arXiv subjects

Qian Zhang

Publications and source records attributed to Qian Zhang.

At least 109 records · Page 6Linked to original sources

arXiv2026

DINO-Tok: Adapting DINO for Visual Tokenizers

Recent advances in visual generation have emphasized the importance of Latent Generative Models (LGMs), which critically depend on effective visual tokenizers to bridge pixels and semantic representations. However, tokenizers constructed on pre-trained vision foundation models (VFMs) often struggle to balance semantic richness and reconstruction fidelity in high-dimensional latent spaces. In this paper, we introduce DINO-Tok, a visual tokenizer built upon a frozen DINO encoder that supports both continuous autoencoding (DINO-Tok-AE) and discrete vector-quantization (DINO-Tok-VQ). By unifying hierarchical representations from both shallow fine-grained features and deep global semantics into an information-complete latent space, DINO-Tok preserves texture details while maintaining \textit{semantic consistency} for generation. We further investigate VQ in frozen semantic feature spaces of high dimensionality, where information dilution and codebook collapse frequently arise. To address this issue, we propose Dominant-Subspace Quantization (DSQ), which leverages a global PCA analysis to select principal components while suppressing noisy dimensions, thereby stabilizing codebook optimization and improving reconstruction and generation quality. On ImageNet 256x256, DINO-Tok achieves strong reconstruction performance, achieving 0.28 rFID for continuous autoencoding and 1.10 rFID for discrete VQ, as well as strong few-step generation performance 1.82 gFID for diffusion and 2.44 gFID for autoregressive generation. These results demonstrate that pre-trained VFMs such as DINO can be directly adapted into high-fidelity, semantically aligned visual tokenizers for next-generation latent generative models. Code will be publicly available at https://github.com/MKJia/DINO-Tok.

cs.CV↗

arXiv2026

ClearAIR: A Human-Visual-Perception-Inspired All-in-One Image Restoration

All-in-One Image Restoration (AiOIR) has advanced significantly, offering promising solutions for complex real-world degradations. However, most existing approaches rely heavily on degradation-specific representations, often resulting in oversmoothing and artifacts. To address this, we propose ClearAIR, a novel AiOIR framework inspired by Human Visual Perception (HVP) and designed with a hierarchical, coarse-to-fine restoration strategy. First, leveraging the global priority of early HVP, we employ a Multimodal Large Language Model (MLLM)-based Image Quality Assessment (IQA) model for overall evaluation. Unlike conventional IQA, our method integrates cross-modal understanding to more accurately characterize complex, composite degradations. Building upon this overall assessment, we then introduce a region awareness and task recognition pipeline. A semantic cross-attention, leveraging semantic guidance unit, first produces coarse semantic prompts. Guided by this regional context, a degradation-aware module implicitly captures region-specific degradation characteristics, enabling more precise local restoration. Finally, to recover fine details, we propose an internal clue reuse mechanism. It operates in a self-supervised manner to mine and leverage the intrinsic information of the image itself, substantially enhancing detail restoration. Experimental results show that ClearAIR achieves superior performance across diverse synthetic and real-world datasets.

cs.CV↗

arXiv2026

GravCal: Single-Image Calibration of IMU Gravity Priors with Per-Sample Confidence

Gravity estimation is fundamental to visual-inertial perception, augmented reality, and robotics, yet gravity priors from IMUs are often unreliable under linear acceleration, vibration, and transient motion. Existing methods often estimate gravity directly from images or assume reasonably accurate inertial input, leaving the practical problem of correcting a noisy gravity prior from a single image largely unaddressed. We present GravCal, a feedforward model for single-image gravity prior calibration. Given one RGB image and a noisy gravity prior, GravCal predicts a corrected gravity direction and a per-sample confidence score. The model combines two complementary predictions, including a residual correction of the input prior and a prior-independent image estimate, and uses a learned gate to fuse them adaptively. Extensive experiments show strong gains over raw inertial priors: GravCal reduces mean angular error from 22.02° (IMU prior) to 14.24°, with larger improvements when the prior is severely corrupted. We also introduce a novel dataset of over 148K frames with paired VIO-derived ground-truth gravity and Mahony-filter IMU priors across diverse scenes and arbitrary camera orientations. The learned gate also correlates with prior quality, making it a useful confidence signal for downstream systems.

