Searcharxiv⌕ Search

arXiv subjects

Zhe Li

Publications and source records attributed to Zhe Li.

At least 55 records · Page 3Linked to original sources
arXiv2026

Utilizing Improper Gaussian Signaling for Downlink Rate-Splitting Multiple Access systems with Imperfect Successive Interference Cancellation

To mitigate the residual interference from imperfect successive interference cancellation (SIC) in Rate-Splitting Multiple Access (RSMA), this paper incorporates improper Gaussian signaling (IGS) into the downlink RSMA framework. Unlike existing RSMA--IGS works that embed impropriety within IQ-imbalanced frameworks, we show that IGS alone effectively counters SIC-induced residual interference. A basic Single-Input Single-Output (SISO) setup is considered with IGS on the common stream and proper Gaussian signaling (PGS) on the private streams, and it is analyzed in three parts. First, we show that the private rate maximization is achieved at the maximum signal impropriety. Second, we derive closed-form optimal solutions with rigorous monotonicity conditions for the common rate maximization. Third, to address the non-convex sum rate maximization, we develop a soft actor-critic (SAC) based algorithm that efficiently explores the solution space. Theoretical analysis and numerical results demonstrate that IGS consistently surpasses PGS, and its performance advantage becomes more pronounced when the SIC imperfection becomes severe.

eess.SP↗
arXiv2026

DeSeG: Decoupling Semantic Intent and Geometric Constraints for Physically Plausible Human-Scene Interaction

Synthesizing physically plausible human-scene interactions (HSI) remains a critical challenge in computer vision and the development of human avatars. Although recent generative models enable diverse motion synthesis, they suffer from an inductive bias referred to as semantic-geometric entanglement. Because spatial constraints often strongly correlate with specific actions in training data, monolithic models will learn the shortcut bias, aggressively overriding the semantic intent when faced with strict geometric cues. Furthermore, this entanglement exacerbates physical hallucinations, such as body-scene penetrations. To address these limitations, we propose DeSeG, a hierarchical framework that explicitly decouples semantic intent from geometric constraints. First, we introduce a Residual Semantic Planner that encodes textual instructions and canonicalized goal voxels into a compact latent space, enabling fine-grained semantic control independent of spatial trajectories. Second, we propose a physics regularized diffusion executor that incorporates differentiable repulsive potential fields directly into the diffusion objective, enforcing collision-aware motion generation. Extensive experiments on the Lingo dataset demonstrate that DeSeG achieves state-of-the-art performance, reducing mean scene penetration by 47% and improving semantic alignment by 29% over the SOTA baselines.

cs.CV↗
arXiv2026

DT-Guard: Intent-Driven Reasoning-Active Training for Reasoning-Free LLM Safety Guardrail

Large language models deployed in open-world applications require safety guardrails that are both robust to complex risks and efficient enough for low-latency runtime moderation. Existing guardrails face a practical trade-off between lightweight classification-based models, which are efficient but often struggle with concealed intent, ambiguous semantics, and borderline safety decisions, and reasoning-based guards, which improve judgment quality but introduce additional token generation and inference latency. We present DT-Guard, a content safety guardrail model based on a Reasoning-Active Training, Reasoning-Free Inference paradigm. The key idea is to use reasoning supervision during training while emitting only structured safety labels at inference time. DT-Guard formulates safety judgment as a progressive decision process, Intent - Category - Safety, and constructs an intent-driven dataset with intent labels, risk categories, safety labels, and structured reasoning trajectories. To further improve hard-case robustness, we propose Rollout-Guided Progressive Hard-Case Optimization (RG-PHO), which uses multi-rollout consistency to identify stably mastered, persistently failed, and preference-unstable samples, and applies targeted supervised and preference optimization accordingly. At inference time, DT-Guard directly generates structured labels without explicit reasoning traces, preserving deployment efficiency. Experiments on prompt-side and response-side safety benchmarks show that DT-Guard achieves average F1 scores of 0.886 and 0.870, respectively. With only a 4B backbone, it reaches a dual-side average F1 of 0.878, outperforming strong 8B guardrail baselines. These results demonstrate that reasoning supervision can be effectively internalized into low-latency safety discrimination.

cs.AI↗
arXiv2026

Consistent Selection of the Number of Groups in Panel Models via Cross-Validation

