arXiv subjects

Anja Bielefeld

Publications and source records attributed to Anja Bielefeld.

3 recordsLinked to original sources

arXiv2026

What DINO saw: ALiBi positional encoding reduces positional bias in Vision Transformers

Vision transformers (ViTs) - especially feature foundation models like DINOv2 - learn rich representations useful for many downstream tasks. However, architectural choices (such as positional encoding) can lead to these models displaying positional biases and artefacts independent of semantic content. This makes zero-shot adaption difficult in fields like material science, where images are often cross-sections of homogeneous microstructure (i.e. having no preferred direction). In this work, we investigate the positional bias in ViTs via linear probing, finding it present across a range of objectives and positional encodings, and subsequently reduce it by finetuning models to use ALiBi relative positional encoding. We demonstrate that these models retain desirable general semantics and their unbiased features can be used successfully in trainable segmentation of complex microscopy images.

cs.CV↗

arXiv2026

Virtual materials testing of ASSB cathodes combining AI-based stochastic 3D modeling and numerical simulations

The performance of all-solid-state battery (ASSB) cathodes strongly depends on their microstructure. Optimizing the cathode morphology can therefore enhance effective macroscopic properties such as ionic and electronic conductivity. The search for optimized microstructures can be facilitated by virtual materials testing: By integrating image analysis and stochastic microstructure modeling to generate a wide range of realistic 3D microstructures and evaluate their effective macroscopic properties by means of numerical simulations, thereby reducing the need for extensive physical experiments. This approach allows for the investigation of structure-property relationships through parametric regression models that incorporate relevant geometrical descriptors of microstructures such as volume fractions, mean geodesic tortuosities, specific surface areas, and constrictivities. By linking these geometrical descriptors to macroscopic properties, virtual materials testing provides quantitative insight into how microstructure influences material performance. In the present paper, this framework is applied for ASSB cathodes. In addition, by systematically varying model parameters, a broad range of 3D microstructures can be generated, which remain close to the original cathode morphology while inducing targeted changes in selected geometrical descriptors. The resulting database enables the calibration of regression models whose predictive performance is assessed by comparing predicted and simulated effective properties such as the ionic and electronic conductivity, thereby quantifying how accurately combinations of geometrical descriptors can explain and predict variations in effective macroscopic properties.

cond-mat.mtrl-sci↗

arXiv2025

Generative adversarial framework to calibrate excursion set models for the 3D morphology of all-solid-state battery cathodes

This paper presents a computational method for generating virtual 3D morphologies of functional materials using low-parametric stochastic geometry models, i.e., digital twins, calibrated with 2D microscopy images. These digital twins allow systematic parameter variations to simulate various morphologies, that can be deployed for virtual materials testing by means of spatially resolved numerical simulations of macroscopic properties. Generative adversarial networks (GANs) have gained popularity for calibrating models to generate realistic 3D morphologies. However, GANs often comprise of numerous uninterpretable parameters make systematic variation of morphologies for virtual materials testing challenging. In contrast, low-parametric stochastic geometry models (e.g., based on Gaussian random fields) enable targeted variation but may struggle to mimic complex morphologies. Combining GANs with advanced stochastic geometry models (e.g., excursion sets of more general random fields) addresses these limitations, allowing model calibration solely from 2D image data. This approach is demonstrated by generating a digital twin of all-solid-state battery (ASSB) cathodes. Since the digital twins are parametric, they support systematic exploration of structural scenarios and their macroscopic properties. The proposed method facilitates simulation studies for optimizing 3D morphologies, benefiting not only ASSB cathodes but also other materials with similar structures.

stat.ML↗