Dataset Distillation by Tabular Alignment via Moment Embeddings
Machine Learning and Knowledge Extraction · 2026-08-12 · 8 (8) · 241
DOI: 10.3390/make8080241 ↗At a glance
Journal article in Machine Learning and Knowledge Extraction, published online on 2026-08-12. Written by Eduard Barnoviciu and 1 co-authors. Cited 1 times so far (Crossref); 42 references. The full text is freely available (open access, CC BY 4.0). Funded by: Unitatea Executiva Pentru Finantarea Invatamantului Superior Si A Cercetarii Stiintifice Universitare.
Abstract (original language)
Dataset distillation has achieved strong results in computer vision, but is largely underexplored in the tabular domain. We introduce a tabular dataset distillation method that projects the data through many random embedders (views) to achieve invariance to transformation and to focus on the consistency between real and synthetic sets. The synthetic set is determined through a formulation of distribution matching between the many-view projection of the original and distilled dataset. Building on this approach, our proposal achieves three goals: (1) we formulate the Tabular Alignment via Moment Embeddings (TAME) method and, by extensive empirical evaluation, we show its efficiency; (2) we evaluate TAME on a benchmark of 18 tabular datasets, with strong baselines and evaluation metrics; and (3) we present a structured set of studies analyzing the impact of losses, dataset geometry, embedder architecture, instances per class (IPC) budget and downstream classifiers. We show that the proposed TAME method consistently surpasses baselines on neural classifiers, while remaining competitive with strong coreset baselines on tree-based classifiers (RF, XGBoost). Performance is further increased, especially for tree-based classifiers with a lightweight validation method. Extensive evaluation, including on additional large-scale sets and ablation experiments, allow a better understanding of the method.
Funding
- Unitatea Executiva Pentru Finantarea Invatamantului Superior Si A Cercetarii Stiintifice Universitare
Cite
APA
Barnoviciu, E., & Florea, C. (2026). Dataset Distillation by Tabular Alignment via Moment Embeddings. Machine Learning and Knowledge Extraction, 8(8), 241. https://doi.org/10.3390/make8080241
Vancouver
Barnoviciu E, Florea C. Dataset Distillation by Tabular Alignment via Moment Embeddings. Machine Learning and Knowledge Extraction. 2026;8(8):241. doi:10.3390/make8080241
BibTeX
@article{Barnoviciu2026,
author = {Barnoviciu, Eduard and Florea, Corneliu},
title = {Dataset Distillation by Tabular Alignment via Moment Embeddings},
journal = {Machine Learning and Knowledge Extraction},
year = {2026},
volume = {8},
number = {8},
pages = {241},
doi = {10.3390/make8080241}
}
EUVIMED