Group number selection is a key problem for group panel data modeling. In this work, we develop a cross-validation (CV) method to tackle this problem. Specifically, we split the panel data into two data folds on the time span with a buffer zone, with group structure preserved for individuals. We first estimate the group memberships and parameters on one data fold, then plug in the estimates and utilize the other data fold to evaluate a designed criterion. Subsequently, the group number is estimated by minimizing the average criterion across all data folds. The proposed CV method has two advantages compared to existing approaches. First, the method is totally data-driven; thus no further model-specific tuning parameters are involved. Second, the method can be flexibly applied to a wide range of panel data models. Theoretically, we establish the estimation consistency by taking advantage of the optimization process on the training data fold. Experiments are carried out with a variety of synthetic datasets and panel models to further illustrate the advantages of the proposed method. Lastly, the CV method is employed to analyze the heterogeneous patterns of stock volatilities in the Chinese stock market during the 2008 financial crisis.

stat.ME↗
arXiv2026

Ultrafast optical route to coupled ferroelectric and altermagnetic switching

Exploring novel magnetoelectric coupling mechanisms to achieve control of ferroelectric polarization and magnetism is highly significant for both fundamental science and electronic device applications. Although extensive studies have been conducted on electrical switching of magnetism in multiferroic materials, simultaneous ultrafast laser switching of ferroelectric polarization and altermagnetism remains unexplored. In this letter, we propose that the ultrafast laser can be used to switch ferroelectric polarization and altermagnetism concurrently in charge-order-induced altermagnetic ferroelectrics. Building on this idea, we further demonstrate that such dual switching can be realized in charge-order-induced altermagnetic ferroelectric LiV$_2$F$_6$ by symmetry analysis and time-dependent density functional theory (TDDFT) calculation. Given that LiV$_2$F$_6$ has already been experimentally synthesized, our work not only provides an ideal material platform for experimentally realizing simultaneous switching of ferroelectric polarization and altermagnetism but also holds potential application value in future ultrafast spintronic devices.

cond-mat.mtrl-sci↗
arXiv2026

A Unified Critical Scaling Theory for Macroscopic Lightning and Quantum Avalanches: From Three-Dimensional Directed Percolation to Testable Predictions

Lightning, the most colossal discharge in nature, and flux avalanches in quantum superconductors--phenomena separated by twenty orders of magnitude in scale--display striking fractal similarity. We demonstrate that this is no mere analogy but reveals a deep physical unity. Here, we establish a universal theoretical framework that connects them. By mapping both onto the same three-dimensional reaction-diffusion-advection equation grounded in non-equilibrium statistical physics, we show they belong to the same critical universality class. We demonstrate that both systems belong to the three-dimensional Directed Percolation (3D-DP) universality class near their critical point, sharing a unified set of universal critical exponents (e.g., avalanche size distribution exponent $τ\approx 1.41$ , fractal dimension $D_f \approx 2.5$). Furthermore, by incorporating the anisotropy and turbulence coupling intrinsic to real thunderstorm environments, we predict novel effects such as anisotropic fractality of lightning channels and the systematic shift of critical exponents by turbulence. The core theoretical breakthrough lies in proposing a geometric correspondence of quantum phase information: through a rigorous mapping, the microscopic quantum phase coherence of superconductors is translated into the curvature and torsion distributions of macroscopic lightning channels, revealing a quantum statistical fingerprint emergent in classical geometry. This framework not only provides a unified paradigm for understanding dissipative structures across scales but also, via seven testable predictions, opens avenues for simulating natural lightning with laboratory quantum systems and developing novel physical early-warning methods.

cond-mat.stat-mech↗
arXiv2026

Visual-Seeker: Towards Visual-Native Multimodal Agentic Search via Active Visual Reasoning

Multimodal large language models (MLLMs) have demonstrated impressive capabilities in many visual tasks, but they often struggle with factual grounding when confronted with complex, open-world scenarios. While recent multimodal deep search agents attempt to address this issue by utilizing external tools, the visual-native search paradigm remains underexplored. Existing methods primarily rely on simple images with explicit semantics and text-only evidence trajectories, limiting the agent's ability to perform multi-hop, cross-modal reasoning and search. To address these limitations, we propose Visual-Seeker, a visual-native multimodal deep search agent via active visual reasoning. Rather than treating vision as a static input, our agent actively attends to fine-grained visual details, dynamically harvests visual evidence throughout the search process. To unlock its visual-native potential, we design an active visual reasoning data pipeline and synthesize 5K high-quality multimodal trajectories for model training. Extensive experiments demonstrate the state-of-the-art performance across five challenging multimodal search benchmarks, even surpassing several proprietary models, validating robust visual-native reasoning and search in real-world web environments. The code and data can be accessed at: https://github.com/ZhengboZhang/Visual-Seeker.

