arXiv subjects

Zehan Wu

Publications and source records attributed to Zehan Wu.

4 recordsLinked to original sources

arXiv2026

Towards unified brain-to-text decoding across speech production and perception

Speech production and perception are the main ways humans communicate daily. Prior brain-to-text decoding studies have largely focused on a single modality and alphabetic languages. Here, we present a unified brain-to-sentence decoding framework for both speech production and perception in Mandarin Chinese. The framework exhibits strong generalization ability, enabling sentence-level decoding when trained only on single-character data and supporting characters and syllables unseen during training. In addition, it allows direct and controlled comparison of neural dynamics across modalities. Mandarin speech is decoded by first classifying syllable components in Hanyu Pinyin, namely initials and finals, from neural signals, followed by a post-trained large language model (LLM) that maps sequences of toneless Pinyin syllables to Chinese sentences. To enhance LLM decoding, we designed a three-stage post-training and two-stage inference framework based on a 7-billion-parameter LLM, achieving overall performance that exceeds larger commercial LLMs with hundreds of billions of parameters or more. In addition, several characteristics were observed in Mandarin speech production and perception: speech production involved neural responses across broader cortical regions than auditory perception; channels responsive to both modalities exhibited similar activity patterns, with speech perception showing a temporal delay relative to production; and decoding performance was broadly comparable across hemispheres. Our work not only establishes the feasibility of a unified decoding framework but also provides insights into the neural characteristics of Mandarin speech production and perception. These advances contribute to brain-to-text decoding in logosyllabic languages and pave the way toward neural language decoding systems supporting multiple modalities.

q-bio.NC↗

arXiv2025

WishGI: Lightweight Static Global Illumination Baking via Spherical Harmonics Fitting

Global illumination combines direct and indirect lighting to create realistic lighting effects, bringing virtual scenes closer to reality. Static global illumination is a crucial component of virtual scene rendering, leveraging precomputation and baking techniques to significantly reduce runtime computational costs. Unfortunately, many existing works prioritize visual quality by relying on extensive texture storage and massive pixel-level texture sampling, leading to large performance overhead. In this paper, we introduce an illumination reconstruction method that effectively reduces sampling in fragment shader and avoids additional render passes, making it well-suited for low-end platforms. To achieve high-quality global illumination with reduced memory usage, we adopt a spherical harmonics fitting approach for baking effective illumination information and propose an inverse probe distribution method that generates unique probe associations for each mesh. This association, which can be generated offline in the local space, ensures consistent lighting quality across all instances of the same mesh. As a consequence, our method delivers highly competitive lighting effects while using only approximately 5% of the memory required by mainstream industry techniques.

cs.GR↗

arXiv2025

Precise constraint on properties of neutron stars through new universal relations and astronomical observations

In view of the great uncertainty of the equation of state (EOS) of high-density nuclear matter, establishing EOS-independent universal relations between global properties of neutron stars provides a practical way to constrain the unobservable or difficult-to-observe properties through astronomical observations. It is common to construct universal relations between EOS-dependent properties (e.g., moment of inertia, tidal deformation, etc.) or combined properties (e.g., compactness). Improving the precision of the universal relations may provide stricter constraint on the properties of neutron star. We find that in 3-dimensional space with mass and radius as the base coordinates, the points corresponding to a certain property of neutron star described by different EOSs are almost located in the same surface. Thus the universal relation between the property and the stellar mass-radius can be expressed through describing the surface. It is shown that the resulting universal relations have higher precisions. As an example, we construct high-precision universal relations for the moment of inertia, the $f$-mode frequency, and the dimensionless tidal deformation respect to the mass-radius. As the observational data of neutron star mass and radius from NICER grows in data and accuracy, these universal relations allow for more precise constraints on the unobservable or difficult-to-observe properties.

astro-ph.HE↗

arXiv2023

From masses and radii of neutron stars to EOS of nuclear matter through neural network

The equation of state (EOS) of dense nuclear matter is a key factor to determine the internal structure and properties of neutron stars. However, the EOS of high-density nuclear matter has great uncertainty mainly because the terrestrial nuclear experiments cannot reproduce matter as dense as that in the inner core of a neutron star. Fortunately, continuous improvements in astronomical observations of neutron stars provide the opportunity to inversely constrain the EOS of high-density nuclear matter. A number of methods have been proposed to implement this inverse constraint, such as the Bayesian analysis algorithm, the Lindblom's approach, and so on. Neural network algorithm is an effective new method developed in recent years. By employing a set of isospin-dependent parametric EOSs as the training sample of neural network algorithm, we set up an effective way to reconstruct the EOS with relative accuracy through a few mass-radius data. Based on the obtained neural network algorithms and according to the NICER observations on masses and radii of neutron stars with assumed precision, we get the inversely constrained EOS and further calculate the corresponding macroscopic properties of the neutron star. The results are basically consistent with the constraint on EOS from the Huth $et~ al.$ based on Bayesian analysis. Moreover, the results show that even though the neural network algorithm was obtained by using the finite parameterized EOS as the training set, it is valid for any rational parameter combination of the parameterized EOS model.

nucl-th↗