Rui Huang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Rui Huang has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
- Tianyu Sun
Kuaishou Technology, Beijing, China
2 shared publications - Zhaojie Liu
Kuaishou Technology, Beijing, China
2 shared publications - Changqing Qiu
Kuaishou Technology, Beijing, China
1 shared publication - Guorui Zhou
Kuaishou Technology, Beijing, China
1 shared publication - Guorui Zhou
Kuaishou Inc., Beijing, China
1 shared publication - Hezheng Lin
Kuaishou Technology, Beijing, China
1 shared publication
Latest Publications
- Top Cited PapersOpen Access
- 10 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 5915-5922
QARM: Quantitative Alignment Multi-Modal Recommendation at Kuaishou
- 10 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 5915-5922
Top Cited PapersOpen AccessQARM: Quantitative Alignment Multi-Modal Recommendation at Kuaishou
- Open Access
- 13 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 694-697
De-Anchor: Mitigating Attention Polarization for Lifelong User Behavior Modeling in Click-Through Rate Prediction
User lifelong behavior sequences are essential for click-through rate (CTR) prediction tasks in industrial recommender systems. Attention-based module, especially multi-head target attention (MHTA), has been proven to be effective in aggregating behavior features given a certain target item. However, we found a common phenomenon that attention weights in MHTA tend to over-concentrate on merely a small subset of a user's historical behaviors, producing a sparse one-hot distributed attention weights in training gradually, which we callAttention Polarization (AP). These polarized weights on certain behaviors (which we call "\textitattention anchor ") could make the model fail to capture a user's diversified interests, and harm the learning of behavior embedding as the gradients on these features are nearly zero. We introduce two indicators:anchor rate andattention entropy to measure the magnitude of AP, and proposeDe-Anchor , a novel method to alleviate it, which can serve as a stand-alone and parameter-efficient plug-in to existing CTR backbones. De-Anchor contains two modules:Anchor-aware gradient dropout (AGD) forcing the model to capture diversified interest information from behavior sequences by discarding gradients of non-behavior features, andTarget-aware attention anchor (TAA) providing a pseudo behavior to offload excessive weights of MHTA. Extensive offline experiments and industrial online A/B tests demonstrate the efficacy of our method.
- 13 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 694-697
Open AccessDe-Anchor: Mitigating Attention Polarization for Lifelong User Behavior Modeling in Click-Through Rate Prediction
User lifelong behavior sequences are essential for click-through rate (CTR) prediction tasks in industrial recommender systems. Attention-based module, especially multi-head target attention (MHTA), has been proven to be effective in aggregating behavior features given a certain target item. However, we found a common phenomenon that attention weights in MHTA tend to over-concentrate on merely a small subset of a user's historical behaviors, producing a sparse one-hot distributed attention weights in training gradually, which we callAttention Polarization (AP). These polarized weights on certain behaviors (which we call "\textitattention anchor ") could make the model fail to capture a user's diversified interests, and harm the learning of behavior embedding as the gradients on these features are nearly zero. We introduce two indicators:anchor rate andattention entropy to measure the magnitude of AP, and proposeDe-Anchor , a novel method to alleviate it, which can serve as a stand-alone and parameter-efficient plug-in to existing CTR backbones. De-Anchor contains two modules:Anchor-aware gradient dropout (AGD) forcing the model to capture diversified interest information from behavior sequences by discarding gradients of non-behavior features, andTarget-aware attention anchor (TAA) providing a pseudo behavior to offload excessive weights of MHTA. Extensive offline experiments and industrial online A/B tests demonstrate the efficacy of our method.