cs.AI↗
arXiv2026

MVAD: A Benchmark Dataset for Multimodal AI-Generated Video-Audio Detection

The rapid advancement of AI-generated multimodal video-audio content has raised significant concerns regarding information security and content authenticity. Existing synthetic video datasets predominantly focus on the visual modality alone, while the few incorporating audio are largely confined to facial deepfakes--a limitation that fails to address the expanding landscape of general multimodal AI-generated content and substantially impedes the development of trustworthy detection systems. To bridge this critical gap, we introduce the Multimodal Video-Audio Dataset (MVAD), the first comprehensive dataset specifically designed for detecting AI-generated multimodal video-audio content. Our dataset exhibits three key characteristics: (1) genuine multimodality with samples generated according to three realistic video-audio forgery patterns; (2) high perceptual quality achieved through diverse state-of-the-art generative models; and (3) comprehensive diversity spanning realistic and anime visual styles, four content categories (humans, animals, objects, and scenes), and four video-audio multimodal data types. Our dataset will be available at https://github.com/HuMengXue0104/MVAD.

cs.CV↗
arXiv2026

Towards stable and accurate electron dynamics via neural network based time-dependent variational Monte Carlo

Real-time dynamics of interacting electrons lies at the interface between quantum mechanics and non-equilibrium physics, governing the microscopic origin of ultrafast phenomena of molecules and nano-materials. Though neural network variational Monte Carlo has achieved unprecedented accuracy for stationary state calculations, its extension to real-time evolution remains challenging. In this work, we introduce the neural basis time-dependent variational Monte Carlo framework, which achieves stable and highly accurate simulations of electron dynamics. By constraining the time evolution to a compact, customized manifold spanned by the neural basis, we effectively bypass instability issues and achieve long-term stable evolution. Moreover, we demonstrate that this framework yields benchmark-quality accuracy in simulating the laser-driven dipole responses of the hydrogen atom and a stretched hydrogen molecule, and accurately extracts the dynamic polarizabilities of helium and beryllium atoms. Our work reveals the vast potential of neural network wavefunctions for accurately describing real-time electron dynamics and establishes a promising new route for first-principles simulations of complex, time-dependent electronic phenomena.

physics.comp-ph↗
arXiv2026

Enhancing Adversarial Robustness with Signed Distance Fields for Harmonizing Geometric Invariance and Texture

Deep neural networks demonstrate impressive performance in visual recognition but remain highly vulnerable to imperceptible adversarial attacks. Existing defense strategies such as adversarial training and diffusion-based purification have achieved significant progress but are frequently constrained by high computational cost, information loss, and inference latency. To address these challenges, we propose a Geometric and Texture balancing Purification (GeoTexPuri) framework that enhances adversarial robustness by harmonizing invariant geometric structures with textural features. Specifically, the framework integrates dense geometric guidance into the training phase by transforming discrete image masks into continuous spatial fields via Signed Distance Fields (SDF). This process establishes stable structural anchors that shield the model from local pixel noise. Through a multi-stream training objective, the model learns to internalize purified representations that effectively align semantic textural cues with these underlying geometric invariants. Extensive experiments on ImageNet demonstrate the efficacy of our approach. GeoTexPuri achieves 84.79\% clean accuracy and 83.52\% robust accuracy under the AutoAttack. Crucially, GeoTexPuri functions as a deterministic classifier during inference, requiring only the input image without any auxiliary geometric modules or additional computational costs, thereby ensuring a scalable and efficient solution for real-time applications.

cs.CV↗
arXiv2026

A Topological Soliton Model for Ball Lightning: Theory and Numerical Verification with the 3D Gross-Pitaevskii Equation

Ball lightning remains one of the most enigmatic atmospheric phenomena, characterized by its long lifetime, ability to penetrate materials, and stable spherical structure. Here we propose a novel theoretical framework interpreting ball lightning as a three-dimensional projection of a high-dimensional topological soliton. The system is described by a nonlinear Schrödinger equation with attractive interactions, stabilized by a non-zero topological charge. Through comprehensive numerical simulations of the three-dimensional Gross-Pitaevskii equation, we verify the model's core predictions: (1) long-lived stability protected by topological invariants, (2) low transmission probability due to wavefunction orthogonality, and (3) energy and size scales consistent with observational data. The soliton lifetime $τ\sim \hbar/Γ$ naturally explains the observed second-scale durations. Our work provides a self-consistent physical explanation for ball lightning while offering concrete pathways for experimental realization of three-dimensional topological solitons in Bose-Einstein condensates and nonlinear optical systems. This theoretical framework gains additional support from recent experimental breakthroughs in laboratory generation of ball-lightning-like structures.

