Selected work

Representative Papers

Decoupling Time and Space: A Temporally Conditioned Refinement for EEG Source Imaging

Oct 05, 2026

This study addresses the ill-posed nature of spatial inversion in EEG source imaging and the inherent trade-off between temporal modeling capability and inference cost. To this end, we propose a dual-stream decoupled framework that explicitly separates spatiotemporal processing to reduce computational overhead. Methodologically, a Transformer-based temporal encoder extracts global dynamic conditions, while a factorized spatiotemporal attention mechanism preserves sensor-level features. A source-space refiner then enables high-fidelity, time-step-wise reconstruction. Experimental results demonstrate that the proposed model significantly outperforms baselines under challenging high-noise, multi-source scenarios on synthetic data. Furthermore, it successfully transfers to real-world EEG age-group decoding tasks, achieving synergistic improvements in both accuracy and efficiency.

0 citationsRead paper

Learning Deterministic and Stochastic Forced Hamiltonian Systems

Aug 20, 2026

本文提出一种基于拉格朗日-达朗贝尔原理的几何框架，通过构建广义强迫哈密顿神经网络(GFHNNs)来学习确定性和随机性强迫哈密顿系统，提高了长期稳定性和准确性。

0 citationsRead paper

"Nobody should control the end user": Exploring Privacy Perspectives of Indian Internet Users in Light of DPDPA

Aug 25, 2025

This study investigates Indian internet users’ awareness, attitudes, and concerns regarding cookie banners, online privacy, and key provisions of the Digital Personal Data Protection Act (DPDPA)—notably government exemptions and consent mechanisms—during its initial implementation phase. Employing a mixed-methods design, it administered an anonymous survey to 428 users and conducted thematic coding of 143 open-ended responses. The study provides the first empirical evidence of a pronounced gap between users’ stated privacy concerns and actual behavioral practices, identifying deep public distrust in governmental data powers as a central determinant of privacy stance. It further innovatively reveals how the DPDPA’s accountability exemptions and proceduralized consent framework exacerbate legitimacy anxieties. Based on these findings, the study proposes user-centered policy refinements: enhancing transparency, redesigning consent as a dynamic, context-aware mechanism, and institutionalizing regulatory checks and balances.

0 citationsRead paper

Learning Image Fractals Using Chaotic Differentiable Point Splatting

Feb 24, 2025

This work addresses the fractal inverse problem in image modeling—recovering Iterated Function System (IFS) parameters from a single natural image to capture self-similarity and enable artifact-free, arbitrary-scale synthesis. We propose the first joint optimization framework integrating chaotic optimization with differentiable point-splatting rendering: chaotic dynamics enhance global search capability, enabling escape from local minima in complex energy landscapes; differentiable rendering facilitates end-to-end gradient-based optimization of IFS affine parameters. This hybrid stochastic-deterministic algorithm significantly improves fractal code reconstruction fidelity. In comprehensive benchmark evaluations, it achieves an average PSNR gain of 3.2 dB over state-of-the-art methods. Moreover, it supports depth scaling up to 100× while preserving rich hierarchical detail without ringing artifacts or blocking effects.

0 citationsRead paper

Recent publications

Latest Papers

Decoupling Time and Space: A Temporally Conditioned Refinement for EEG Source Imaging

Oct 05, 2026

This study addresses the ill-posed nature of spatial inversion in EEG source imaging and the inherent trade-off between temporal modeling capability and inference cost. To this end, we propose a dual-stream decoupled framework that explicitly separates spatiotemporal processing to reduce computational overhead. Methodologically, a Transformer-based temporal encoder extracts global dynamic conditions, while a factorized spatiotemporal attention mechanism preserves sensor-level features. A source-space refiner then enables high-fidelity, time-step-wise reconstruction. Experimental results demonstrate that the proposed model significantly outperforms baselines under challenging high-noise, multi-source scenarios on synthetic data. Furthermore, it successfully transfers to real-world EEG age-group decoding tasks, achieving synergistic improvements in both accuracy and efficiency.

0 citationsRead paper

Learning Deterministic and Stochastic Forced Hamiltonian Systems

Aug 20, 2026

本文提出一种基于拉格朗日-达朗贝尔原理的几何框架，通过构建广义强迫哈密顿神经网络(GFHNNs)来学习确定性和随机性强迫哈密顿系统，提高了长期稳定性和准确性。

0 citationsRead paper

"Nobody should control the end user": Exploring Privacy Perspectives of Indian Internet Users in Light of DPDPA

Aug 25, 2025

This study investigates Indian internet users’ awareness, attitudes, and concerns regarding cookie banners, online privacy, and key provisions of the Digital Personal Data Protection Act (DPDPA)—notably government exemptions and consent mechanisms—during its initial implementation phase. Employing a mixed-methods design, it administered an anonymous survey to 428 users and conducted thematic coding of 143 open-ended responses. The study provides the first empirical evidence of a pronounced gap between users’ stated privacy concerns and actual behavioral practices, identifying deep public distrust in governmental data powers as a central determinant of privacy stance. It further innovatively reveals how the DPDPA’s accountability exemptions and proceduralized consent framework exacerbate legitimacy anxieties. Based on these findings, the study proposes user-centered policy refinements: enhancing transparency, redesigning consent as a dynamic, context-aware mechanism, and institutionalizing regulatory checks and balances.

0 citationsRead paper

Learning Image Fractals Using Chaotic Differentiable Point Splatting

Feb 24, 2025

This work addresses the fractal inverse problem in image modeling—recovering Iterated Function System (IFS) parameters from a single natural image to capture self-similarity and enable artifact-free, arbitrary-scale synthesis. We propose the first joint optimization framework integrating chaotic optimization with differentiable point-splatting rendering: chaotic dynamics enhance global search capability, enabling escape from local minima in complex energy landscapes; differentiable rendering facilitates end-to-end gradient-based optimization of IFS affine parameters. This hybrid stochastic-deterministic algorithm significantly improves fractal code reconstruction fidelity. In comprehensive benchmark evaluations, it achieves an average PSNR gain of 3.2 dB over state-of-the-art methods. Moreover, it supports depth scaling up to 100× while preserving rich hierarchical detail without ringing artifacts or blocking effects.

0 citationsRead paper