Institution profile

Munich Data Science Institute

Academic institutioneurope · de
Official website
Research library6linked papers
Opportunities0open roles
Selected work

Representative Papers

Physics-Aligned Electronic Ground-State Learning Improves Generalization

Oct 07, 2026

This study addresses the limited out-of-distribution generalization of machine learning interatomic potentials and their inherent trade-off between computational cost and accuracy by proposing a physics-aligned electronic ground-state descriptor model. Methodologically, Kohn-Sham density functional theory constraints are embedded into the learning architecture, and three techniques—ON-Loss, GROOT, and ROCKET—are introduced to eliminate unphysical degrees of freedom and enable label-free self-consistent fine-tuning. Experimental results demonstrate that energy and force prediction errors are reduced by over 95%, achieving a mean absolute error of 0.07 mHa on the QMugs dataset, with reaction chemistry errors falling below chemical accuracy thresholds. These improvements significantly enhance the cross-scale generalization performance of the model.

0 citationsRead paper

Sparse cubical complexes for efficient topology-preservation in image data

Sep 29, 2026

This study addresses the prohibitive computational overhead of persistent homology (PH) in image segmentation, which hinders its scalability to large-scale 3D data. To overcome this limitation, we introduce sparse cubical complexes into PH computation for the first time and propose a topology-aware deep learning framework based on sparse cubical filtration. By eliminating redundant information, this approach accelerates topological feature extraction and optimization. The proposed method achieves an effective balance between computational efficiency and topological precision, yielding a hundredfold speedup and an 80% improvement in topological accuracy. Furthermore, it enables end-to-end training on large-scale 3D data, establishing a new paradigm for the scalable application of topological methods in 3D vision.

0 citationsRead paper
Recent publications

Latest Papers

Physics-Aligned Electronic Ground-State Learning Improves Generalization

Oct 07, 2026

This study addresses the limited out-of-distribution generalization of machine learning interatomic potentials and their inherent trade-off between computational cost and accuracy by proposing a physics-aligned electronic ground-state descriptor model. Methodologically, Kohn-Sham density functional theory constraints are embedded into the learning architecture, and three techniques—ON-Loss, GROOT, and ROCKET—are introduced to eliminate unphysical degrees of freedom and enable label-free self-consistent fine-tuning. Experimental results demonstrate that energy and force prediction errors are reduced by over 95%, achieving a mean absolute error of 0.07 mHa on the QMugs dataset, with reaction chemistry errors falling below chemical accuracy thresholds. These improvements significantly enhance the cross-scale generalization performance of the model.

0 citationsRead paper

Sparse cubical complexes for efficient topology-preservation in image data

Sep 29, 2026

This study addresses the prohibitive computational overhead of persistent homology (PH) in image segmentation, which hinders its scalability to large-scale 3D data. To overcome this limitation, we introduce sparse cubical complexes into PH computation for the first time and propose a topology-aware deep learning framework based on sparse cubical filtration. By eliminating redundant information, this approach accelerates topological feature extraction and optimization. The proposed method achieves an effective balance between computational efficiency and topological precision, yielding a hundredfold speedup and an 80% improvement in topological accuracy. Furthermore, it enables end-to-end training on large-scale 3D data, establishing a new paradigm for the scalable application of topological methods in 3D vision.

0 citationsRead paper