arXiv subjects

Yanwei Liu

Publications and source records attributed to Yanwei Liu.

14 recordsLinked to original sources

arXiv2026

Rotation-Invariant Spherical Watermarking via Third-Order SO(3) Representation Coupling

Reliable watermarking of panoramic imagery is fundamentally challenged by arbitrary 3D rotations. As panoramas are defined on the sphere, they naturally transform under the action of $SO(3)$, rendering conventional planar representations and augmentation-based robustness strategies inadequate and devoid of theoretical guarantees. To address this, we formulate panoramas as spherical signals and leverage $SO(3)$ representation theory to derive provably rotation-invariant descriptors. While spherical harmonic coefficients transform equivariantly under rotations, the natural invariant constructions are typically limited to zeroth-order statistics which eliminate directional information and severely constrain embedding capacity. In this work, we introduce a principled third-order invariant construction by coupling higher-order $SO(3)$ irreducible representations via tensor products and projecting onto the trivial representation. This yields a spherical invariant bispectrum that preserves phase information while remaining strictly rotation-invariant. Leveraging this property, we embed watermarks into higher-order spherical harmonic coefficients and recover them from invariant bispectral scalars, enabling reliable extraction under arbitrary 3D rotations. We provide a theoretical proof of $SO(3)$ invariance for it and demonstrate experimentally its near-perfect robustness to continuous rotations while maintaining high visual fidelity.

cs.CV↗

arXiv2026

Rel-Zero: Harnessing Patch-Pair Invariance for Robust Zero-Watermarking Against AI Editing

Recent advancements in diffusion-based image editing pose a significant threat to the authenticity of digital visual content. Traditional embedding-based watermarking methods often introduce perceptible perturbations to maintain robustness, inevitably compromising visual fidelity. Meanwhile, existing zero-watermarking approaches, typically relying on global image features, struggle to withstand sophisticated manipulations. In this work, we uncover a key observation: while individual image patches undergo substantial alterations during AI-based editing, the relational distance between patch pairs remains relatively invariant. Leveraging this property, we propose Relational Zero-Watermarking (Rel-Zero), a novel framework that requires no modification to the original image but derives a unique zero-watermark from these editing-invariant patch relations. By grounding the watermark in intrinsic structural consistency rather than absolute appearance, Rel-Zero provides a non-invasive yet resilient mechanism for content authentication. Extensive experiments demonstrate that Rel-Zero achieves substantially improved robustness across diverse editing models and manipulations compared to prior zero-watermarking approaches.

cs.CV↗

arXiv2026

DRGW: Learning Disentangled Representations for Robust Graph Watermarking

Graph-structured data is foundational to numerous web applications, and watermarking is crucial for protecting their intellectual property and ensuring data provenance. Existing watermarking methods primarily operate on graph structures or entangled graph representations, which compromise the transparency and robustness of watermarks due to the information coupling in representing graphs and uncontrollable discretization in transforming continuous numerical representations into graph structures. This motivates us to propose DRGW, the first graph watermarking framework that addresses these issues through disentangled representation learning. Specifically, we design an adversarially trained encoder that learns an invariant structural representation against diverse perturbations and derives a statistically independent watermark carrier, ensuring both robustness and transparency of watermarks. Meanwhile, we devise a graph-aware invertible neural network to provide a lossless channel for watermark embedding and extraction, guaranteeing high detectability and transparency of watermarks. Additionally, we develop a structure-aware editor that resolves the issue of latent modifications into discrete graph edits, ensuring robustness against structural perturbations. Experiments on diverse benchmark datasets demonstrate the superior effectiveness of DRGW.

cs.LG↗

arXiv2026

Attesting Model Lineage by Consisted Knowledge Evolution with Fine-Tuning Trajectory

The fine-tuning technique in deep learning gives rise to an emerging lineage relationship among models. This lineage provides a promising perspective for addressing security concerns such as unauthorized model redistribution and false claim of model provenance, which are particularly pressing in \textcolor{blue}{open-weight model} libraries where robust lineage verification mechanisms are often lacking. Existing approaches to model lineage detection primarily rely on static architectural similarities, which are insufficient to capture the dynamic evolution of knowledge that underlies true lineage relationships. Drawing inspiration from the genetic mechanism of human evolution, we tackle the problem of model lineage attestation by verifying the joint trajectory of knowledge evolution and parameter modification. To this end, we propose a novel model lineage attestation framework. In our framework, model editing is first leveraged to quantify parameter-level changes introduced by fine-tuning. Subsequently, we introduce a novel knowledge vectorization mechanism that refines the evolved knowledge within the edited models into compact representations by the assistance of probe samples. The probing strategies are adapted to different types of model families. These embeddings serve as the foundation for verifying the arithmetic consistency of knowledge relationships across models, thereby enabling robust attestation of model lineage. Extensive experimental evaluations demonstrate the effectiveness and resilience of our approach in a variety of adversarial scenarios in the real world. Our method consistently achieves reliable lineage verification across a broad spectrum of model types, including classifiers, diffusion models, and large language models.

