A Non-Linear Neuron Based Detection of Isolated Pixels in Binary and Grayscale Images using Contrast Sensitive Receptive Fields
本文提出了一种基于非线性神经元模型的方法，通过对比敏感感受野检测二值和灰度图像中的孤立像素点，解决了现有方法在噪声敏感性和参数设定上的局限。
0 citationsRead paper
Selected work
Representative Papers
Robust Validation to Geometric Perturbations for Autonomous Pose Estimation
该研究针对视觉导航中几何扰动导致的自主姿态估计问题，采用全局Lipschitz优化方法有效验证并提高了系统的鲁棒性。
0 citationsRead paper
Propaganda Forensics: Recovering the Generation Pipeline of an AI-Driven Influence Campaign
This study addresses the challenge of tracing generative pipelines in AI-driven influence operations by constructing Propagia, the first French propaganda corpus. By integrating topic modeling, sentiment analysis, prompt leakage detection, and rewriting-based attribution techniques, this work reverse-engineers the content generation process. The research reveals distinct stylistic characteristics of AI-generated propaganda, identifies evidence of prompt leakage across 50 websites, and successfully attributes generated content to Llama-3 and Mistral models. Collectively, these findings establish a systematic methodological framework for the forensic analysis and provenance identification of AI-generated content, offering critical insights into detecting and mitigating automated disinformation campaigns.
0 citationsRead paper
Homomorphic Aggregation of Continuous-Variable GKP States
In distributed quantum computing, logical states encoded in continuous-variable Gottesman–Kitaev–Preskill (GKP) codes face significant challenges in achieving high-fidelity fusion via passive linear optics due to finite squeezing and entanglement-induced decoherence. This work proposes an active, measurement-based protocol that leverages auxiliary GKP Bell states, homodyne detection, and feedforward control to implement a completely positive trace-preserving map preserving the geometric structure of the logical code space. The protocol enables, for the first time, homomorphic logical addition of GKP states across multiple nodes. It further exhibits approximate quantum non-demolition characteristics, rigorously bounds information leakage under one-time-pad encryption, and derives an analytical upper bound on logical fidelity under finite squeezing. These results demonstrate the scheme’s ability to simultaneously suppress decoherence and information leakage while effectively preserving logical information integrity.
0 citationsRead paper
From Keypoints to Predictive Distributions: Post-Hoc Uncertainty for YOLO-Pose Models
This work addresses the lack of spatial uncertainty modeling in YOLO-Pose for keypoint localization. The authors propose a lightweight, post-hoc probabilistic extension that introduces an additional probability head to predict input-dependent 2×2 covariance matrices, enabling calibrated bivariate Gaussian or Student-t distributions over original keypoints. This is the first approach to equip YOLO-Pose with keypoint-level predictive distributions. A novel evaluation protocol is introduced, combining distribution calibration diagnostics with Average Keypoint Precision (AKP). Experiments on COCO demonstrate that the method effectively supports reliability-based keypoint ranking, with the Student-t formulation yielding more accurate residual distribution fitting. Furthermore, in an aircraft visual landing task, the calibrated covariance enables uncertainty-aware pose estimation and sensor fusion.
0 citationsRead paper
Recent publications
Latest Papers
A Non-Linear Neuron Based Detection of Isolated Pixels in Binary and Grayscale Images using Contrast Sensitive Receptive Fields
本文提出了一种基于非线性神经元模型的方法，通过对比敏感感受野检测二值和灰度图像中的孤立像素点，解决了现有方法在噪声敏感性和参数设定上的局限。
0 citationsRead paper
Robust Validation to Geometric Perturbations for Autonomous Pose Estimation
该研究针对视觉导航中几何扰动导致的自主姿态估计问题，采用全局Lipschitz优化方法有效验证并提高了系统的鲁棒性。
0 citationsRead paper
Propaganda Forensics: Recovering the Generation Pipeline of an AI-Driven Influence Campaign
This study addresses the challenge of tracing generative pipelines in AI-driven influence operations by constructing Propagia, the first French propaganda corpus. By integrating topic modeling, sentiment analysis, prompt leakage detection, and rewriting-based attribution techniques, this work reverse-engineers the content generation process. The research reveals distinct stylistic characteristics of AI-generated propaganda, identifies evidence of prompt leakage across 50 websites, and successfully attributes generated content to Llama-3 and Mistral models. Collectively, these findings establish a systematic methodological framework for the forensic analysis and provenance identification of AI-generated content, offering critical insights into detecting and mitigating automated disinformation campaigns.
0 citationsRead paper
Homomorphic Aggregation of Continuous-Variable GKP States
In distributed quantum computing, logical states encoded in continuous-variable Gottesman–Kitaev–Preskill (GKP) codes face significant challenges in achieving high-fidelity fusion via passive linear optics due to finite squeezing and entanglement-induced decoherence. This work proposes an active, measurement-based protocol that leverages auxiliary GKP Bell states, homodyne detection, and feedforward control to implement a completely positive trace-preserving map preserving the geometric structure of the logical code space. The protocol enables, for the first time, homomorphic logical addition of GKP states across multiple nodes. It further exhibits approximate quantum non-demolition characteristics, rigorously bounds information leakage under one-time-pad encryption, and derives an analytical upper bound on logical fidelity under finite squeezing. These results demonstrate the scheme’s ability to simultaneously suppress decoherence and information leakage while effectively preserving logical information integrity.
0 citationsRead paper
From Keypoints to Predictive Distributions: Post-Hoc Uncertainty for YOLO-Pose Models
This work addresses the lack of spatial uncertainty modeling in YOLO-Pose for keypoint localization. The authors propose a lightweight, post-hoc probabilistic extension that introduces an additional probability head to predict input-dependent 2×2 covariance matrices, enabling calibrated bivariate Gaussian or Student-t distributions over original keypoints. This is the first approach to equip YOLO-Pose with keypoint-level predictive distributions. A novel evaluation protocol is introduced, combining distribution calibration diagnostics with Average Keypoint Precision (AKP). Experiments on COCO demonstrate that the method effectively supports reliability-based keypoint ranking, with the Student-t formulation yielding more accurate residual distribution fitting. Furthermore, in an aircraft visual landing task, the calibrated covariance enables uncertainty-aware pose estimation and sensor fusion.
0 citationsRead paper