Dynamic Spectral Denoising with Global-Context Attention for Multi-Behavior Recommendation

Multi-behavior recommendation improves target-behavior prediction by exploiting heterogeneous auxiliary feedback (e.g., view, collect, and cart), yet its robustness is often undermined by behavior-dependent noise and inconsistency. We argue that the key bottleneck is not merely noisy behaviors, but a representation-level failure caused by two coupled heterogeneities. First, intra-behavior representation entanglement arises when multi-hop propagation blends incidental signals with true preferences in the embedding space. This entanglement renders coarse spatial denoising ineffective, since it cannot suppress noise without sacrificing weak-but-informative niche signals. Second, inter-behavior reliability heterogeneity complicates cross-behavior fusion, as the predictive value of auxiliary behaviors varies substantially across users and contexts. Without reliability calibration, aggregation can be dominated by frequent yet untrustworthy signals, leading to target-intent drift. Existing methods typically address these issues in isolation and often fail when entanglement makes reliability estimation itself unstable. To resolve this robustness bottleneck, we propose Dynamic Spectral Denoising with Global-Context Attention for Multi-Behavior Recommendation (SpectraMB), a target-oriented model that performs representation purification before reliability-aware fusion. To mitigate intra-behavior entanglement, SpectraMB introduces Dynamic Feature-Level Spectral Filtering, which re-parameterizes embeddings along the feature dimension into a feature-frequency space and learns view-adaptive spectral modulation end-to-end under target supervision, enabling component-wise purification without hand-crafted frequency assumptions. Built on purified representations, SpectraMB further proposes Global-Context Attention Fusion, which uses the purified global representation as a stable context anchor to assess view compatibility and perform reliability-aware aggregation, while a residual global backbone preserves stable collaborative structure. Extensive experiments on three real-world datasets show that SpectraMB achieves the best results in most evaluation settings and exhibits improved robustness under noisy interactions. Our implementation is available at https://github.com/miaomiao-cai2/SpectraMB-KDD2026.