cs.CR↗

arXiv2025

Training Report of TeleChat3-MoE

TeleChat3-MoE is the latest series of TeleChat large language models, featuring a Mixture-of-Experts (MoE) architecture with parameter counts ranging from 105 billion to over one trillion,trained end-to-end on Ascend NPU cluster. This technical report mainly presents the underlying training infrastructure that enables reliable and efficient scaling to frontier model sizes. We detail systematic methodologies for operator-level and end-to-end numerical accuracy verification, ensuring consistency across hardware platforms and distributed parallelism strategies. Furthermore, we introduce a suite of performance optimizations, including interleaved pipeline scheduling, attention-aware data scheduling for long-sequence training,hierarchical and overlapped communication for expert parallelism, and DVM-based operator fusion. A systematic parallelization framework, leveraging analytical estimation and integer linear programming, is also proposed to optimize multi-dimensional parallelism configurations. Additionally, we present methodological approaches to cluster-level optimizations, addressing host- and device-bound bottlenecks during large-scale training tasks. These infrastructure advancements yield significant throughput improvements and near-linear scaling on clusters comprising thousands of devices, providing a robust foundation for large-scale language model development on hardware ecosystems.

cs.CL↗

arXiv2025

The micro-Doppler Attack Against AI-based Human Activity Classification from Wireless Signals

A subset of Human Activity Classification (HAC) systems are based on AI algorithms that use passively collected wireless signals. This paper presents the micro-Doppler attack targeting HAC from wireless orthogonal frequency division multiplexing (OFDM) signals. The attack is executed by inserting artificial variations in a transmitted OFDM waveform to alter its micro-Doppler signature when it reflects off a human target. We investigate two variants of our scheme that manipulate the waveform at different time scales resulting in altered receiver spectrograms. HAC accuracy with a deep convolutional neural network (CNN) can be reduced to less than 10%.

eess.SP↗

arXiv2025

Towards Understanding How Knowledge Evolves in Large Vision-Language Models

Large Vision-Language Models (LVLMs) are gradually becoming the foundation for many artificial intelligence applications. However, understanding their internal working mechanisms has continued to puzzle researchers, which in turn limits the further enhancement of their capabilities. In this paper, we seek to investigate how multimodal knowledge evolves and eventually induces natural languages in LVLMs. We design a series of novel strategies for analyzing internal knowledge within LVLMs, and delve into the evolution of multimodal knowledge from three levels, including single token probabilities, token probability distributions, and feature encodings. In this process, we identify two key nodes in knowledge evolution: the critical layers and the mutation layers, dividing the evolution process into three stages: rapid evolution, stabilization, and mutation. Our research is the first to reveal the trajectory of knowledge evolution in LVLMs, providing a fresh perspective for understanding their underlying mechanisms. Our codes are available at https://github.com/XIAO4579/Vlm-interpretability.

cs.CV↗

arXiv2025

Cross-Layer-Optimized Link Selection for Hologram Video Streaming over Millimeter Wave Networks

Holographic-type communication brings an immersive tele-holography experience by delivering holographic contents to users. As the direct representation of holographic contents, hologram videos are naturally three-dimensional representation, which consist of a huge volume of data. Advanced multi-connectivity (MC) millimeter-wave (mmWave) networks are now available to transmit hologram videos by providing the necessary bandwidth. However, the existing link selection schemes in MC-based mmWave networks neglect the source content characteristics of hologram videos and the coordination among the parameters of different protocol layers in each link, leading to sub-optimal streaming performance. To address this issue, we propose a cross-layer-optimized link selection scheme for hologram video streaming over mmWave networks. This scheme optimizes link selection by jointly adjusting the video coding bitrate, the modulation and channel coding schemes (MCS), and link power allocation to minimize the end-to-end hologram distortion while guaranteeing the synchronization and quality balance between real and imaginary components of the hologram. Results show that the proposed scheme can effectively improve the hologram video streaming performance in terms of PSNR by 1.2dB to 6.4dB against the non-cross-layer scheme.

eess.SP↗

arXiv2025

Power-Efficient Deceptive Wireless Beamforming Against Eavesdroppers

Eavesdroppers of wireless signals want to infer as much as possible regarding the transmitter (Tx). Popular methods to minimize information leakage to the eavesdropper include covert communication, directional modulation, and beamforming with nulling. In this paper we do not attempt to prevent information leakage to the eavesdropper like the previous methods. Instead we propose to beamform the wireless signal at the Tx in such a way that it incorporates deceptive information. The beamformed orthogonal frequency division multiplexing (OFDM) signal includes a deceptive value for the Doppler (velocity) and range of the Tx. To design the optimal baseband waveform with these characteristics, we define and solve an optimization problem for power-efficient deceptive wireless beamforming (DWB). The relaxed convex Quadratic Program (QP) is solved using a heuristic algorithm. Our simulation results indicate that our DWB scheme can successfully inject deceptive information with low power consumption, while preserving the shape of the created beam.

