arXiv cs.CL· Eilam Shapira, Alan Arazi, Shoham Grunblat, Roi Reichart·· 2 天前AI 评分31
LightTAR:面向多模态表格学习的轻量快速目标感知表示方法
LightTAR: Light and Fast Target-Aware Representations for Multimodal Tabular Learning
AI 导读
LightTAR 是一种轻量低成本方法,用正则化偏最小二乘(PLS)从编码器最终层和中间层联合提取少量目标感知特征,增强标准 PCA 表示。在 21 个表格模型(含树集成、神经网络和表格基础模型)上评估,LightTAR 提升了所有模型的平均预测性能,在 MulTaBench 几乎每个数据集上均有改善,总运行时间仅小幅增加。该方法可直接集成到现有编码流程,无需修改编码器或学习器。
正文 · 原文
Abstract:Multimodal tabular learning involves classification and regression on tables combining structured numerical and categorical features with unstructured text and images. These unstructured fields are commonly represented by pretrained encoder embeddings reduced with principal component analysis (PCA). This reduction prioritizes input variance but may overlook information relevant to predicting the target. We introduce LightTAR, a simple, inexpensive method that adapts dimensionality reduction to the prediction task. LightTAR augments the standard PCA representation with a few target-aware features extracted jointly from the final and an intermediate encoder layer using regularized partial least squares (PLS). We evaluate LightTAR with 21 tabular models, including tree ensembles, neural networks, and tabular foundation models. LightTAR improves average predictive performance for all models and, averaged across models, improves prediction on almost every MulTaBench dataset. These gains come with a small increase in total runtime relative to PCA representations. LightTAR integrates directly into existing encoding pipelines, improving tabular learners' predictive performance without modifying the encoder or the learner.
| Comments: | 21 pages, 3 figures. Code: this https URL |
| Subjects: | Machine Learning (cs.LG); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2611.17600 [cs.LG] |
| (or arXiv:2611.17600v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2611.17600 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Eilam Shapira [view email]
[v1]
Thu, 12 Nov 2026 23:45:58 UTC (161 KB)
来源:arXiv cs.CL · arxiv.org