Technology Category
Machine Learning: Causal LearningReasoning under Uncertainty: CausalityComputer Vision: Interpretability, Explainability, and Transparency
A Model of Causal Explanation on Neural Networks for Tabular Data
🤖 AI Summary
This work addresses key interpretability bottlenecks in neural network prediction on tabular data—namely, spurious correlations, unclear causal mechanisms, and difficulty in combinatorial attribution—by proposing Causal-Enhanced Neural Networks (CENNET). CENNET is the first approach to deeply embed Structural Causal Models (SCMs) into the neural inference pipeline, enabling end-to-end joint optimization of prediction accuracy and causal interpretability. It further introduces an information-entropy-based Explanatory Power Index to quantitatively assess attribution quality. Extensive experiments on synthetic and semi-real-world datasets demonstrate that CENNET significantly outperforms state-of-the-art explanation methods in classification tasks: it maintains high predictive accuracy while substantially improving the accuracy, robustness, and transparency of causal attribution. By unifying predictive modeling with causal reasoning, CENNET establishes a novel paradigm for trustworthy AI modeling on tabular data.
Technology Category
Application Category
📝 Abstract
The problem of explaining the results produced by machine learning methods continues to attract attention. Neural network (NN) models, along with gradient boosting machines, are expected to be utilized even in tabular data with high prediction accuracy. This study addresses the related issues of pseudo-correlation, causality, and combinatorial reasons for tabular data in NN predictors. We propose a causal explanation method, CENNET, and a new explanation power index using entropy for the method. CENNET provides causal explanations for predictions by NNs and uses structural causal models (SCMs) effectively combined with the NNs although SCMs are usually not used as predictive models on their own in terms of predictive accuracy. We show that CEN-NET provides such explanations through comparative experiments with existing methods on both synthetic and quasi-real data in classification tasks.
Problem
Research questions and friction points this paper is trying to address.
Addresses pseudo-correlation and causality in neural network predictions
Proposes a causal explanation method for tabular data using NNs
Introduces an entropy-based explanation power index for evaluation
Innovation
Methods, ideas, or system contributions that make the work stand out.
CENNET method for causal explanations in neural networks
Uses structural causal models combined with neural networks
Introduces entropy-based explanation power index
🔎 Similar Papers
No similar papers found.
