Selected work

Representative Papers

Constructing self-referential instances for the clique problem

Jan 27, 2026

This study investigates the intrinsic computational hardness of the clique problem in its critical region. By constructing pairs of graphs at the phase transition point of the Erdős–Rényi random graph model that share identical numbers of vertices, edges, and degree sequences yet exhibit opposite solution statuses—i.e., one contains a k-clique while the other does not—the authors employ degree-preserving symmetric transformations to reveal the indistinguishability of their solution spaces. This work presents the first family of self-referential instances that rigorously establishes the existence of an exact phase transition threshold for the clique problem. Theoretically, it demonstrates that near this threshold, any algorithm must almost surely explore the entire solution space to determine the existence of a solution, thereby explaining the unavoidable necessity of exhaustive search in this regime.

1 citationsRead paper

ViSymRe: Vision-guided Multimodal Symbolic Regression

Dec 15, 2024arXiv.org

Traditional symbolic regression struggles with low efficiency, overfitting, and poor cross-modal convergence when discovering underlying mechanisms from high-dimensional complex data. To address these challenges, we propose the first vision–symbol–numerical tri-modal symbolic regression framework. Our method introduces a novel vision-guided paradigm that leverages image features to enhance equation structural plausibility and interpretability; designs a meta-learning-driven symbolic search strategy enabling transfer of historical experience; and optimizes for structural simplicity and physical plausibility—moving beyond conventional “fit-first” objectives. Key components include a multi-modal encoder, cross-modal attention, and a vision-driven equation structure constraint module. Experiments across multiple benchmarks demonstrate a 12.6% improvement in fitting accuracy, a 37% gain in expression conciseness, a 29.4% increase in structural accuracy, along with strong robustness to noise and superior cross-task generalization capability.

1 citationsRead paper

Hierarchical and Permutation-Invariant Feature Transformation Learning via Policy-Guided Embedding Search

Sep 09, 2026

该研究针对特征转换中的层次关系忽视、顺序敏感性和非凸空间搜索问题，提出了一种结合自注意力机制和策略引导强化学习的框架来优化表格式数据的预测性能。

0 citationsRead paper

Polarity-Asymmetric Structural Calibration for Link Sign Prediction

Sep 05, 2026

Link sign prediction (LSP) aims to infer the positive or negative polarity of unobserved links in signed networks. Signed Graph Neural Networks (SGNNs) usually rely on signed-graph structural priors, including structural balance and homophily-like similarity, to guide message passing and prediction. These priors describe population-level tendencies, not guarantees for individual target edges. Their failures are especially costly under severe sign imbalance, where errors on minority and locally conflicting relations are harder to detect and correct. We propose Polarity-Asymmetric Structural Calibration (PASC), a target-edge structural-prior calibration framework for signed link prediction. PASC constructs a structure-only prior representation, estimates a target-edge structural prior score, and compares this score with a local signed-context cue to derive a conflict residual. The residual calibrates signed attention aggregation, target-edge gated fusion, and regime-adaptive optimization. Experiments on five real-world signed network datasets show that PASC consistently achieves the best Macro-F1 among representative baselines, with competitive AUC, Binary-F1, and Micro-F1. Structural-shift experiments further suggest reduced dependence on dense-neighborhood and local-closure shortcuts. Source code is available at https://github.com/iqqGGGGGGG/PASC-for-LSP.

0 citationsRead paper

Respect Your Zero-Shot Uncertainty: Conservative Calibration for Test-Time-Adapted Vision-Language Models

Aug 06, 2026

Test-time adaptation (TTA) improves the accuracy of vision-language models under distribution shifts but often degrades calibration due to a phenomenon termed “prediction-preserving sharpening,” wherein predictions remain unchanged while confidence becomes excessively high and predictive entropy is overly compressed. This work is the first to establish the link between this phenomenon and calibration error, and introduces Zero-shot Anchor Entropy Calibration (ZAEC)—a label-free, parameter-free post-processing method that leverages zero-shot entropy as a reference for uncertainty and selectively applies temperature scaling to restore predictive entropy without altering class rankings or accuracy. Evaluated across five TTA methods and fifteen datasets on both ViT-B/16 and RN50 architectures, ZAEC consistently achieves substantial reductions in macro-averaged Expected Calibration Error (ECE), yielding state-of-the-art calibration performance.