cs.CV↗

arXiv2026

Robust Beamforming for Practical RIS-Aided RSMA Systems with Imperfect SIC under Transceiver Hardware Impairments

Reconfigurable intelligent surface (RIS)-aided rate-splitting multiple access (RSMA) systems have demonstrated remarkable potential in enhancing spectral efficiency. However, most existing works rely on ideal hardware, which is unrealistic.In practical deployments, RIS elements suffer from amplitude-phase coupling, where transceivers are subject to hardware impairments (HWI), and successive interference cancellation (SIC) in RSMA networks cannot achieve perfect interference elimination for decoded signals.To address these limitations, we investigate a robust beamforming design for RIS-aided RSMA systems under practical hardware imperfections. We first characterize the asymptotic signal-to-noise ratio (SNR) of practical RIS systems when the beamformer is designed based on ideal RIS model, thereby theoretically quantifying the resulting performance degradation. We then derive a closed-form expression for the distortion noise power induced by transceiver HWI, while also accounting for residual interference due to imperfect SIC. Building on these insights, we establish a comprehensive system model that jointly incorporates all hardware-induced impairments and formulate a multiuser sum rate maximization problem. To solve the resulting non-convex optimization problem, we develop an efficient block variable relaxation algorithm. Simulation results verify that the proposed scheme significantly outperforms conventional non-orthogonal multiple access (NOMA) approaches, and achieves superior robustness compared with benchmark schemes neglecting HWI, imperfect SIC, or amplitude-phase coupling.

cs.IT↗

arXiv2026

OccTENS: 3D Occupancy World Model via Temporal Next-Scale Prediction

In this paper, we propose OccTENS, a generative occupancy world model that enables controllable, high-fidelity long-term occupancy generation while maintaining computational efficiency. Different from visual generation, the occupancy world model must capture the fine-grained 3D geometry and dynamic evolution of the 3D scenes, posing great challenges for the generative models. Recent approaches based on autoregression (AR) have demonstrated the potential to predict vehicle movement and future occupancy scenes simultaneously from historical observations, but they typically suffer from \textbf{inefficiency}, \textbf{temporal degradation} in long-term generation and \textbf{lack of controllability}. To holistically address these issues, we reformulate the occupancy world model as a temporal next-scale prediction (TENS) task, which decomposes the temporal sequence modeling problem into the modeling of spatial scale-by-scale generation and temporal scene-by-scene prediction. With a \textbf{TensFormer}, OccTENS can effectively manage the temporal causality and spatial relationships of occupancy sequences in a flexible and scalable way. To enhance the pose controllability, we further propose a holistic pose aggregation strategy, which features a unified sequence modeling for occupancy and ego-motion. Experiments show that OccTENS outperforms the state-of-the-art method with both higher occupancy quality and faster inference time.

cs.CV↗

arXiv2026

Joint Precoding and Phase-Shift Optimization for Beyond-Diagonal RIS-Aided ISAC System

Beyond diagonal reconfigurable intelligent surfaces (BD-RIS) can realize the interconnection between reflecting elements through the impedance network, thereby providing a new approach for the performance improvement of integrated sensing and communication (ISAC) systems. This paper investigates the optimization problem of BD-RIS-aided multiuser ISAC system, aiming to achieve the flexible design of trade-offs between communication and sensing performance. Specifically, we propose an optimization framework jointly combining the multiuser interference management and sensing beam gain approximation method. By jointly optimizing the precoding vector and RIS phase-shift matrix, improving the multiuser communication sum rate through the proposed interference management method, and enhancing the system sensing performance through the beam gain approximation method. For the resulting non-convex weighted optimization problem, we employ the alternating optimization (AO) algorithm to decouple it into two subproblems of precoding vector and phase-shift matrix optimization, with each step admitting closed-form solutions.Simulation results demonstrate that the proposed BD-RIS-aided ISAC system can achieve significant improvement in the trade-offs between communication and sensing performance than the traditional diagonal RIS, verifying the effectiveness of the proposed optimization framework.

