arXiv subjects

Yuxin Zhang

Publications and source records attributed to Yuxin Zhang.

At least 91 records · Page 5Linked to original sources

arXiv2025

GEN3D: Generating Domain-Free 3D Scenes from a Single Image

Despite recent advancements in neural 3D reconstruction, the dependence on dense multi-view captures restricts their broader applicability. Additionally, 3D scene generation is vital for advancing embodied AI and world models, which depend on diverse, high-quality scenes for learning and evaluation. In this work, we propose Gen3d, a novel method for generation of high-quality, wide-scope, and generic 3D scenes from a single image. After the initial point cloud is created by lifting the RGBD image, Gen3d maintains and expands its world model. The 3D scene is finalized through optimizing a Gaussian splatting representation. Extensive experiments on diverse datasets demonstrate the strong generalization capability and superior performance of our method in generating a world model and Synthesizing high-fidelity and consistent novel views.

cs.CV↗

arXiv2025

Semantic Alignment and Reinforcement for Data-Free Quantization of Vision Transformers

Data-free quantization (DFQ) enables model quantization without accessing real data, addressing concerns regarding data security and privacy. With the growing adoption of Vision Transformers (ViTs), DFQ for ViTs has garnered significant attention. However, existing DFQ methods exhibit two limitations: (1) semantic distortion, where the semantics of synthetic images deviate substantially from those of real images, and (2) semantic inadequacy, where synthetic images contain extensive regions with limited content and oversimplified textures, leading to suboptimal quantization performance. To address these limitations, we propose SARDFQ, a novel Semantics Alignment and Reinforcement Data-Free Quantization method for ViTs. To address semantic distortion, SARDFQ incorporates Attention Priors Alignment (APA), which optimizes synthetic images to follow randomly generated structure attention priors. To mitigate semantic inadequacy, SARDFQ introduces Multi-Semantic Reinforcement (MSR), leveraging localized patch optimization to enhance semantic richness across synthetic images. Furthermore, SARDFQ employs Soft-Label Learning (SL), wherein multiple semantic targets are adapted to facilitate the learning of multi-semantic images augmented by MSR. Extensive experiments demonstrate the effectiveness of SARDFQ, significantly surpassing existing methods. For example, SARDFQ improves top-1 accuracy on ImageNet by 15.52% for W4A4 ViT-B. The code is at https://github.com/zysxmu/SARDFQ.

cs.CV↗

arXiv2025

Enabling Near-realtime Remote Sensing via Satellite-Ground Collaboration of Large Vision-Language Models

Large vision-language models (LVLMs) have recently demonstrated great potential in remote sensing (RS) tasks (e.g., disaster monitoring) conducted by low Earth orbit (LEO) satellites. However, their deployment in real-world LEO satellite systems remains largely unexplored, hindered by limited onboard computing resources and brief satellite-ground contacts. We propose Grace, a satellite-ground collaborative system designed for near-realtime LVLM inference in RS tasks. Accordingly, we deploy compact LVLM on satellites for realtime inference, but larger ones on ground stations (GSs) to guarantee end-to-end performance. Grace is comprised of two main phases that are asynchronous satellite-GS Retrieval-Augmented Generation (RAG), and a task dispatch algorithm. Firstly, we still the knowledge archive of GS RAG to satellite archive with tailored adaptive update algorithm during limited satellite-ground data exchange period. Secondly, propose a confidence-based test algorithm that either processes the task onboard the satellite or offloads it to the GS. Extensive experiments based on real-world satellite orbital data show that Grace reduces the average latency by 76-95% compared to state-of-the-art methods, without compromising inference accuracy.

cs.NI↗

arXiv2025

Leveraging Learning Bias for Noisy Anomaly Detection

This paper addresses the challenge of fully unsupervised image anomaly detection (FUIAD), where training data may contain unlabeled anomalies. Conventional methods assume anomaly-free training data, but real-world contamination leads models to absorb anomalies as normal, degrading detection performance. To mitigate this, we propose a two-stage framework that systematically exploits inherent learning bias in models. The learning bias stems from: (1) the statistical dominance of normal samples, driving models to prioritize learning stable normal patterns over sparse anomalies, and (2) feature-space divergence, where normal data exhibit high intra-class consistency while anomalies display high diversity, leading to unstable model responses. Leveraging the learning bias, stage 1 partitions the training set into subsets, trains sub-models, and aggregates cross-model anomaly scores to filter a purified dataset. Stage 2 trains the final detector on this dataset. Experiments on the Real-IAD benchmark demonstrate superior anomaly detection and localization performance under different noise conditions. Ablation studies further validate the framework's contamination resilience, emphasizing the critical role of learning bias exploitation. The model-agnostic design ensures compatibility with diverse unsupervised backbones, offering a practical solution for real-world scenarios with imperfect training data. Code is available at https://github.com/hustzhangyuxin/LLBNAD.