nlin.PS↗
arXiv2026

3DThinkVLA: Endowing Vision-Language-Action Models with Latent 3D Priors via 3D-Thinking-Guided Co-training

We propose a 3D-thinking-guided co-training framework that enables vision-language-action (VLA) models to perform 3D spatial reasoning implicitly during action prediction. Our core insight is that 3D geometry perception and 3D spatial reasoning are distinct capabilities that can be disentangled and injected at different feature hierarchies. During training, three tightly coupled components work in concert primarily within the latent space: (1) To gain geometric priors, a latent 3D geometry perception module aligns intermediate visual features with a 3D foundation model, acquiring low-level geometric cues without architectural modifications to the VLM backbone. (2) Complementing this, an online 3D reasoning distillation module mitigates the prompt-induced reasoning gap via a shared reasoning anchor token. During 3D VLM co-training, this anchor is emitted as the first output token to robustly encode spatial priors. During VLA training, it serves as an input token inserted between the task and action instructions, transferring high-level spatial thinking from explicit teacher reasoning prompts to student action prompts without chain-of-thought text generation. (3) These disentangled geometric and reasoning features are then united by a spatially augmented action integration, which jointly injects them into the action-query tokens as hierarchical spatial conditions to prevent action shortcuts. At deployment, our method retains only its lightweight adapters to perform implicit 3D reasoning, discarding the 3D foundation model and the teacher branch used for supervision. Consequently, it operates purely on 2D images without 3D sensors, external models, or explicit text generation while preventing catastrophic forgetting of the pretrained VLM, achieving state-of-the-art performance on LIBERO, LIBERO-PLUS, SimplerEnv, and real-world manipulation tasks.

cs.CV↗
arXiv2026

One-Loop Tensor Power Spectrum from a Non-Canonical Spectator Field during Inflation

We compute the full one-loop corrections to the primordial tensor power spectrum in an inflationary scenario with a {non-canonical spectator field}, using the in-in formalism. We derive semi-analytic results for the scalar-sourced one-loop tensor spectrum and the effective tensor-to-scalar ratio, $r_{\mathrm{eff}}$. We consider two representative coupling functions: a localized Gaussian dip (Model G), which leads to moderate loop corrections, and a rapidly oscillatory coupling (Model O), which can yield much larger loop contributions. For Model G, we find a $\mathcal{O}(1)$ correction to $r_{\mathrm{eff}}$ while Model O can significantly enhance $r_{\mathrm{eff}}$ by several orders of magnitude (relative to the tree-level value). We further calculate the energy density of primordial gravitational waves. Assuming that primordial black holes with mass $10^{-12}M_{\odot}$ generated in this scenario, constitute all of the dark matter, we find that the results are several orders of magnitude lower than the sensitivities of Taiji/TianQin/LISA.

astro-ph.CO↗
arXiv2026

Scaling Pre-training to One Hundred Billion Data for Vision Language Models

We provide an empirical investigation of the potential of pre-training vision-language models on an unprecedented scale: 100 billion examples. We find that model performance tends to saturate at this scale on many common Western-centric classification and retrieval benchmarks, such as COCO Captions. Nevertheless, tasks of cultural diversity achieve more substantial gains from the 100-billion scale web data, thanks to its coverage of long-tail concepts. Furthermore, we analyze the model's multilinguality and show gains in low-resource languages as well. In addition, we observe that reducing the size of the pretraining dataset via quality filters like using CLIP, typically used to enhance performance, may inadvertently reduce the cultural diversity represented in large-scale datasets. Our results highlight that while traditional benchmarks may not benefit significantly from scaling noisy, raw web data to 100 billion examples, this data scale is vital for building truly inclusive multimodal systems.