cs.IT↗

arXiv2026

LongStream: Long-Sequence Streaming Autoregressive Visual Geometry

Long-sequence streaming 3D reconstruction remains a significant open challenge. Existing autoregressive models often fail when processing long sequences because they anchor poses to the first frame, leading to attention decay, scale drift, and extrapolation errors. We introduce LongStream, a novel gauge-decoupled streaming visual geometry model for metric-scale scene reconstruction across thousands of frames under a strictly online, future-invisible setting. Our approach is threefold. First, we discard the first-frame anchor and predict keyframe-relative poses. This reformulates long-range extrapolation into a constant-difficulty local task. Second, we introduce orthogonal scale learning. This method fully disentangles geometry from scale estimation to suppress drift. Finally, we identify attention bias issues in Transformers, including attention-sink reliance and long-term KV-cache saturation. We propose cache-consistent training combined with periodic cache refresh. This approach suppresses attention biases and contamination over ultra-long sequences and reduces the gap between training and inference. Experiments show that LongStream achieves state-of-the-art performance, enabling stable, metric-scale reconstruction over kilometer-scale sequences at 18 FPS. Project Page: https://3dagentworld.github.io/longstream/

cs.CV↗

arXiv2026

Composing Driving Worlds through Disentangled Control for Adversarial Scenario Generation

A major challenge in autonomous driving is the "long tail" of safety-critical edge cases, which often emerge from unusual combinations of common traffic elements. Synthesizing these scenarios is crucial, yet current controllable generative models provide incomplete or entangled guidance, preventing the independent manipulation of scene structure, object identity, and ego actions. We introduce CompoSIA, a compositional driving video simulator that disentangles these traffic factors, enabling fine-grained control over diverse adversarial driving scenarios. To support controllable identity replacement of scene elements, we propose a noise-level identity injection, allowing pose-agnostic identity generation across diverse element poses, all from a single reference image. Furthermore, a hierarchical dual-branch action control mechanism is introduced to improve action controllability. Such disentangled control enables adversarial scenario synthesis-systematically combining safe elements into dangerous configurations that entangled generators cannot produce. Extensive comparisons demonstrate superior controllable generation quality over state-of-the-art baselines, with a 17% improvement in FVD for identity editing and reductions of 30% and 47% in rotation and translation errors for action control. Furthermore, downstream stress-testing reveals substantial planner failures: across editing modalities, the average collision rate of 3s increases by 173%.

cs.CV↗

arXiv2026

Senna-2: Aligning VLM and End-to-End Driving Policy for Consistent Decision Making and Planning

Vision-language models (VLMs) enhance the planning capability of end-to-end (E2E) driving policy by leveraging high-level semantic reasoning. However, existing approaches often overlook the dual-system consistency between VLM's high-level decision and E2E's low-level planning. As a result, the generated trajectories may misalign with the intended driving decisions, leading to weakened top-down guidance and decision-following ability of the system. To address this issue, we propose Senna-2, an advanced VLM-E2E driving policy that explicitly aligns the two systems for consistent decision-making and planning. Our method follows a consistency-oriented three-stage training paradigm. In the first stage, we conduct driving pre-training to achieve preliminary decision-making and planning, with a decision adapter transmitting VLM decisions to E2E policy in the form of implicit embeddings. In the second stage, we align the VLM and the E2E policy in an open-loop setting. In the third stage, we perform closed-loop alignment via bottom-up Hierarchical Reinforcement Learning in 3DGS environments to reinforce the safety and efficiency. Extensive experiments demonstrate that Senna-2 achieves superior dual-system consistency (19.3% F1 score improvement) and significantly enhances driving safety in both open-loop (5.7% FDE reduction) and closed-loop settings (30.6% AF-CR reduction).

cs.CV↗

arXiv2026

Beamforming Optimization for Extremely Large-Scale RIS-Aided Near-Field Secure Communications

