I am a Ph.D. student with a keen interest in the dynamic and ever-evolving field of Computer Vision. Under the supervision of Prof. Dr. Ing. Margret Keuper at the University of Mannheim in cooperation with Prof. Dr. Janis Keuper at IMLA, Offenburg, I am able to explore this exciting field at the highest level.
My research focuses on signal processing fundamentals, which I apply in modern deep learning approaches to create innovative solutions that push the boundaries of what’s possible. Especially, my current focus lies at the intersection of signal processing fundamentals and convolutional neural networks (CNNs). Through extensive analysis of various components within CNNs, I aim to leverage my signal processing expertise to address current limitations and optimize their performance. Adversarial attacks, in particular, provide a valuable means to assess these vulnerabilities of CNNs and refine their robustness for real-world tasks. As such, I am equally intrigued by adversarial attacks and defences.
News
05/2024 Our paper As large as it gets – Studying Infinitely Large Convolutions via Neural Implicit Frequency Filters got accepted at TMLR with Featured Certification!
05/2024 I am excited to announce that I will be a visiting researcher at Tampere University in the computer vision group of Prof. Dr. Esa Rathu for the next three month!
08/2023 Our paper On the unreasonable vulnerability of transformers for image restoration-and an easy fix got accepted to the ICCV 2023 Workshop on Adversarial Robustness In the Real World!
09/2022 Our paper Robust Models are less Over-Confident got accepted at NeurIPS 2022!
07/2022 Our paper FrequencyLowCut Pooling–Plug & Play against Catastrophic Overfitting got accepted at ECCV 2022!
06/2022 Our paper Aliasing and adversarial robust generalization of CNNs got accepted at ECML 2022!
Publications
As large as it gets – Studying Infinitely Large Convolutions via Neural Implicit Frequency Filters
J. Grabinski, J. Keuper, M. Keuper
TMLR 2024 (Featured Certification)
PDF | Code
Improving Stability during Upsampling–on the Importance of Spatial Context
S. Agnihotri, J. Grabinski, M. Keuper
ECCV 2024
PDF
On the unreasonable vulnerability of transformers for image restoration-and an easy fix
S. Agnihotri, KV. Gandikota, J. Grabinski, P. Chandramouli, M. Keuper
ICCV 2023, Workshop on Adversarial Robustness In the Real World
PDF
Robust Models are less Over-Confident
J. Grabinski, P. Gavrikov, J. Keuper, M. Keuper
NeurIPS 2022
PDF | Code
FrequencyLowCut Pooling–Plug & Play against Catastrophic Overfitting
J. Grabinski, S. Jung, J. Keuper, M. Keuper
ECCV 2022
PDF | Code
Aliasing and adversarial robust generalization of CNNs
J. Grabinski, J. Keuper, M. Keuper
ECML 2022
PDF
Robust Models are less Over-Confident
J. Grabinski, P. Gavrikov, J. Keuper, M. Keuper
ICML 2022, Workshop New Frontiers in Adversarial Machine Learning
PDF
Aliasing coincides with CNNs vulnerability towards adversarial attacks
J. Grabinski, J. Keuper, M. Keuper
AAAI 2022, Workshop on Adversarial Machine Learning and Beyond
PDF