Kai Sun

Kai Sun

Auto-generated profile, based on publicly available information

Kai Sun

Tsinghua University, Beijing, China

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2020 in Neurocomputing
Verified
Journal Article
Closed access
Cai MengCai Meng
Cai Meng
KSKai Sun
Kai Sun
SGShaoya Guan
Shaoya Guan
QWQi Wang
Qi Wang
RZRui Zong
Rui Zong
LLLei Liu
Lei Liu

21 January 2026 in Applied Soft Computing
Verified
Journal Article
Closed access
XWXiaofeng Wen
Xiaofeng Wen
YFYaoyao Fan
Yaoyao Fan
FSFuchun Sun
Fuchun Sun
XZXiaohong Zhang
Xiaohong Zhang
KSKai Sun
Kai Sun
PLPengkun Liu
Pengkun Liu

17 March 2025 in IEEE Transactions on Image Processing
Verified
Journal Article
Closed access
LWLing Wang
Ling Wang
RCRunfa Chen
Runfa Chen
FSFuchun Sun
Fuchun Sun
XWXinzhou Wang
Xinzhou Wang
KSKai Sun
Kai Sun
YWYikai Wang
Yikai Wang
Unsupervised non-rigid point cloud shape correspondence underpins a multitude of 3D vision tasks, yet itself is non-trivial given the exponential complexity stemming from inter-point degree-of-freedom, i.e., pose transformations. Based on the assumption of local rigidity, one solution for reducing complexity is to decompose the overall shape into independent local regions using Local Reference Frames (LRFs) that are equivariant to SE(3) transformations. However, the focus solely on local structure neglects global geometric contexts, resulting in less distinctive LRFs that lack crucial semantic information necessary for effective matching. Furthermore, such complexity introduces out-of-distribution geometric contexts during inference, thus complicating generalization. To this end, we introduce 1) EquiShape, a novel structure tailored to learn pair-wise LRFs with global structural cues for both spatial and semantic consistency, and 2) LRF-Refine, an optimization strategy generally applicable to LRF-based methods, aimed at addressing the generalization challenges. Specifically, for EquiShape, we employ cross-talk within separate equivariant graph neural networks (Cross-GVP) to build long-range dependencies to compensate for the lack of semantic information in local structure modeling, deducing pair-wise independent SE(3)-equivariant LRF vectors for each point. For LRF-Refine, the optimization adjusts LRFs within specific contexts and knowledge, enhancing the geometric and semantic generalizability of point features. Our overall framework surpasses the state-of-the-art methods by a large margin on three benchmarks. Codes are available at https://github.com/2019EPWL/EquiShape.

27 January 2025 in IEEE Sensors Journal
Verified
Journal Article
Closed access
XWXinzhou Wang
Xinzhou Wang
FSFuchun Sun
Fuchun Sun
LWLing Wang
Ling Wang
KSKai Sun
Kai Sun
XZXudong Zhang
Xudong Zhang
HZHuaidong Zhou
Huaidong Zhou
6D pose estimation from a monocular RGB sensor is essential for robotic assembly. While deep learning approaches leverage priors from synthetic and labeled real-world data to estimate 6D poses, their generalizability is often constrained by the limited scale and realism of training datasets. Moreover, scale ambiguity is an inherent issue when image datasets are captured without calibration or depth sensing. To address these limitations, we propose Phys-Field, a physics-aware neural surface framework for joint 6D pose and scale estimation in robotic assembly without requiring 3D models or depth sensing. We incorporate an isotropic scale factor into the 6D pose estimation process, formulating the task as an iterative optimization problem within the Sim(3) group. The framework integrates multiple differentiable neural surfaces to estimate corresponding Sim(3) transformations by minimizing photometric differences between rendered and observed images through rendering inversion. To achieve conflict-free compositional rendering and efficient collision detection, we introduce deep convex decomposition. Concurrently, Phys-Field leverages dynamic simulations with physics parameters derived from large language models to evaluate the plausibility of the estimated Sim(3) transformations, effectively resolving scale ambiguity and accelerating convergence. We validate our framework’s effectiveness in a real-world assembly task, with experimental results showing notable improvements in reconstruction fidelity, Sim(3) transformations estimation accuracy, and assembly success rate.

12 February 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
HSHaoran Sun
Haoran Sun
WLWenwen Liu
Wenwen Liu
VCVictor Yu Cui
Victor Yu Cui
KSKai Sun
Kai Sun
XZXiaohui Zhang
Xiaohui Zhang
WLWannian Liang
Wannian Liang
Background: Population ageing in China creates urgent need for tools that anticipate functional decline before costly disability events.Methods: Using nationally representative CHARLS waves (2011–2018), we trained nine supervised algorithms to predict next-wave Activities of Daily Living (ADL) impairment from demographic, chronic-disease, psychosocial, and behavioral features. Class imbalance was handled with class weights; Hyperparameters were tuned on a validation set using grid search, and final model performance, including discrimination (ROC–AUC, PR–AUC) and threshold metrics (accuracy, precision, recall, F1, specificity), was evaluated on a held-out test set. We repeated the workflow in four age strata (<60, 60–69, 70–79, ≥80 years) and used SHAP to explain global and case-level contributions.Findings: In the full sample, tree-based learners offered modest gains over logistic regression; CatBoost delivered the most stable cross-stratum AUC (≈0·70–0·72 for <70; ≈0·65–0·68 for 70–79; ≈0·58–0·61 for ≥80) with favorable precision–recall trade-offs. SHAP consistently identified age and multimorbidity—especially arthritis/rheumatism and hypertension—as principal risk-increasing drivers, whereas higher income, education, and life satisfaction were protective.Interpretation An explainable, tabular-data model can be embedded in primary care and hospital discharge workflows to flag high-risk older adults and to target long-term care insurance resources toward conditions most strongly associated with subsequent ADL impairment.

11 June 2026 in Intelligent Medicine
Verified
Journal Article
Open access
KSKai Sun
Kai Sun
TLTianyu Liu
Tianyu Liu
XWXinzhou Wang
Xinzhou Wang
LWLing Wang
Ling Wang
FSFuchun Sun
Fuchun Sun
JDJiahong Dong
Jiahong Dong