This paper studies an extremely large-scale reconfigurable intelligent surface (XL-RIS)-aided near-field physical layer security (PLS) communication system, aiming to maximize the secrecy rate by jointly optimizing precoding vector at the BS and the reflection coefficient matrix at the XL-RIS. Artifi-cial jamming was introduced to further enhance communication security. To solve the non-convex secrecy rate problem, an alternate optimization-based algorithm is adopted to decompose it into two sub-problems. Specifically, when optimizing the transmit beamformer at the BS, the non-convex prob-lem is transformed into a convex one through the weighted minimum mean-square error and the successive convex approximation-based algorithms. For the optimization problem of the XL-RIS phase-shifting matrix, a low-complexity alternating direction method of multipliers-based algorithm is employed to enhance the flexibility of the design. The proposed algorithm is capable of accommodating discrete phase optimization for the XL-RIS, thus better aligning with practical system requirements. Simulation results demonstrate that when the eavesdropper reside in the same direction as the legitimate user and is located closer to the XL-RIS, the proposed scheme in this paper can still ensure the secure communication.

cs.IT↗

arXiv2026

ForesightSafety Bench: A Frontier Risk Evaluation and Governance Framework towards Safe AI

Rapidly evolving AI exhibits increasingly strong autonomy and goal-directed capabilities, accompanied by derivative systemic risks that are more unpredictable, difficult to control, and potentially irreversible. However, current AI safety evaluation systems suffer from critical limitations such as restricted risk dimensions and failed frontier risk detection. The lagging safety benchmarks and alignment technologies can hardly address the complex challenges posed by cutting-edge AI models. To bridge this gap, we propose the "ForesightSafety Bench" AI Safety Evaluation Framework, beginning with 7 major Fundamental Safety pillars and progressively extends to advanced Embodied AI Safety, AI4Science Safety, Social and Environmental AI risks, Catastrophic and Existential Risks, as well as 8 critical industrial safety domains, forming a total of 94 refined risk dimensions. To date, the benchmark has accumulated tens of thousands of structured risk data points and assessment results, establishing a widely encompassing, hierarchically clear, and dynamically evolving AI safety evaluation framework. Based on this benchmark, we conduct systematic evaluation and in-depth analysis of over twenty mainstream advanced large models, identifying key risk patterns and their capability boundaries. The safety capability evaluation results reveals the widespread safety vulnerabilities of frontier AI across multiple pillars, particularly focusing on Risky Agentic Autonomy, AI4Science Safety, Embodied AI Safety, Social AI Safety and Catastrophic and Existential Risks. Our benchmark is released at https://github.com/Beijing-AISI/ForesightSafety-Bench. The project website is available at https://foresightsafety-bench.beijing-aisi.ac.cn/.

cs.AI↗

arXiv2026

Secure Transmission for Fluid Antenna-Aided ISAC Systems

Fluid antenna (FA) has become a highly promising technology and has recently been used to enhance the integrated sensing and communication (ISAC) system. However, the scenario where sensing targets act as eavesdroppers in ISAC and how to maximize the sum secrecy rate has not been addressed. This letter investigates secure transmission in FA-aided ISAC systems, where the spatial agility of FAs enables enhanced physical layer security. We jointly optimize antenna position vector (APV) and beamforming to maximize the multiuser sum secrecy rate, which complicates the solution process. To solve the resulting non-convex problem, we use a block successive upper-bound minimization (BSUM) algorithm, which incorporates the proximal distance algorithm (PDA) for closed-form beamformer updates and extrapolated projected gradient (EPG) for APV optimization. Simulation results show that the proposed FA-ISAC scheme achieves over 20$\%$ sum secrecy rate gain compared to fixed-position antenna (FPA) systems.