0 citationsRead paper

Recent publications

Latest Papers

Hierarchical and Permutation-Invariant Feature Transformation Learning via Policy-Guided Embedding Search

Sep 09, 2026

该研究针对特征转换中的层次关系忽视、顺序敏感性和非凸空间搜索问题，提出了一种结合自注意力机制和策略引导强化学习的框架来优化表格式数据的预测性能。

0 citationsRead paper

Polarity-Asymmetric Structural Calibration for Link Sign Prediction

Sep 05, 2026

Link sign prediction (LSP) aims to infer the positive or negative polarity of unobserved links in signed networks. Signed Graph Neural Networks (SGNNs) usually rely on signed-graph structural priors, including structural balance and homophily-like similarity, to guide message passing and prediction. These priors describe population-level tendencies, not guarantees for individual target edges. Their failures are especially costly under severe sign imbalance, where errors on minority and locally conflicting relations are harder to detect and correct. We propose Polarity-Asymmetric Structural Calibration (PASC), a target-edge structural-prior calibration framework for signed link prediction. PASC constructs a structure-only prior representation, estimates a target-edge structural prior score, and compares this score with a local signed-context cue to derive a conflict residual. The residual calibrates signed attention aggregation, target-edge gated fusion, and regime-adaptive optimization. Experiments on five real-world signed network datasets show that PASC consistently achieves the best Macro-F1 among representative baselines, with competitive AUC, Binary-F1, and Micro-F1. Structural-shift experiments further suggest reduced dependence on dense-neighborhood and local-closure shortcuts. Source code is available at https://github.com/iqqGGGGGGG/PASC-for-LSP.

0 citationsRead paper

Respect Your Zero-Shot Uncertainty: Conservative Calibration for Test-Time-Adapted Vision-Language Models

Aug 06, 2026

Test-time adaptation (TTA) improves the accuracy of vision-language models under distribution shifts but often degrades calibration due to a phenomenon termed “prediction-preserving sharpening,” wherein predictions remain unchanged while confidence becomes excessively high and predictive entropy is overly compressed. This work is the first to establish the link between this phenomenon and calibration error, and introduces Zero-shot Anchor Entropy Calibration (ZAEC)—a label-free, parameter-free post-processing method that leverages zero-shot entropy as a reference for uncertainty and selectively applies temperature scaling to restore predictive entropy without altering class rankings or accuracy. Evaluated across five TTA methods and fifteen datasets on both ViT-B/16 and RN50 architectures, ZAEC consistently achieves substantial reductions in macro-averaged Expected Calibration Error (ECE), yielding state-of-the-art calibration performance.

0 citationsRead paper

CosFly-VLA: A Spatially Aware Vision-Language-Action Model for UAV Tracking

Jul 16, 2026

This work addresses the challenge of target loss in urban drone tracking caused by occlusions by proposing a spatially aware vision-language-action model that jointly performs target localization, visibility estimation, and continuous flight action generation. The model bridges the gap from imitation learning on visible frames to spatial closed-loop control through spatial grounding continual pretraining (CPT), a three-stage curriculum-supervised fine-tuning (SFT) strategy, chain-of-thought (CoT) reasoning, and closed-loop reinforcement learning. Compared to OpenVLA, the proposed approach reduces open-loop trajectory error (ADE) by 34.1% and 35.3% on seen and unseen test scenarios, respectively, and improves closed-loop tracking success rates by 29.8% and 2.5%, substantially enhancing robustness in occlusion-prone environments.

0 citationsRead paper

gDMC: A Generic Distributed Model Counting Framework via Work-Stealing

Jul 15, 2026

This work addresses the limited scalability of propositional model counting (#SAT) on single-core architectures and the inefficiencies of existing distributed approaches, which often suffer from high initialization overhead or inflexible designs. The paper proposes a general-purpose distributed framework for exact model counting that, for the first time, decouples solver logic from parallel scheduling via C++ template metaprogramming, enabling mainstream solvers to be parallelized with minimal modifications. Additionally, it introduces an adaptive work-stealing mechanism that dynamically balances computational load across workers. Evaluated on standard competition benchmarks, the approach achieves near-linear speedup and significantly outperforms current distributed #SAT solvers.

0 citationsRead paper