cs.CV↗

arXiv2025

Spotlight Attention: Towards Efficient LLM Generation via Non-linear Hashing-based KV Cache Retrieval

Reducing the key-value (KV) cache burden in Large Language Models (LLMs) significantly accelerates inference. Dynamically selecting critical KV caches during decoding helps maintain performance. Existing methods use random linear hashing to identify important tokens, but this approach is inefficient due to the orthogonal distribution of queries and keys within two narrow cones in LLMs. We introduce Spotlight Attention, a novel method that employs non-linear hashing functions to optimize the embedding distribution of queries and keys, enhancing coding efficiency and robustness. We also developed a lightweight, stable training framework using a Bradley-Terry ranking-based loss, enabling optimization of the non-linear hashing module on GPUs with 16GB memory in 8 hours. Experimental results show that Spotlight Attention drastically improves retrieval precision while shortening the length of the hash code at least 5$\times$ compared to traditional linear hashing. Finally, we exploit the computational advantages of bitwise operations by implementing specialized CUDA kernels, achieving hashing retrieval for 512K tokens in under 100$μ$s on a single A100 GPU, with end-to-end throughput up to 3$\times$ higher than vanilla decoding.

cs.CL↗

arXiv2025

Randomized Quantum Singular Value Transformation

We introduce the first randomized algorithms for Quantum Singular Value Transformation (QSVT), a unifying framework for many quantum algorithms. Standard implementations of QSVT rely on block encodings of the Hamiltonian, which are costly to construct, requiring a logarithmic number of ancilla qubits, intricate multi-qubit control, and circuit depth scaling linearly with the number of Hamiltonian terms. In contrast, our algorithms use only a single ancilla qubit and entirely avoid block encodings. We develop two methods: (i) a direct randomization of QSVT, where block encodings are replaced by importance sampling, and (ii) an approach that integrates qDRIFT into the generalized quantum signal processing framework, with the dependence on precision exponentially improved through classical extrapolation. Both algorithms achieve gate complexity independent of the number of Hamiltonian terms, a hallmark of randomized methods, while incurring only quadratic dependence on the degree of the target polynomial. We identify natural parameter regimes where our methods outperform even standard QSVT, making them promising for early fault-tolerant quantum devices. We also establish a fundamental lower bound showing that the quadratic dependence on the polynomial degree is optimal within this framework. We apply our framework to two fundamental tasks: solving quantum linear systems and estimating ground-state properties of Hamiltonians, obtaining polynomial advantages over prior randomized algorithms. Finally, we benchmark our ground-state property estimation algorithm on electronic structure Hamiltonians and the transverse-field Ising model with long-range interactions. In both cases, our approach outperforms prior work by several orders of magnitude in circuit depth, establishing randomized QSVT as a practical and resource-efficient alternative for early fault-tolerant quantum devices.

quant-ph↗

arXiv2025

Evolaris: A Roadmap to Self-Evolving Software Intelligence Management

In recent years, the landscape of software threats has become significantly more dynamic and distributed. Security vulnerabilities are no longer discovered and shared only through formal channels such as public vulnerability databases or vendor advisories. Increasingly, criti- cal threat information emerges informally through blogs, social media, developer forums, open source repositories, and even underground com- munities. To this end, capturing such intelligence in a timely manner is essential for maintaining situational awareness and enabling prompt security responses. However, this remains a complex challenge due to the fragmented nature of data sources and the technical difficulty of collecting, parsing, mapping, and validating information at scale. To ad- dress this, we propose Evolaris, a self-evolving software intelligence sys- tem built on a multi-agent framework. Evolaris is designed to support a full-stack workflow, where agents operate independently but coordinate through shared context to perform tasks such as information discovery, reasoning, gap completion, validation, and risk detection. This archi- tecture enables the platform to learn from new inputs, refine its internal knowledge, and adapt to emerging threat patterns over time, which could continuously improve the precision, timeliness, and scalability of software threat analysis, and offers a sustainable foundation for proactive secu- rity decision-making and strengthens the broader ecosystem of security threat understanding.