cs.IT↗

arXiv2026

AI-Driven Structure Refinement of X-ray Diffraction

Artificial intelligence can rapidly propose candidate phases and structures from X-ray diffraction (XRD), but these hypotheses often fail in downstream refinement because peak intensities cannot be stably assigned under severe overlap and diffraction consistency is enforced only weakly. Here we introduce the whole-pattern expectation--maximization (WPEM) algorithm, a physics-constrained whole-pattern decomposition and refinement workflow that turns Bragg's law into an explicit constraint within a batch expectation--maximization framework. WPEM models the full profile as a probabilistic mixture density and iteratively infers component-resolved intensities while keeping peak centres Bragg-consistent, producing a continuous, physically admissible intensity representation that remains stable in heavily overlapped regions and in the presence of mixed radiation or multiple phases. We benchmark WPEM on standard reference patterns (PbSO$_4$ and Tb$_2$BaCoO$_5$), where it yields lower $R_p/R_{wp}$ than widely used packages (FullProf and TOPAS) under matched refinement conditions. We further demonstrate generality across realistic experimental scenarios, including phase-resolved decomposition in multiphase materials, quantitative recovery of mixture compositions, separation of crystalline peaks from amorphous backgrounds in semicrystalline systems, high-throughput operando lattice tracking, automated refinement of compositionally disordered solid solutions, and quantitative phase-resolved analysis of complex archaeological samples from synchrotron powder XRD. By providing Bragg-consistent, uncertainty-aware intensity partitioning as a refinement-ready interface, WPEM closes the gap between AI-generated hypotheses and diffraction-admissible structure refinement on challenging XRD data.

cond-mat.mtrl-sci↗

arXiv2026

Optimality Analysis of RSMA Degenerating to SDMA Under Imperfect SIC

This document serves as supplementary material for our journal submission, providing detailed mathematical proofs and derivations that support the results presented in the main manuscript. Specifically, under a modeling framework that jointly considers transceiver hardware impairments and imperfect successive interference cancellation (SIC), we systematically derive and prove from an optimality perspective that: when the residual interference coefficient approaches 1 (i.e., SIC becomes severely ineffective), there exists an optimal solution such that the common stream beamformer satisfies $\bm w_c^\star=\bm 0$, and hence the optimal rate-splitting multiple access (RSMA) transmission structure degenerates into space division multiple access (SDMA). This conclusion provides a verifiable theoretical justification for the convergence phenomenon observed in simulations, namely that "the RSMA performance gradually approaches that of SDMA as SIC degrades", and can also serve as a reference for multiple-access selection and system design in SIC-limited scenarios.

cs.IT↗

arXiv2026

Learning Proposes, Geometry Disposes: A Modular Framework for Efficient Spatial Reasoning

Spatial perception aims to estimate camera motion and scene structure from visual observations, a problem traditionally addressed through geometric modeling and physical consistency constraints. Recent learning-based methods have demonstrated strong representational capacity for geometric perception and are increasingly used to augment classical geometry-centric systems in practice. However, whether learning components should directly replace geometric estimation or instead serve as intermediate modules within such pipelines remains an open question. In this work, we address this gap and investigate an end-to-end modular framework for effective spatial reasoning, where learning proposes geometric hypotheses, while geometric algorithms dispose estimation decisions. In particular, we study this principle in the context of relative camera pose estimation on RGB-D sequences. Using VGGT as a representative learning model, we evaluate learning-based pose and depth proposals under varying motion magnitudes and scene dynamics, followed by a classical point-to-plane RGB-D ICP as the geometric backend. Our experiments on the TUM RGB-D benchmark reveal three consistent findings: (1) learning-based pose proposals alone are unreliable; (2) learning-proposed geometry, when improperly aligned with camera intrinsics, can degrade performance; and (3) when learning-proposed depth is geometrically aligned and followed by a geometric disposal stage, consistent improvements emerge in moderately challenging rigid settings. These results demonstrate that geometry is not merely a refinement component, but an essential arbiter that validates and absorbs learning-based geometric observations. Our study highlights the importance of modular, geometry-aware system design for robust spatial perception.

cs.CV↗

arXiv2026

Towards Reliable Evaluation of Adversarial Robustness for Spiking Neural Networks

