arXiv subjects

Jie Zhou

Publications and source records attributed to Jie Zhou.

At least 217 records · Page 12Linked to original sources

arXiv2025

Terra: Explorable Native 3D World Model with Point Latents

World models have garnered increasing attention for comprehensive modeling of the real world. However, most existing methods still rely on pixel-aligned representations as the basis for world evolution, neglecting the inherent 3D nature of the physical world. This could undermine the 3D consistency and diminish the modeling efficiency of world models. In this paper, we present Terra, a native 3D world model that represents and generates explorable environments in an intrinsic 3D latent space. Specifically, we propose a novel point-to-Gaussian variational autoencoder (P2G-VAE) that encodes 3D inputs into a latent point representation, which is subsequently decoded as 3D Gaussian primitives to jointly model geometry and appearance. We then introduce a sparse point flow matching network (SPFlow) for generating the latent point representation, which simultaneously denoises the positions and features of the point latents. Our Terra enables exact multi-view consistency with native 3D representation and architecture, and supports flexible rendering from any viewpoint with only a single generation process. Furthermore, Terra achieves explorable world modeling through progressive generation in the point latent space. We conduct extensive experiments on the challenging indoor scenes from ScanNet v2. Terra achieves state-of-the-art performance in both reconstruction and generation with high 3D consistency.

cs.CV↗

arXiv2025

CurriFlow: Curriculum-Guided Depth Fusion with Optical Flow-Based Temporal Alignment for 3D Semantic Scene Completion

Semantic Scene Completion (SSC) aims to infer complete 3D geometry and semantics from monocular images, serving as a crucial capability for camera-based perception in autonomous driving. However, existing SSC methods relying on temporal stacking or depth projection often lack explicit motion reasoning and struggle with occlusions and noisy depth supervision. We propose CurriFlow, a novel semantic occupancy prediction framework that integrates optical flow-based temporal alignment with curriculum-guided depth fusion. CurriFlow employs a multi-level fusion strategy to align segmentation, visual, and depth features across frames using pre-trained optical flow, thereby improving temporal consistency and dynamic object understanding. To enhance geometric robustness, a curriculum learning mechanism progressively transitions from sparse yet accurate LiDAR depth to dense but noisy stereo depth during training, ensuring stable optimization and seamless adaptation to real-world deployment. Furthermore, semantic priors from the Segment Anything Model (SAM) provide category-agnostic supervision, strengthening voxel-level semantic learning and spatial consistency. Experiments on the SemanticKITTI benchmark demonstrate that CurriFlow achieves state-of-the-art performance with a mean IoU of 16.9, validating the effectiveness of our motion-guided and curriculum-aware design for camera-based 3D semantic scene completion.

cs.CV↗

arXiv2025

DualResearch: Entropy-Gated Dual-Graph Retrieval for Answer Reconstruction

The deep-research framework orchestrates external tools to perform complex, multi-step scientific reasoning that exceeds the native limits of a single large language model. However, it still suffers from context pollution, weak evidentiary support, and brittle execution paths. To address these issues, we propose DualResearch, a retrieval and fusion framework that matches the epistemic structure of tool-intensive reasoning by jointly modeling two complementary graphs: a breadth semantic graph that encodes stable background knowledge, and a depth causal graph that captures execution provenance. Each graph has a layer-native relevance function, seed-anchored semantic diffusion for breadth, and causal-semantic path matching with reliability weighting for depth. To reconcile their heterogeneity and query-dependent uncertainty, DualResearch converts per-layer path evidence into answer distributions and fuses them in log space via an entropy-gated rule with global calibration. The fusion up-weights the more certain channel and amplifies agreement. As a complement to deep-research systems, DualResearch compresses lengthy multi-tool execution logs into a concise reasoning graph, and we show that it can reconstruct answers stably and effectively. On the scientific reasoning benchmarks HLE and GPQA, DualResearch achieves competitive performance. Using log files from the open-source system InternAgent, its accuracy improves by 7.7% on HLE and 6.06% on GPQA.

cs.AI↗

arXiv2025

$\bf{D^3}$QE: Learning Discrete Distribution Discrepancy-aware Quantization Error for Autoregressive-Generated Image Detection

The emergence of visual autoregressive (AR) models has revolutionized image generation while presenting new challenges for synthetic image detection. Unlike previous GAN or diffusion-based methods, AR models generate images through discrete token prediction, exhibiting both marked improvements in image synthesis quality and unique characteristics in their vector-quantized representations. In this paper, we propose to leverage Discrete Distribution Discrepancy-aware Quantization Error (D$^3$QE) for autoregressive-generated image detection that exploits the distinctive patterns and the frequency distribution bias of the codebook existing in real and fake images. We introduce a discrete distribution discrepancy-aware transformer that integrates dynamic codebook frequency statistics into its attention mechanism, fusing semantic features and quantization error latent. To evaluate our method, we construct a comprehensive dataset termed ARForensics covering 7 mainstream visual AR models. Experiments demonstrate superior detection accuracy and strong generalization of D$^3$QE across different AR models, with robustness to real-world perturbations. Code is available at \href{https://github.com/Zhangyr2022/D3QE}{https://github.com/Zhangyr2022/D3QE}.

