TriLens: Per-Layer Logit-Lens Entropy for White-Box Hallucination Detection

Bohan Yang, Yijun Gong, Zhi Zhang, Ge Zhang, Wenpeng Xing^#, Meng Han^#

Under review First Author

TriLens introduces per-layer logit-lens entropy as a white-box signal for hallucination detection in large language models.

[Paper] [Project]

QuadBox: Accelerating 3D Gaussian Splatting with Geometry-Aware Boxes

Xinze Li, Bohan Yang, Pengxu Chen, Yiyuan Wang, Hongcheng Luo, Wentao Cheng^#, Weifeng Su^#

IEEE International Conference on Image Processing (ICIP) 2026 Published

QuadBox accelerates 3D Gaussian Splatting using geometry-aware boxes for more efficient scene representation and rendering.

[Paper] [Code]

Latent Fusion Jailbreak: Blending Harmful and Harmless Representations to Elicit Unsafe LLM Outputs

Wenpeng Xing, Bohan Yang, Zaifeng Yang, Changting Lin, Meng Han^#

Under review Student First Author

Latent Fusion Jailbreak studies how blending harmful and harmless representations can elicit unsafe outputs from large language models.

[Paper]

DECENT: Distribution-aware Prototypical Calibration with Fine-grained Voting for VLM Misclassification Detection

Yijun Gong*, Bohan Yang*, Zhi Zhang, Yupeng Qin, Wenpeng Xing, Changting Lin, Meng Han, Ziyue Qiao^#, Xiao Luo^#

Under review Co-First Author

DECENT studies distribution-aware prototypical calibration and fine-grained voting for detecting misclassifications in vision-language models.

COLOR: Coarse-to-fine Outlier Detection with Subspace Purification for Open-Vocabulary Semantic Segmentation

Yijun Gong*, Bohan Yang*, Ziyue Qiao^#, Xiao Luo^#

Under review Co-First Author

COLOR explores coarse-to-fine outlier detection with subspace purification for open-vocabulary semantic segmentation.