Information Theoretic Counterfactual Learning from Missing-Not-At-Random Feedback

2020/09/06 by Zifeng Wang, Wang, Zifeng, Xi Chen +9 · 1 citation
Computer Science · Decision Sciences · #Advanced Bandit Algorithms Research #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Information Retrieval (cs.IR) #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Methodology (stat.ME) #Recommender Systems and Techniques

paper · pdf · doi:10.48550/arxiv.2009.02623

openalex publication_date 2020/09/06 · openalex created_date 2020/09/11 · openalex updated_date 2026/07/28

Abstract

Counterfactual learning for dealing with missing-not-at-random data (MNAR) is an intriguing topic in the recommendation literature since MNAR data are ubiquitous in modern recommender systems. Missing-at-random (MAR) data, namely randomized controlled trials (RCTs), are usually required by most previous counterfactual learning methods for debiasing learning. However, the execution of RCTs is extraordinarily expensive in practice. To circumvent the use of RCTs, we build an information-theoretic counterfactual variational information bottleneck (CVIB), as an alternative for debiasing learning without RCTs. By separating the task-aware mutual information term in the original information bottleneck Lagrangian into factual and counterfactual parts, we derive a contrastive information loss and an additional output confidence penalty, which facilitates balanced learning between the factual and counterfactual domains. Empirical evaluation on real-world datasets shows that our CVIB significantly enhances both shallow and deep models, which sheds light on counterfactual learning in recommendation that goes beyond RCTs.

Citations

• Adam: A Method for Stochastic Optimization
• The information bottleneck method
• Doubly Robust Policy Evaluation and Learning
• Matrix Factorization Techniques for Recommender Systems
• The central role of the propensity score in observational studies for causal effects
• Opening the Black Box of Deep Neural Networks via Information
• Compressing Neural Networks using the Variational Information Bottleneck
• DeepFM: A Factorization-Machine based Neural Network for CTR Prediction
• InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets
• PyTorch: An Imperative Style, High-Performance Deep Learning Library

Cited by

• Addressing Correlated Latent Exogenous Variables in Debiased Recommender Systems

Related

• CounterCLR: Counterfactual Contrastive Learning with Non-random Missing Data in Recommendation
• Contrastive Counterfactual Learning for Causality-aware Interpretable Recommender Systems
• Cycle-Balanced Representation Learning For Counterfactual Inference
• Adversarial Counterfactual Learning and Evaluation for Recommender System
• Adversary-Free Counterfactual Prediction via Information-Regularized Representations