Tianyu Sun
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Tianyu Sun has a total of 31 co-authors. Here are the top co-authors by number of shared publications:
- Rui Huang
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 - Chuansheng Li
National Institute of Metrology, Beijing, China
1 shared publication - Feipeng Lin
National Institute of Metrology, Beijing, China
1 shared publication - Guorui Zhou
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.
- 21 July 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 71 (00189456) , 1-9
Measurement Uncertainty in Voltage Coefficients for Compressed-Gas Capacitor Based on a Simplified Tilting Method
High-voltage compressed-gas capacitor is a type of precision equipment commonly used in high-voltage laboratories. The voltage dependence of capacitance, in other words, the voltage coefficient (VC) of capacitance, is an important factor to ensure the measurement accuracy of capacitors. The simplified tilting method (STM) derived from PTB, which provides a solution for measuring the VC of the high-voltage capacitor with compressed-gas coaxial cylindrical electrodes structure. However, it is not an easy work for most laboratories. In order to evaluate the VC based on the STM, a special experimental platform was developed. The high-voltage capacitor installed on the platform can be tilted and rotated at high voltage and can be remote-controlled. Based on the developed platform, the VC of a 350-kV high-voltage capacitor was measured. In order to verify, it was remeasured by the direct measurement method with a dummy high-voltage capacitor, and before that, the VC of the dummy capacitor was calibrated with a near zero-VC capacitor. The results show a good agreement, and the consistency of the measurement results is better than 1.2 $\times \,\,10^{-6}$ between the two methods. Furthermore, the evaluation of measurement uncertainty in the VC determination of capacitor based on STM is introduced in detail. The evaluation results show that the expanded uncertainty is significantly better than 2.0 $\times \,\,10^{-6}$ , $k =2$ .
- 21 July 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 71 (00189456) , 1-9
Measurement Uncertainty in Voltage Coefficients for Compressed-Gas Capacitor Based on a Simplified Tilting Method
High-voltage compressed-gas capacitor is a type of precision equipment commonly used in high-voltage laboratories. The voltage dependence of capacitance, in other words, the voltage coefficient (VC) of capacitance, is an important factor to ensure the measurement accuracy of capacitors. The simplified tilting method (STM) derived from PTB, which provides a solution for measuring the VC of the high-voltage capacitor with compressed-gas coaxial cylindrical electrodes structure. However, it is not an easy work for most laboratories. In order to evaluate the VC based on the STM, a special experimental platform was developed. The high-voltage capacitor installed on the platform can be tilted and rotated at high voltage and can be remote-controlled. Based on the developed platform, the VC of a 350-kV high-voltage capacitor was measured. In order to verify, it was remeasured by the direct measurement method with a dummy high-voltage capacitor, and before that, the VC of the dummy capacitor was calibrated with a near zero-VC capacitor. The results show a good agreement, and the consistency of the measurement results is better than 1.2 $\times \,\,10^{-6}$ between the two methods. Furthermore, the evaluation of measurement uncertainty in the VC determination of capacitor based on STM is introduced in detail. The evaluation results show that the expanded uncertainty is significantly better than 2.0 $\times \,\,10^{-6}$ , $k =2$ .