Selected work

Representative Papers

MOTIP2: Spatial Priors for End-to-End Multi-Object Tracking

Oct 07, 2026

This study addresses the identity mismatch problem in end-to-end multi-object tracking caused by the absence of spatial priors. Building upon the DEIM detection transformer architecture, this work proposes explicit spatial prior strategies across three levels: data, loss, and representation. Specifically, it introduces Spatial ID Switches, Spatial ID Loss, and Spatial Anchor techniques to optimize trajectory association and attention mechanisms. These components effectively correct long-range identity switch errors without requiring additional annotations while preserving fully end-to-end inference. Experimental results demonstrate that the proposed method achieves state-of-the-art performance on benchmarks such as DanceTrack, attaining a HOTA score of 73.4. Furthermore, a lightweight variant of the model accelerates inference speed by more than threefold, offering an efficient yet highly accurate solution for real-time tracking applications.

0 citationsRead paper

On Pairwise Quantile Regression -- Statistical Guarantees and Applications

Jul 05, 2026

This work addresses the limitation of traditional quantile regression in modeling pairwise similarity responses—such as image similarity scores in face recognition—by introducing, for the first time, a pairwise quantile regression framework. The proposed method predicts conditional quantiles of similarity scores for input pairs by minimizing a pairwise pinball loss. On the theoretical front, sharp concentration inequalities based on U-processes are employed to establish generalization error bounds, yielding fast learning rates under mild assumptions. Empirically, the approach is validated through both simulated data and real-world face recognition similarity scoring tasks, demonstrating its effectiveness and practical utility.

0 citationsRead paper

Sharper Perturbed-Kullback-Leibler Exponential Tail Bounds for Beta and Dirichlet Distributions

Aug 11, 2025

Existing exponential tail bounds for Beta and Dirichlet distributions—particularly KL-divergence-based bounds—are loose, especially in high-dimensional or nonparametric Bayesian settings. Method: We propose a novel KL-divergence perturbation framework that introduces a mean-zero shift to tighten these bounds systematically. This perturbation is optimized to enhance bound sharpness while preserving statistical validity. Contribution/Results: Our approach is the first to extend perturbation-based optimization from univariate Beta to multivariate Dirichlet distributions and, further, to the nonparametric Dirichlet process. Theoretically, increasing the perturbation parameter strictly improves bound tightness; the resulting bounds dominate classical Chernoff-type and standard KL-type bounds across diverse parameter regimes. Empirical evaluation demonstrates superior performance in constructing confidence intervals and analyzing posterior contraction rates in high-dimensional probabilistic inference. By bridging exponential inequalities with Bayesian nonparametrics, our work expands the theoretical applicability of concentration bounds in modern Bayesian statistics.

0 citationsRead paper

Recent publications

Latest Papers

MOTIP2: Spatial Priors for End-to-End Multi-Object Tracking

Oct 07, 2026

This study addresses the identity mismatch problem in end-to-end multi-object tracking caused by the absence of spatial priors. Building upon the DEIM detection transformer architecture, this work proposes explicit spatial prior strategies across three levels: data, loss, and representation. Specifically, it introduces Spatial ID Switches, Spatial ID Loss, and Spatial Anchor techniques to optimize trajectory association and attention mechanisms. These components effectively correct long-range identity switch errors without requiring additional annotations while preserving fully end-to-end inference. Experimental results demonstrate that the proposed method achieves state-of-the-art performance on benchmarks such as DanceTrack, attaining a HOTA score of 73.4. Furthermore, a lightweight variant of the model accelerates inference speed by more than threefold, offering an efficient yet highly accurate solution for real-time tracking applications.

0 citationsRead paper

On Pairwise Quantile Regression -- Statistical Guarantees and Applications

Jul 05, 2026

This work addresses the limitation of traditional quantile regression in modeling pairwise similarity responses—such as image similarity scores in face recognition—by introducing, for the first time, a pairwise quantile regression framework. The proposed method predicts conditional quantiles of similarity scores for input pairs by minimizing a pairwise pinball loss. On the theoretical front, sharp concentration inequalities based on U-processes are employed to establish generalization error bounds, yielding fast learning rates under mild assumptions. Empirically, the approach is validated through both simulated data and real-world face recognition similarity scoring tasks, demonstrating its effectiveness and practical utility.

0 citationsRead paper

Sharper Perturbed-Kullback-Leibler Exponential Tail Bounds for Beta and Dirichlet Distributions

Aug 11, 2025

Existing exponential tail bounds for Beta and Dirichlet distributions—particularly KL-divergence-based bounds—are loose, especially in high-dimensional or nonparametric Bayesian settings. Method: We propose a novel KL-divergence perturbation framework that introduces a mean-zero shift to tighten these bounds systematically. This perturbation is optimized to enhance bound sharpness while preserving statistical validity. Contribution/Results: Our approach is the first to extend perturbation-based optimization from univariate Beta to multivariate Dirichlet distributions and, further, to the nonparametric Dirichlet process. Theoretically, increasing the perturbation parameter strictly improves bound tightness; the resulting bounds dominate classical Chernoff-type and standard KL-type bounds across diverse parameter regimes. Empirical evaluation demonstrates superior performance in constructing confidence intervals and analyzing posterior contraction rates in high-dimensional probabilistic inference. By bridging exponential inequalities with Bayesian nonparametrics, our work expands the theoretical applicability of concentration bounds in modern Bayesian statistics.

0 citationsRead paper