eess.SP↗

arXiv2024

Know2Vec: A Black-Box Proxy for Neural Network Retrieval

For general users, training a neural network from scratch is usually challenging and labor-intensive. Fortunately, neural network zoos enable them to find a well-performing model for directly use or fine-tuning it in their local environments. Although current model retrieval solutions attempt to convert neural network models into vectors to avoid complex multiple inference processes required for model selection, it is still difficult to choose a suitable model due to inaccurate vectorization and biased correlation alignment between the query dataset and models. From the perspective of knowledge consistency, i.e., whether the knowledge possessed by the model can meet the needs of query tasks, we propose a model retrieval scheme, named Know2Vec, that acts as a black-box retrieval proxy for model zoo. Know2Vec first accesses to models via a black-box interface in advance, capturing vital decision knowledge from models while ensuring their privacy. Next, it employs an effective encoding technique to transform the knowledge into precise model vectors. Secondly, it maps the user's query task to a knowledge vector by probing the semantic relationships within query samples. Furthermore, the proxy ensures the knowledge-consistency between query vector and model vectors within their alignment space, which is optimized through the supervised learning with diverse loss functions, and finally it can identify the most suitable model for a given task during the inference stage. Extensive experiments show that our Know2Vec achieves superior retrieval accuracy against the state-of-the-art methods in diverse neural network retrieval tasks.

cs.LG↗

arXiv2024

Generation of Spatially Coherent Light at Extreme Ultraviolet Wavelengths

We present spatial coherence measurements of extreme-ultraviolet light generated using the process of high-harmonic upconversion of a femtosecond laser. Using a phase-matched hollow-fiber geometry, the generated beam is found to exhibit essentially full spatial coherence. The coherence of this laser-like EUV source is demonstrated by recording Gabor holograms of small objects. This work demonstrates the capability to do EUV holography using a tabletop experimental setup. Such an EUV source, with low divergence and high spatial coherence, can be used for experiments such as high-precision metrology, inspection of optical components for EUV lithography (1), and for microscopy and holography (2) with nanometer resolution. Furthermore, the short time duration of the EUV radiation (a few femtoseconds) will enable EUV microscopy and holography to be performed with ultrahigh time resolution.

physics.optics↗

arXiv2024

X-ray Microscopy and Talbot Imaging with the Matter in Extreme Conditions X-ray Imager at LCLS

The last decade has shown the great potential that X-ray Free Electron Lasers (FEL) have to study High Energy Density (HED) physics. Experiments at FELs have made significant breakthroughs in Shock Physics and Dynamic Diffraction, Dense Plasma Physics and Warm Dense Matter Science, using techniques such as isochoric heating, inelastic scattering, small angle scattering and X-ray diffraction. In addition, and complementary to these techniques, the coherent properties of the FEL beam can be used to image HED samples with high fidelity. We present new imaging diagnostics and techniques developed at the Matter in Extreme Conditions (MEC) instrument at Linac Coherent Light Source (LCLS) over the last few years. We show results in Phase Contrast Imaging geometry, where the X-ray beam propagates from the target to a camera revealing its phase, as well as in Direct Imaging geometry, where a real image of the sample plane is produced in the camera with a spatial resolution down to 200 nm. Last, we show an implementation of the Talbot Imaging method allowing both X-ray phase and intensity measurements change introduced by a target with sub-micron resolution.

physics.ins-det↗

arXiv2023

A compact single-shot soft X-ray photon spectrometer for free electron laser diagnostics

The photon spectrum from free-electron laser (FEL) light sources offers valuable information in time-resolved experiments and machine optimization in the spectral and temporal domains. We have developed a compact single-shot photon spectrometer to diagnose soft X-ray spectra. The spectrometer consists of an array of off-axis Fresnel zone plates (FZP) that act as transmission-imaging gratings, a Ce-YAG scintillator, and a microscope objective to image the scintillation target onto a two-dimensional imaging detector. This spectrometer operates in an energy range which covers absorption edges associated with several atomic constituents carbon, nitrogen, oxygen, and neon. The spectrometer's performance is demonstrated at a repetition rate of 120 Hz, but our detection scheme can be easily extended to 200 kHz spectral collection by employing a fast complementary metal oxide semiconductor (CMOS) line-scan camera to detect the light from the scintillator. This compact photon spectrometer provides an opportunity for monitoring the spectrum downstream of an endstation in a limited space environment with subelectronvolt energy resolution.

physics.ins-det↗

arXiv2016

Striped nanoscale phase separation at the metal-insulator transition of heteroepitaxial nickelates

Nucleation processes of mixed-phase states are an intrinsic characteristic of first-order phase transitions, typically related to local symmetry breaking. Direct observation of emerging mixed-phase regions in materials showing a first-order metal-insulator transition (MIT) offers unique opportunities to uncover their driving mechanism. Using photoemission electron microscopy, we image the nanoscale formation and growth of insulating domains across the temperature-driven MIT in NdNiO3 epitaxial thin films. Heteroepitaxy is found to strongly determine the nanoscale nature of the phase transition, inducing preferential formation of striped domains along the terraces of atomically flat stepped surfaces. We show that the distribution of transition temperatures is an intrinsic local property, set by surface morphology and stable across multiple temperature cycles. Our data provides new insights into the MIT of heteroepitaxial nickelates and points to a rich, nanoscale phenomenology in this strongly correlated material.

cond-mat.str-el↗