HD-Rec outperforms existing models by up to 15% on cross-domain benchmarks like Amazon and MovieLens datasets?

HD-Rec outperforms existing models by up to 15% on cross-domain benchmarks like Amazon and MovieLens datasets.

Uses hierarchical quantization with domain-adaptive sparse routing to handle heterogeneous item semantics?

Uses hierarchical quantization with domain-adaptive sparse routing to handle heterogeneous item semantics.

Incorporates a mixture-of-experts module with shared and specialized experts for dynamic adaptation.

Research & Papers

arXiv cs.IR August 10, 2026

⚡HD-Rec outperforms existing models by 15% on cross-domain benchmarks...

Deep Dive

Key Points

HD-Rec outperforms existing models by up to 15% on cross-domain benchmarks like Amazon and MovieLens datasets.
Uses hierarchical quantization with domain-adaptive sparse routing to handle heterogeneous item semantics.
Incorporates a mixture-of-experts module with shared and specialized experts for dynamic adaptation.

HD-Rec could revolutionize recommendation systems by enabling seamless cross-domain personalization at scale.