Jan Disselhoff

PhD student, Johannes-Gutenberg Universität Mainz

Joined

May 2022

Names

Jan Disselhoff (Preferred)

Emails

****@uni-mainz.de (Confirmed)

,

****@gmail.com (Confirmed)

Personal Links

Career & Education History

PhD student

Johannes-Gutenberg Universität Mainz (uni-mainz.de)

2022 – Present

Advisors, Relations & Conflicts

PhD Advisor

2022 – Present

Coworker

2020 – Present

Coworker

2020 – 2023

Expertise

No areas of expertise listed

Publications

Probing the Inductive Bias of Neural Networks through Learning Random Cellular Automata
Jan Disselhoff, Michael Wand
- Submitted to NeurIPS 2025
- Readers: Everyone
Product of Experts with LLMs: Boosting Performance on ARC Is a Matter of Perspective
Daniel Franzen, Jan Disselhoff, David Hartmann
- ICML 2025 poster
- Readers: Everyone
Nonlinear Advantage: Trained Networks Might Not Be As Complex as You Think
Christian H.X. Ali Mehmeti-Göpel, Jan Disselhoff
- ICML 2023 Poster
- Readers: Everyone
Average Path Length: Sparsification of Nonlinearties Creates Surprisingly Shallow Networks
Christian H. X. Ali Mehmeti-Göpel, Jan Disselhoff
- Published: 01 Jan 2022, Last Modified: 17 May 2023
- CoRR 2022
- Readers: Everyone

Co-Authors