Cross-Treatment Effect Estimation for Multi-Category, Multi-Valued Causal Inference via Dynamic Neural Masking

2025-11-03

AI Summary

Addressing the challenges of modeling heterogeneous cross-effects under multi-class, multi-value interventions—where existing methods are constrained by binary or single-treatment assumptions, suffer from poor scalability, and lack appropriate evaluation frameworks—this paper proposes XTNet, an end-to-end causal inference model based on dynamic neural masking and effect decomposition. Its core innovations include: (i) a learnable masking mechanism that adaptively captures combinatorial intervention effects, and (ii) explicit disentanglement of baseline and interaction effects. Additionally, we introduce MCMV-AUCC, a novel evaluation metric that jointly accounts for treatment cost and interaction impact. Extensive experiments on synthetic data and multiple real-world datasets—including industrial-scale A/B tests—demonstrate that XTNet significantly outperforms state-of-the-art methods in both counterfactual prediction accuracy and ranking quality.

Technology Category

• Reasoning under Uncertainty: Causality
• Machine Learning: Causal Learning
• Knowledge Representation and Reasoning: Action, Change, and Causality

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Counterfactual causal inference faces significant challenges when extended to multi-category, multi-valued treatments, where complex cross-effects between heterogeneous interventions are difficult to model. Existing methodologies remain constrained to binary or single-type treatments and suffer from restrictive assumptions, limited scalability, and inadequate evaluation frameworks for complex intervention scenarios. We present XTNet, a novel network architecture for multi-category, multi-valued treatment effect estimation. Our approach introduces a cross-effect estimation module with dynamic masking mechanisms to capture treatment interactions without restrictive structural assumptions. The architecture employs a decomposition strategy separating basic effects from cross-treatment interactions, enabling efficient modeling of combinatorial treatment spaces. We also propose MCMV-AUCC, a suitable evaluation metric that accounts for treatment costs and interaction effects. Extensive experiments on synthetic and real-world datasets demonstrate that XTNet consistently outperforms state-of-the-art baselines in both ranking accuracy and effect estimation quality. The results of the real-world A/B test further confirm its effectiveness.

Problem

• Estimating multi-category multi-valued treatment causal effects
• Modeling complex cross-effects between heterogeneous interventions
• Addressing limitations of binary treatment methodologies

Innovation

• Dynamic masking mechanisms capture treatment interactions
• Decomposition separates basic effects from cross-interactions
• Proposes MCMV-AUCC metric for cost-aware evaluation