Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

RMBRec: Robust Multi-Behavior Recommendation towards Target Behaviors

Miaomiao Cai, Zhijie Zhang, Junfeng Fang, Zhiyong Cheng, Xiang Wang, Meng Wang

TL;DR

RMBRec addresses robustness gaps in multi-behavior recommendation by treating auxiliary signals as heterogeneous sources of information rather than uniformly reliable inputs. It couples a target-centered Representation Robustness Module (RRM) that aligns auxiliary user representations with the target via a target-anchored contrastive objective, with an Optimization Robustness Module (ORM) that enforces invariant preferences across behavioral environments through a Risk Extrapolation-based regularizer. The approach is grounded in information theory, maximizing predictive information while minimizing its variance across behaviors, and is validated on three real-world datasets where RMBRec achieves superior accuracy and stability under noise. The framework also introduces Behavioral Alignment Ratio (BAR) as a diagnostic metric to quantify cross-behavior alignment and guide robustness design, offering practical insights for deployment in noisy, real-world settings.

Abstract

Multi-behavior recommendation faces a critical challenge in practice: auxiliary behaviors (e.g., clicks, carts) are often noisy, weakly correlated, or semantically misaligned with the target behavior (e.g., purchase), which leads to biased preference learning and suboptimal performance. While existing methods attempt to fuse these heterogeneous signals, they inherently lack a principled mechanism to ensure robustness against such behavioral inconsistency. In this work, we propose Robust Multi-Behavior Recommendation towards Target Behaviors (RMBRec), a robust multi-behavior recommendation framework grounded in an information-theoretic robustness principle. We interpret robustness as a joint process of maximizing predictive information while minimizing its variance across heterogeneous behavioral environments. Under this perspective, the Representation Robustness Module (RRM) enhances local semantic consistency by maximizing the mutual information between users' auxiliary and target representations, whereas the Optimization Robustness Module (ORM) enforces global stability by minimizing the variance of predictive risks across behaviors, which is an efficient approximation to invariant risk minimization. This local-global collaboration bridges representation purification and optimization invariance in a theoretically coherent way. Extensive experiments on three real-world datasets demonstrate that RMBRec not only outperforms state-of-the-art methods in accuracy but also maintains remarkable stability under various noise perturbations. For reproducibility, our code is available at https://github.com/miaomiao-cai2/RMBRec/.

RMBRec: Robust Multi-Behavior Recommendation towards Target Behaviors

TL;DR

RMBRec addresses robustness gaps in multi-behavior recommendation by treating auxiliary signals as heterogeneous sources of information rather than uniformly reliable inputs. It couples a target-centered Representation Robustness Module (RRM) that aligns auxiliary user representations with the target via a target-anchored contrastive objective, with an Optimization Robustness Module (ORM) that enforces invariant preferences across behavioral environments through a Risk Extrapolation-based regularizer. The approach is grounded in information theory, maximizing predictive information while minimizing its variance across behaviors, and is validated on three real-world datasets where RMBRec achieves superior accuracy and stability under noise. The framework also introduces Behavioral Alignment Ratio (BAR) as a diagnostic metric to quantify cross-behavior alignment and guide robustness design, offering practical insights for deployment in noisy, real-world settings.

Abstract

Multi-behavior recommendation faces a critical challenge in practice: auxiliary behaviors (e.g., clicks, carts) are often noisy, weakly correlated, or semantically misaligned with the target behavior (e.g., purchase), which leads to biased preference learning and suboptimal performance. While existing methods attempt to fuse these heterogeneous signals, they inherently lack a principled mechanism to ensure robustness against such behavioral inconsistency. In this work, we propose Robust Multi-Behavior Recommendation towards Target Behaviors (RMBRec), a robust multi-behavior recommendation framework grounded in an information-theoretic robustness principle. We interpret robustness as a joint process of maximizing predictive information while minimizing its variance across heterogeneous behavioral environments. Under this perspective, the Representation Robustness Module (RRM) enhances local semantic consistency by maximizing the mutual information between users' auxiliary and target representations, whereas the Optimization Robustness Module (ORM) enforces global stability by minimizing the variance of predictive risks across behaviors, which is an efficient approximation to invariant risk minimization. This local-global collaboration bridges representation purification and optimization invariance in a theoretically coherent way. Extensive experiments on three real-world datasets demonstrate that RMBRec not only outperforms state-of-the-art methods in accuracy but also maintains remarkable stability under various noise perturbations. For reproducibility, our code is available at https://github.com/miaomiao-cai2/RMBRec/.
Paper Structure (33 sections, 21 equations, 5 figures, 5 tables, 1 algorithm)

This paper contains 33 sections, 21 equations, 5 figures, 5 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Works
  3. Model
  4. Multi-Behavior Encoding
  5. Representation Robustness Module (RRM)
  6. Optimization Robustness Module (ORM)
  7. Behavioral Environments as Optimization Domains
  8. Learning Invariant Preferences for Global Robustness
  9. Target Prediction and Optimization
  10. Information-Theoretic Analysis
  11. Experiments
  12. Experiments Settings
  13. Datasets
  14. Baselines and Implementation Details
  15. Overall Performance
  16. ...and 18 more sections

Figures (5)

  • Figure 1: Illustration of distributional shift and auxiliary noise in multi-behavior recommendation.
  • Figure 2: The overview of RMBRec, which improves robustness via Representation Robustness and Optimization Robustness.
  • Figure 3: Robustness analysis under different perturbation ratios on the Taobao dataset (HR@10).
  • Figure 4: Hyperparameter sensitivity analysis of RMBRec with $\lambda_1$ and $\lambda_2$.
  • Figure 5: Robustness analysis under different perturbation ratios on the Taobao dataset (NDCG@10).