cs.SE↗

arXiv2025

Pandora: A Code-Driven Large Language Model Agent for Unified Reasoning Across Diverse Structured Knowledge

Unified Structured Knowledge Reasoning (USKR) aims to answer natural language questions (NLQs) by using structured sources such as tables, databases, and knowledge graphs in a unified way. Existing USKR methods either rely on employing task-specific strategies or custom-defined representations, which struggle to leverage the knowledge transfer between different SKR tasks or align with the prior of LLMs, thereby limiting their performance. This paper proposes a novel USKR framework named \textsc{Pandora}, which takes advantage of \textsc{Python}'s \textsc{Pandas} API to construct a unified knowledge representation for alignment with LLM pre-training. It employs an LLM to generate textual reasoning steps and executable Python code for each question. Demonstrations are drawn from a memory of training examples that cover various SKR tasks, facilitating knowledge transfer. Extensive experiments on four benchmarks involving three SKR tasks demonstrate that \textsc{Pandora} outperforms existing unified frameworks and competes effectively with task-specific methods.

cs.CL↗

arXiv2025

Material Experience: An Evaluation Model for Creative Materials Based on Visual-Tactile Sensory Properties

This study adopts a design-oriented approach to integrate traditional braids with commonly used matrix materials, developing creative materials with different sensory properties by altering matrix material types and braid patterns. Based on these creative materials, a quantitative and structured model is proposed to assist designers understanding the material experience process and guide material selection by analyzing the relationship between material properties and sensory perception. Specifically, participants evaluated the creative materials under visual-tactile conditions using a 7-point semantic differential (SD) scale. Correlation analysis was performed to explore the data. The main and interaction effects of matrix materials and braid patterns on impression evaluation were analyzed using two-way analysis of variance (ANOVA). A structural equation model (SEM) was constructed based on exploratory factor analysis (EFA), and path coefficients were computed to assess the relative importance of material properties in determining material attractiveness. The results show that, compared to braids, the creative materials resulted in significant changes in impression evaluation. Furthermore, the creative materials can be understood through intrinsic, aesthetic, and physical properties, with their standardized regression coefficients for material attractiveness of 0.486, 0.650, and 0.103, respectively. These properties are interrelated and under their combined influence affect the attractiveness of the material. Therefore, designers should consider utilizing these relationships to enhance sensory experience in order to achieve design objectives. Moreover, designers should also consider balancing technology and experience, using materials according to the principle of "form follows function".

cs.HC↗

arXiv2025

Quantum singular value transformation without block encodings: Near-optimal complexity with minimal ancilla

We develop new algorithms for Quantum Singular Value Transformation (QSVT), a unifying framework that encapsulates most known quantum algorithms and serves as the foundation for new ones. Existing implementations of QSVT rely on block encoding, incurring an intrinsic $O(\log L)$ ancilla overhead and circuit depth $\widetilde{O}(L dλ)$ for polynomial transformations of a Hamiltonian $H=\sum_{k=1}^L H_k$, where $d$ is the polynomial degree and $λ=\sum_{k}\|H_k\|$. We introduce a simple yet powerful approach that utilizes only basic Hamiltonian simulation techniques, namely, Trotter methods, to: (i) eliminate the need for block encoding, (ii) reduce the ancilla overhead to only a single qubit, and (iii) still maintain near-optimal complexity. Our method achieves a circuit depth of $\widetilde{O}(L(dλ_{\mathrm{comm}})^{1+o(1)})$, without requiring any complicated multi-qubit controlled gates. Moreover, $λ_{\mathrm{comm}}$ depends on the nested commutators of the terms of $H$ and can be substantially smaller than $λ$ for many physically relevant Hamiltonians, a feature absent in standard QSVT. To achieve these results, we make use of Richardson extrapolation in a novel way, systematically eliminating errors in any interleaved sequence of arbitrary unitaries and Hamiltonian evolution operators, thereby establishing a general framework that encompasses QSVT but is more broadly applicable. As applications, we develop end-to-end quantum algorithms for solving linear systems and estimating ground state properties of Hamiltonians, both achieving near-optimal complexity without relying on oracular access. Overall, our results establish a new framework for quantum algorithms, significantly reducing hardware overhead while maintaining near-optimal performance, with implications for both near-term and fault-tolerant quantum computing.

