TabSHAP

2026-04-22

AI Summary

This work addresses the lack of faithful local interpretability and inadequate characterization of predictive uncertainty in large language models for tabular classification tasks. The authors propose a model-agnostic interpretability framework that, for the first time, targets the holistic shift in the output class distribution—rather than merely prediction flips—as the attribution objective. By applying masking at the level of serialized key-value fields and combining Shapley value estimation with Jensen–Shannon divergence, the method quantifies each feature’s contribution to changes in the predicted probability distribution. Experimental results on the Adult Income and Heart Disease benchmarks demonstrate that the proposed approach significantly outperforms random baselines and XGBoost surrogates, achieving state-of-the-art performance in deletion-based faithfulness metrics.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Large Language Models (LLMs) fine-tuned on serialized tabular data are emerging as powerful alternatives to traditional tree-based models, particularly for heterogeneous or context-rich datasets. However, their deployment in high-stakes domains is hindered by a lack of faithful interpretability; existing methods often rely on global linear proxies or scalar probability shifts that fail to capture the model's full probabilistic uncertainty. In this work, we introduce TabSHAP, a model-agnostic interpretability framework designed to directly attribute local query decision logic in LLM-based tabular classifiers. By adapting a Shapley-style sampled-coalition estimator with Jensen-Shannon divergence between full-input and masked-input class distributions, TabSHAP quantifies the distributional impact of each feature rather than simple prediction flips. To align with tabular semantics, we mask at the level of serialized key:value fields (atomic in the prompt string), not individual subword tokens. Experimental validation on the Adult Income and Heart Disease benchmarks demonstrates that TabSHAP isolates critical diagnostic features, achieving significantly higher faithfulness than random baselines and XGBoost proxies. We further run a distance-metric ablation on the same test instances and TabSHAP settings: attributions are recomputed with KL or L1 replacing JSD in the similarity step (results cached per metric), and we compare deletion faithfulness across all three.

Problem

• interpretability
• large language models
• tabular data
• probabilistic uncertainty
• faithfulness

Innovation

• TabSHAP
• Shapley values
• Jensen-Shannon divergence
• tabular interpretability
• LLM-based classification