TN-SHAP-G: Graph-Structured Tensor Network Surrogates for Shapley Values and Interactions

2026-05-31

AI Summary

This work addresses the computational bottleneck in evaluating Shapley values and their higher-order interactions for graph-structured inputs, where the exponential size of the subset space renders exact computation infeasible. To overcome this challenge, the authors propose a graph-aligned multilinear tensor network surrogate model that incorporates graph structural priors to accurately approximate the behavior of black-box models under masked inputs. By integrating graph priors directly into the tensor network architecture, the method enables efficient and deterministic computation of both Shapley values and higher-order interaction indices without requiring additional model queries or Monte Carlo sampling. Experimental results on molecular benchmarks demonstrate that the approach closely approximates exact solutions on small graphs and significantly outperforms existing sampling-based methods on large graphs, achieving a favorable balance between accuracy and scalability.

Abstract

Shapley values are a widely used tool for attributing importance and interactions among input variables in black-box models, but their computation involves a function defined over an exponentially large space of subsets. We propose TN-SHAP-G, a framework that exploits structure in graph-structured inputs to compute Shapley values and higher-order interaction indices efficiently. Given a predictor and a fixed masking scheme, TN-SHAP-G learns a compact, graph-aligned multilinear surrogate that approximates the masked-input behavior, represented as a tensor network whose topology mirrors the input graph. Once trained from a small number of oracle queries, the surrogate enables deterministic recovery of first- and higher-order Shapley indices via the multilinear extension, without additional model queries or Monte Carlo variance. Experiments on molecular benchmarks show that the learned factorization closely matches exact Shapley values on small graphs and scales efficiently to larger graphs where sampling-based methods become infeasible.

Problem

• Shapley values
• graph-structured inputs
• computational complexity
• higher-order interactions
• black-box models

Innovation

• Tensor Network
• Shapley Values
• Graph-Structured Data
• Surrogate Model
• Multilinear Extension