Spiking Neural Networks (SNNs) utilize spike-based activations to mimic the brain's energy-efficient information processing. However, the binary and discontinuous nature of spike activations causes vanishing gradients, making adversarial robustness evaluation via gradient descent unreliable. While improved surrogate gradient methods have been proposed, their effectiveness under strong adversarial attacks remains unclear. We propose a more reliable framework for evaluating SNN adversarial robustness. We theoretically analyze the degree of gradient vanishing in surrogate gradients and introduce the Adaptive Sharpness Surrogate Gradient (ASSG), which adaptively evolves the shape of the surrogate function according to the input distribution during attack iterations, thereby enhancing gradient accuracy while mitigating gradient vanishing. In addition, we design an adversarial attack with adaptive step size under the $L_\infty$ constraint-Stable Adaptive Projected Gradient Descent (SA-PGD), achieving faster and more stable convergence under imprecise gradients. Extensive experiments show that our approach substantially increases attack success rates across diverse adversarial training schemes, SNN architectures and neuron models, providing a more generalized and reliable evaluation of SNN adversarial robustness. The experimental results further reveal that the robustness of current SNNs has been significantly overestimated and highlighting the need for more dependable adversarial training methods. The code is released at https://github.com/craree/ASSG-SNNs-Robustness-Evaluation

cs.LG↗

arXiv2026

Entropy-Guided k-Guard Sampling for Long-Horizon Autoregressive Video Generation

Autoregressive (AR) architectures have achieved significant successes in LLMs, inspiring explorations for video generation. In LLMs, top-p/top-k sampling strategies work exceptionally well: language tokens have high semantic density and low redundancy, so a fixed size of token candidates already strikes a balance between semantic accuracy and generation diversity. In contrast, video tokens have low semantic density and high spatio-temporal redundancy. This mismatch makes static top-k/top-p strategies ineffective for video decoders: they either introduce unnecessary randomness for low-uncertainty regions (static backgrounds) or get stuck in early errors for high-uncertainty regions (foreground objects). Prediction errors will accumulate as more frames are generated and eventually severely degrade long-horizon quality. To address this, we propose Entropy-Guided k-Guard (ENkG) sampling, a simple yet effective strategy that adapts sampling to token-wise dispersion, quantified by the entropy of each token's predicted distribution. ENkG uses adaptive token candidate sizes: for low-entropy regions, it employs fewer candidates to suppress redundant noise and preserve structural integrity; for high-entropy regions, it uses more candidates to mitigate error compounding. ENkG is model-agnostic, training-free, and adds negligible overhead. Experiments demonstrate consistent improvements in perceptual quality and structural stability compared to static top-k/top-p strategies.

cs.CV↗

arXiv2026

FireFly-S: Exploiting Dual-Side Sparsity for Spiking Neural Networks Acceleration with Reconfigurable Spatial Architecture

Spiking Neural Networks (SNNs), with brain-inspired structure using discrete spikes instead of continuous activations, are gaining attention for their efficient processing on neuromorphic chips. While current SNN hardware accelerators often prioritize temporal spike sparsity, exploiting sparse synaptic weights offers significant untapped potential for even greater efficiency. To address this, we propose FireFly-S, a Sparse extension of the FireFly series. This co-optimized software-hardware design focuses on leveraging dual-side sparsity for acceleration. On the software side, we propose a algorithmic optimization framework that combines gradient rewiring for pruning and modified Learned Step Size Quantization (LSQ) for SNNs, achieving a weight sparsity exceeding 85\% and enabling efficient 4-bit quantization with negligible accuracy loss. On the hardware side, we present an efficient dual-side sparsity detector employing a Bitmap-based sparse decoding logic to pinpoint the positions of non-zero weights and input spikes. The logic allows for direct bypassing of redundant computations, thereby enhancing computational efficiency. Different from the overlay architecture adopted by previous FireFly series, we adopt a parametric spatial architecture with inter-layer pipelining that can fully exploit the fine-grained programmability and reconfigurability of Field-Programmable Gate Arrays (FPGAs), enabling fast deployment for various models. A spatial-temporal dataflow is also proposed to support such inter-layer pipelining and avoid long-term temporal dependencies. In experiments conducted on the MNIST, DVS-Gesture and CIFAR-10 datasets, the FireFly-S model achieves 85--95\% sparsity with 4-bit quantization and the hardware accelerator effectively leverages the dual-side sparsity, delivering performance metrics of 10,047~FPS/W on MNIST, 3,683~FPS/W on DVS-Gesture, and 2,327~FPS/W on CIFAR-10.

cs.AR↗