Hyperbolic Manifold Constrained Tabular Neural Network

2026-06-23

AI Summary

Existing tabular prediction models are predominantly optimized in Euclidean space, limiting their capacity to capture complex latent structural relationships within the data. This work proposes HTNN, which introduces hyperbolic geometry into supervised learning for tabular data for the first time. By leveraging hyperbolic manifold constraints and a heterogeneous feature representation layer, HTNN enables geometry-aware training to effectively shape latent representations while supporting transformation into Euclidean space for efficient, geometry-free inference. Extensive evaluations on the TALENT benchmark demonstrate that the proposed method ranks first across 200 classification datasets and third across 100 regression datasets, yielding substantial improvements in predictive performance for tabular data.

Abstract

Tabular prediction is central to a wide range of real-world applications. Tabular data typically contain heterogeneous features as well as rich and complex relational information that can imply a latent structural manifold. Hyperbolic geometry can help capture complex structural relations in data. However, most existing tabular prediction models are constructed and optimized in Euclidean space. How to incorporate hyperbolic geometry into supervised tabular learning remains underexplored. We propose \textbf{HTNN}, a supervised hyperbolic manifold constrained tabular neural network for tabular prediction. HTNN consists of a hyperbolic feature-value representation layer for heterogeneous categorical and numerical features, followed by a conventional MLP predictor. HTNN employs a \emph{geometry-aware training} and \emph{geometry-free inference} optimization framework. The \emph{geometry-aware training} allows hyperbolic geometry to shape the latent representation learning of heterogeneous feature values. After training, the latent hyperbolic representations can be converted into ordinary Euclidean space for efficient \emph{geometry-free inference}. We conducted extensive experiments on the TALENT benchmark. HTNN ranks first among 36 methods on 200 classification datasets and third among 34 methods on 100 regression datasets. Experimental results show that the proposed hyperbolic manifold constrained tabular neural network is effective.

Problem

• Tabular prediction
• Hyperbolic geometry
• Heterogeneous features
• Latent structural manifold
• Supervised learning

Innovation

• Hyperbolic Geometry
• Tabular Neural Network
• Geometry-aware Training
• Heterogeneous Features
• Manifold Constraint