cs.CV↗
arXiv2026

Gemini Embedding 2: A Native Multimodal Embedding Model from Gemini

We introduce Gemini Embedding 2, a native multimodal embedding model that allows embedding video, audio, image, and text modalities in a unified representation space. We leverage the multimodal capabilities of Gemini to produce embeddings for arbitrary combinations of interleaved inputs across all these modalities that generalize well across a wide variety of tasks. Applying large-scale contrastive learning in a multi-task multi-stage training setup, we achieve state-of-the-art performance on key embedding benchmarks including unimodal, cross-modal, and multimodal retrieval spanning a diverse set of tasks. We show that our embedding model demonstrates strong performance (with a score of 62.9 R@1 on MSCOCO, 68.8 NDCG@10 on Vatex, 69.9 on MTEB multilingual and 84.0 on MTEB Code) across a variety of tasks surpassing the performance of specialized models. These unified capabilities make Gemini Embedding 2 a promising candidate for downstream use cases such as RAG, recommendation and search. Furthermore, its robust zero-shot performance across distinct fields - from astronomy and bioscience to fine arts and the culinary arts - establishes it as a highly reliable, out-of-the-box representation even for specialized domains.

cs.CV↗
arXiv2026

When Medical Safety Alignment Fails: A Benchmark for Evaluating LLMs on High-Risk Medical Queries

Large language models (LLMs) are increasingly used for medical and health-related questions, yet their safety in high-risk medical scenarios remains poorly understood. We introduce \textsc{MedHarm}\footnote{Code and data will be released upon acceptance. Due to the sensitive nature of high-risk medical queries, data access will be available to qualified researchers upon request.}, a high-risk medical safety benchmark with 1,100 medically grounded queries across 10 safety-critical categories, including toxicology, pharmacology, covert poisoning, anesthesia, and fetal harm. Unlike broad medical QA benchmarks, \textsc{MedHarm} targets realistic clinical, educational, and technical prompts that require refusal, caution, or safe redirection rather than direct helpfulness. We evaluate 15 LLMs spanning general-purpose, medical-purpose, closed-source, and downstream SFT models, together with 4 representative guardrail models. Results reveal a substantial gap between apparent alignment and medical safety: aligned models can still produce unsafe or actionable responses, medical fine-tuning can amplify harmful specificity, and external guardrails reduce some failures while introducing brittle blocking and weak safe helpfulness. These findings show that medical safety cannot be inferred from general alignment or medical capability alone, highlighting the need for domain-specific stress testing before deploying LLMs in safety-critical medical applications.

cs.CY↗
arXiv2026

Boundary-induced helical bulk acoustic transport in LiNbO3 thin films

We experimentally demonstrate boundary-induced helical bulk states (BI-HBSs) for RF acoustic transport in LiNbO3 thin-film phononic crystals (approximately 175-200 MHz). A boundary-symmetry selection rule at an accidental Γ-point fourfold degeneracy creates interior bulk channels that couple to wide-aperture interdigital transducers without edge-aperture mismatch. Near-field vibrometry and two-port RF S-parameters confirm low-loss propagation with strongly suppressed backscattering through wavelength-scale defects. The helical band also provides slow-wave, low-dispersion delay and phase control on chip.

physics.app-ph↗
arXiv2026

YOLO-NAS-Bench: A Surrogate Benchmark with Self-Evolving Predictors for YOLO Architecture Search

Neural Architecture Search (NAS) for object detection is severely bottlenecked by high evaluation cost, as fully training each candidate YOLO architecture on COCO demands days of GPU time. Meanwhile, existing NAS benchmarks largely target image classification, leaving the detection community without a comparable benchmark for NAS evaluation. To address this gap, we introduce YOLO-NAS-Bench, the first surrogate benchmark tailored to YOLO-style detectors. YOLO-NAS-Bench defines a search space spanning channel width, block depth, and operator type across both backbone and neck, covering the core modules of YOLOv8 through YOLO12. We sample 1,000 architectures via random, stratified, and Latin Hypercube strategies, train them on COCO-mini, and build a LightGBM surrogate predictor. To sharpen the predictor in the high-performance regime most relevant to NAS, we propose a Self-Evolving Mechanism that progressively aligns the predictor's training distribution with the high-performance frontier, by using the predictor itself to discover and evaluate informative architectures in each iteration. This method grows the pool to 1,500 architectures and raises the ensemble predictor's R2 from 0.770 to 0.815 and Sparse Kendall Tau from 0.694 to 0.752, demonstrating strong predictive accuracy and ranking consistency. Using the final predictor as the fitness function for evolutionary search, we discover architectures that surpass all official YOLOv8-YOLO12 baselines at comparable latency on COCO-mini, confirming the predictor's discriminative power for top-performing detection architectures. The code is available at https://github.com/VDIGPKU/YOLO-NAS-Bench.

cs.CV↗