quant-ph↗

arXiv2025

Step-Audio 2 Technical Report

This paper presents Step-Audio 2, an end-to-end multi-modal large language model designed for industry-strength audio understanding and speech conversation. By integrating a latent audio encoder and reasoning-centric reinforcement learning (RL), Step-Audio 2 achieves promising performance in automatic speech recognition (ASR) and audio understanding. To facilitate genuine end-to-end speech conversation, Step-Audio 2 incorporates the generation of discrete audio tokens into language modeling, significantly enhancing its responsiveness to paralinguistic information such as speaking styles and emotions. To effectively leverage the rich textual and acoustic knowledge in real-world data, Step-Audio 2 integrates retrieval-augmented generation (RAG) and is able to call external tools such as web search to mitigate hallucination and audio search to switch timbres. Trained on millions of hours of speech and audio data, Step-Audio 2 delivers intelligence and expressiveness across diverse conversational scenarios. Evaluation results demonstrate that Step-Audio 2 achieves state-of-the-art performance on various audio understanding and conversational benchmarks compared to other open-source and commercial solutions. Please visit https://github.com/stepfun-ai/Step-Audio2 for more information.

cs.CL↗

arXiv2025

LumiSculpt: Enabling Consistent Portrait Lighting in Video Generation

Lighting plays a pivotal role in ensuring the naturalness and aesthetic quality of video generation. However, the impact of lighting is deeply coupled with other factors of videos, e.g., objects and scenes. Thus, it remains challenging to disentangle and model coherent lighting conditions independently, limiting the flexibility to control lighting in video generation. In this paper, inspired by the established controllable T2I models, we propose LumiSculpt, which enables precise and consistent lighting control in T2V generation models. LumiSculpt equips the video generation with new interactive capabilities, allowing the input of reference image sequences with customized lighting conditions. Furthermore, the core learnable plug-and-play module of LumiSculpt facilitates direct control over the intensity, position and trajectory of an assumed light source in video diffusion models. To effectively train LumiSculpt and address the issue of insufficient lighting data, we construct LumiHuman, a new lightweight and flexible dataset for portrait lighting of images and videos. Experimental results demonstrate that LumiSculpt achieves precise and high-quality lighting control in video generation. The analysis demonstrates the flexibility of LumiHuman.

cs.CV↗

arXiv2025

SL-ACC: A Communication-Efficient Split Learning Framework with Adaptive Channel-wise Compression

The increasing complexity of neural networks poses a significant barrier to the deployment of distributed machine learning (ML) on resource-constrained devices, such as federated learning (FL). Split learning (SL) offers a promising solution by offloading the primary computing load from edge devices to a server via model partitioning. However, as the number of participating devices increases, the transmission of excessive smashed data (i.e., activations and gradients) becomes a major bottleneck for SL, slowing down the model training. To tackle this challenge, we propose a communication-efficient SL framework, named SL-ACC, which comprises two key components: adaptive channel importance identification (ACII) and channel grouping compression (CGC). ACII first identifies the contribution of each channel in the smashed data to model training using Shannon entropy. Following this, CGC groups the channels based on their entropy and performs group-wise adaptive compression to shrink the transmission volume without compromising training accuracy. Extensive experiments across various datasets validate that our proposed SL-ACC framework takes considerably less time to achieve a target accuracy than state-of-the-art benchmarks.

cs.LG↗

arXiv2025

DS$^2$Net: Detail-Semantic Deep Supervision Network for Medical Image Segmentation

Deep Supervision Networks exhibit significant efficacy for the medical imaging community. Nevertheless, existing work merely supervises either the coarse-grained semantic features or fine-grained detailed features in isolation, which compromises the fact that these two types of features hold vital relationships in medical image analysis. We advocate the powers of complementary feature supervision for medical image segmentation, by proposing a Detail-Semantic Deep Supervision Network (DS$^2$Net). DS$^2$Net navigates both low-level detailed and high-level semantic feature supervision through Detail Enhance Module (DEM) and Semantic Enhance Module (SEM). DEM and SEM respectively harness low-level and high-level feature maps to create detail and semantic masks for enhancing feature supervision. This is a novel shift from single-view deep supervision to multi-view deep supervision. DS$^2$Net is also equipped with a novel uncertainty-based supervision loss that adaptively assigns the supervision strength of features within distinct scales based on their uncertainty, thus circumventing the sub-optimal heuristic design that typifies previous works. Through extensive experiments on six benchmarks captured under either colonoscopy, ultrasound and microscope, we demonstrate that DS$^2$Net consistently outperforms state-of-the-art methods for medical image analysis.

