural network compression, and explainable AI, with a particular focus on developing novel techniques and algorithms. I have a strong foundation in computer science, having completed my Master's degree from Technische Universität Berlin in 2019. Prior to joining the research community, I worked as a software engineer on large-scale projects for various industries.

In addition to machine learning, I'm interested in software engineering and systems architecture, but also formal languages, automata and compiler construction – topics that underpin many modern software systems. As someone with entrepreneurial experience, I value innovation and collaboration in driving technical solutions. I'm currently pursuing my PhD while working on developing federated learning methods highly personalized and privacy-focused recommender systems.

Recent Publications

Software for dataset-wide XAI: From local explanations to global insights with Zennit, CoRelAy, and ViRelAy

January 2, 2026

Christopher J. Anders*, David Neumann*, Wojciech Samek, Klaus-Robert Müller, and Sebastian Lapuschkin (*equal contribution)

Published in PLOS ONE, volume 21, number 1.

@article{bibliography:software-for-dataset-wide-xai,
    author       = {Anders, Christopher J. and Neumann, David and Samek, Wojciech and Müller, {Klaus-Robert} and Lapuschkin, Sebastian},
    title        = {Software for dataset-wide {XAI}},
    subtitle     = {From local explanations to global insights with Zennit, {CoRelAy}, and {ViRelAy}},
    date         = {2026-01-02},
    doi          = {10.1371/journal.pone.0336683},
    journaltitle = {PLOS ONE},
    issn         = {2331-8422},
    volume       = {21},
    number       = {1},
    pagetotal    = {38},
    publisher    = {Public Library of Science},
    address      = {San Francisco, California, United States of America}
}

A Privacy Preserving System for Movie Recommendations Using Federated Learning

November 27, 2024

David Neumann, Andreas Lutz, Karsten Müller, and Wojciech Samek

Published in the special issue 3, number 2, “Trustworthy Recommender Systems” of the ACM Transactions on Recommender Systems (TORS) journal.

@article{bibliography:a-privacy-preserving-system-for-movie-recommendations-using-federated-learning,
    author       = {Neumann, David and Lutz, Andreas and Müller, Karsten and Samek, Wojciech},
    title        = {A Privacy Preserving System for Movie Recommendations Using Federated Learning},
    date         = {2024-11-27},
    doi          = {10.1145/3634686},
    journaltitle = {ACM Transactions on Recommender Systems},
    shortjournal = {TORS},
    issn         = {2770-6699},
    volume       = {3},
    number       = {2},
    issue        = {June 2025},
    issuetitle   = {Special Issue on Trustworthy Recommender Systems},
    eid          = {14},
    pagetotal    = {51},
    publisher    = {Association for Computing Machinery ({ACM})},
    address      = {New York City, New York, United States of America},
}

Finding and Removing Clever Hans: Using Explanation Methods to Debug and Improve Deep Models

August 3, 2021

Christopher J. Anders, Leander Weber, David Neumann, Wojciech Samek, Klaus-Robert Müller, and Sebastian Lapuschkin

Published in Elsevier Information Fusion, volume 77.

@article{bibliography:finding-and-removing-clever-hans,
    author       = {Anders, Christopher J. and Weber, Leander and Neumann, David and Samek, Wojciech and Müller, {Klaus-Robert} and Lapuschkin, Sebastian},
    title        = {Finding and removing Clever Hans},
    subtitle     = {Using explanation methods to debug and improve deep models},
    date         = {2021-08-03},
    doi          = {10.1016/j.inffus.2021.07.015},
    journaltitle = {Information Fusion},
    issn         = {1566-2535},
    volume       = {77},
    pages        = {261--295},
    pagetotal    = {35},
    publisher    = {Elsevier Science Publishers B. V.},
    address      = {PO Box 211 1000 AE Amsterdam, Netherlands}
}