Tabular Few-Shot Generalization Across Heterogeneous Feature Spaces
This addresses a gap in few-shot learning for tabular data, which is incremental by building on existing methods like Dataset2Vec to handle heterogeneous features.
The paper tackles the problem of few-shot learning for tabular datasets with heterogeneous feature spaces, proposing FLAT, which learns dataset and column embeddings to enable knowledge transfer, and demonstrates successful generalization across 118 UCI datasets with significant improvements over baselines.
Despite the prevalence of tabular datasets, few-shot learning remains under-explored within this domain. Existing few-shot methods are not directly applicable to tabular datasets due to varying column relationships, meanings, and permutational invariance. To address these challenges, we propose FLAT-a novel approach to tabular few-shot learning, encompassing knowledge sharing between datasets with heterogeneous feature spaces. Utilizing an encoder inspired by Dataset2Vec, FLAT learns low-dimensional embeddings of datasets and their individual columns, which facilitate knowledge transfer and generalization to previously unseen datasets. A decoder network parametrizes the predictive target network, implemented as a Graph Attention Network, to accommodate the heterogeneous nature of tabular datasets. Experiments on a diverse collection of 118 UCI datasets demonstrate FLAT's successful generalization to new tabular datasets and a considerable improvement over the baselines.