cs.CV↗

arXiv2025

Multi-Prompt Style Interpolation for Fine-Grained Artistic Control

Text-driven image style transfer has seen remarkable progress with methods leveraging cross-modal embeddings for fast, high-quality stylization. However, most existing pipelines assume a \emph{single} textual style prompt, limiting the range of artistic control and expressiveness. In this paper, we propose a novel \emph{multi-prompt style interpolation} framework that extends the recently introduced \textbf{StyleMamba} approach. Our method supports blending or interpolating among multiple textual prompts (eg, ``cubism,'' ``impressionism,'' and ``cartoon''), allowing the creation of nuanced or hybrid artistic styles within a \emph{single} image. We introduce a \textit{Multi-Prompt Embedding Mixer} combined with \textit{Adaptive Blending Weights} to enable fine-grained control over the spatial and semantic influence of each style. Further, we propose a \emph{Hierarchical Masked Directional Loss} to refine region-specific style consistency. Experiments and user studies confirm our approach outperforms single-prompt baselines and naive linear combinations of styles, achieving superior style fidelity, text-image alignment, and artistic flexibility, all while maintaining the computational efficiency offered by the state-space formulation.

cs.GR↗

arXiv2025

Multi-View Reconstruction with Global Context for 3D Anomaly Detection

3D anomaly detection is critical in industrial quality inspection. While existing methods achieve notable progress, their performance degrades in high-precision 3D anomaly detection due to insufficient global information. To address this, we propose Multi-View Reconstruction (MVR), a method that losslessly converts high-resolution point clouds into multi-view images and employs a reconstruction-based anomaly detection framework to enhance global information learning. Extensive experiments demonstrate the effectiveness of MVR, achieving 89.6\% object-wise AU-ROC and 95.7\% point-wise AU-ROC on the Real3D-AD benchmark.

cs.CV↗

arXiv2025

GOAT-SLM: A Spoken Language Model with Paralinguistic and Speaker Characteristic Awareness

Recent advances in end-to-end spoken language models (SLMs) have significantly improved the ability of AI systems to engage in natural spoken interactions. However, most existing models treat speech merely as a vehicle for linguistic content, often overlooking the rich paralinguistic and speaker characteristic cues embedded in human speech, such as dialect, age, emotion, and non-speech vocalizations. In this work, we introduce GOAT-SLM, a novel spoken language model with paralinguistic and speaker characteristic awareness, designed to extend spoken language modeling beyond text semantics. GOAT-SLM adopts a dual-modality head architecture that decouples linguistic modeling from acoustic realization, enabling robust language understanding while supporting expressive and adaptive speech generation. To enhance model efficiency and versatility, we propose a modular, staged training strategy that progressively aligns linguistic, paralinguistic, and speaker characteristic information using large-scale speech-text corpora. Experimental results on TELEVAL, a multi-dimensional evaluation benchmark, demonstrate that GOAT-SLM achieves well-balanced performance across both semantic and non-semantic tasks, and outperforms existing open-source models in handling emotion, dialectal variation, and age-sensitive interactions. This work highlights the importance of modeling beyond linguistic content and advances the development of more natural, adaptive, and socially aware spoken language systems.

cs.CL↗

arXiv2025

GS-Bias: Global-Spatial Bias Learner for Single-Image Test-Time Adaptation of Vision-Language Models

Recent advances in test-time adaptation (TTA) for Vision-Language Models (VLMs) have garnered increasing attention, particularly through the use of multiple augmented views of a single image to boost zero-shot generalization. Unfortunately, existing methods fail to strike a satisfactory balance between performance and efficiency, either due to excessive overhead of tuning text prompts or unstable benefits from handcrafted, training-free visual feature enhancement. In this paper, we present Global-Spatial Bias Learner (GS-Bias), an efficient and effective TTA paradigm that incorporates two learnable biases during TTA, unfolded as the global bias and spatial bias. Particularly, the global bias captures the global semantic features of a test image by learning consistency across augmented views, while spatial bias learns the semantic coherence between regions in the image's spatial visual representation. It is worth highlighting that these two sets of biases are directly added to the logits outputed by the pretrained VLMs, which circumvent the full backpropagation through VLM that hinders the efficiency of existing TTA methods. This endows GS-Bias with extremely high efficiency while achieving state-of-the-art performance on 15 benchmark datasets. For example, it achieves a 2.23% improvement over TPT in cross-dataset generalization and a 2.72% improvement in domain generalization, while requiring only 6.5% of TPT's memory usage on ImageNet.

cs.CV↗