7 papers · 1 filter
Personalized Task Dependency Graphs for Mitigating Signal Erosion in Multi-Task Recommendation
Fuyuan Liu, Tiandeng Wu, Yaqun Fang +8
Optimizing multiple conversion objectives is a core challenge in industrial recommendation, often limited by signal erosion in rigid architectures. Existing Multi-Task Learning (MT…
Adaptive Item-based Collaborative Structures via Noise Rescheduling in Diffusion for Generative Recommendation
Jiaqi Wang, Tianying Liu, Heng Chang +3
Discrete Diffusion Models (DDMs) have recently been introduced to recommendation systems, modeling user history as a token generation process via iterative denoising. However, whil…
SIDInspector: A Mapping-First Diagnostic Resource for Semantic-ID Tokenizers
Jiandong Ding, Heng Chang, Huijie Qin +1
Semantic-ID (SID) tokenizers are increasingly reused as standalone artifacts in generative recommendation: an exported item-to-code mapping becomes the address space that a later s…
Beyond the Flat Sequence: Hierarchical and Preference-Aware Generative Recommendations
Zerui Chen, Heng Chang, Tianying Liu +5
Generative Recommenders (GRs), exemplified by the Hierarchical Sequential Transduction Unit (HSTU), have emerged as a powerful paradigm for modeling long user interaction sequences…
FuXi-Linear: Unleashing the Power of Linear Attention in Long-term Time-aware Sequential Recommendation
Yufei Ye, Wei Guo, Hao Wang +7
Modern recommendation systems primarily rely on attention mechanisms with quadratic complexity, which limits their ability to handle long user sequences and slows down inference. W…
FAIR: Focused Attention Is All You Need for Generative Recommendation
Longtao Xiao, Haolin Zhang, Guohao Cai +6
Recently, transformer-based generative recommendation has garnered significant attention for user behavior modeling. However, it often requires discretizing items into multi-code r…
Act-With-Think: Chunk Auto-Regressive Modeling for Generative Recommendation
Yifan Wang, Weinan Gan, Longtao Xiao +7
Generative recommendation (GR) typically encodes behavioral or semantic aspects of item information into discrete tokens, leveraging the standard autoregressive (AR) generation par…