cs.CV↗

arXiv2025

LiDAR-HMR: 3D Human Mesh Recovery from LiDAR

In recent years, point cloud perception tasks have been garnering increasing attention. This paper presents the first attempt to estimate 3D human body mesh from sparse LiDAR point clouds. We found that the major challenge in estimating human pose and mesh from point clouds lies in the sparsity, noise, and incompletion of LiDAR point clouds. Facing these challenges, we propose an effective sparse-to-dense reconstruction scheme to reconstruct 3D human mesh. This involves estimating a sparse representation of a human (3D human pose) and gradually reconstructing the body mesh. To better leverage the 3D structural information of point clouds, we employ a cascaded graph transformer (graphormer) to introduce point cloud features during sparse-to-dense reconstruction. Experimental results on three publicly available databases demonstrate the effectiveness of the proposed approach. Code: https://github.com/soullessrobot/LiDAR-HMR/

cs.CV↗

arXiv2025

Image Recognition with Online Lightweight Vision Transformer: A Survey

The Transformer architecture has achieved significant success in natural language processing, motivating its adaptation to computer vision tasks. Unlike convolutional neural networks, vision transformers inherently capture long-range dependencies and enable parallel processing, yet lack inductive biases and efficiency benefits, facing significant computational and memory challenges that limit its real-world applicability. This paper surveys various online strategies for generating lightweight vision transformers for image recognition, focusing on three key areas: Efficient Component Design, Dynamic Network, and Knowledge Distillation. We evaluate the relevant exploration for each topic on the ImageNet-1K benchmark, analyzing trade-offs among precision, parameters, throughput, and more to highlight their respective advantages, disadvantages, and flexibility. Finally, we propose future research directions and potential challenges in the lightweighting of vision transformers with the aim of inspiring further exploration and providing practical guidance for the community. Project Page: https://github.com/ajxklo/Lightweight-VIT

cs.CV↗

arXiv2025

Extending coherence time beyond break-even point using only drives and dissipation

Quantum error correction (QEC) aims to mitigate the loss of quantum information to the environment, which is a critical requirement for practical quantum computing. Existing QEC implementations heavily rely on measurement-based feedback, however, constraints on readout fidelity, hardware latency, and system complexity often limit both performance and scalability. Autonomous QEC (AQEC) seeks to overcome these obstacles by stabilizing logical codewords using introduced drives that provide coherent control and engineered dissipation. Here, we propose an AQEC protocol, derived from quantum channel simulation, that is applicable to arbitrary error-correcting codes. As a demonstration, we implement the protocol using a binomial code encoded in a long-lived bosonic mode (lifetime > 1ms), and extend the logical qubit coherence time to 1.04 times that of the best physical qubit in the system. This is the first experimental realization of an AQEC-protected bosonic logical qubit beyond the break-even point, proving that coherence time can indeed be extended by introducing only drives and dissipation. Our results highlight the performance and scalability potential of AQEC, marking an important step toward large-scale, universal quantum computing.

quant-ph↗

arXiv2025

ConCISE: Confidence-guided Compression in Step-by-step Efficient Reasoning

Large Reasoning Models (LRMs) perform strongly in complex reasoning tasks via Chain-of-Thought (CoT) prompting, but often suffer from verbose outputs, increasing computational overhead. Existing fine-tuning-based compression methods either operate post-hoc pruning, risking disruption to reasoning coherence, or rely on sampling-based selection, which fails to remove redundant content thoroughly. To address these limitations, this work begins by framing two key patterns of redundant reflection in LRMs--Confidence Deficit, wherein the model reflects on correct intermediate steps, and Termination Delay, where reflection continues after a verified, confident answer--through a confidence-guided perspective. Based on this, we introduce ConCISE (Confidence-guided Compression In Step-by-step Efficient Reasoning), a framework designed to generate concise reasoning chains, integrating Confidence Injection to boost reasoning confidence, and Early Stopping to terminate reasoning when confidence is sufficient. Extensive experiments demonstrate that compared to baseline methods, fine-tuning LRMs on ConCISE-generated data yields a better balance between compression and task performance, reducing length by up to approximately 50% under SimPO, while maintaining high task accuracy.

cs.LG↗

arXiv2025

Direct Estimation of Eigenvalues of Large Dimensional Precision Matrix

In this paper, we consider directly estimating the eigenvalues of precision matrix, without inverting the corresponding estimator for the eigenvalues of covariance matrix. We focus on a general asymptotic regime, i.e., the large dimensional regime, where both the dimension $N$ and the sample size $K$ tend to infinity whereas their quotient $N/K$ converges to a positive constant. By utilizing tools from random matrix theory, we construct an improved estimator for eigenvalues of precision matrix. We prove the consistency of the new estimator under large dimensional regime. In order to obtain the asymptotic bias term of the proposed estimator, we provide a theoretical result that characterizes the convergence rate of the expected Stieltjes transform (with its derivative) of the spectra of the sample covariance matrix. Using this result, we prove that the asymptotic bias term of the proposed estimator is of order $O(1/K^2)$. Additionally, we establish a central limiting theorem (CLT) to describe the fluctuations of the new estimator. Finally, some numerical examples are presented to validate the excellent performance of the new estimator and to verify the accuracy of the CLT.

math.ST↗

arXiv2025

LiteRSan: Lightweight Memory Safety Via Rust-specific Program Analysis and Selective Instrumentation

Rust is a memory-safe language, and its strong safety guarantees combined with high performance have been attracting widespread adoption in systems programming and security-critical applications. However, Rust permits the use of unsafe code, which bypasses compiler-enforced safety checks and can introduce memory vulnerabilities. A widely adopted approach for detecting memory safety bugs in Rust is Address Sanitizer (ASan). Optimized versions, such as ERASan and RustSan, have been proposed to selectively apply security checks in order to reduce performance overhead. However, these tools still incur significant performance and memory overhead and fail to detect many classes of memory safety vulnerabilities due to the inherent limitations of ASan. In this paper, we present LiteRSan, a novel memory safety sanitizer that addresses the limitations of prior approaches. By leveraging Rust's unique ownership model, LiteRSan performs Rust-specific static analysis that is aware of pointer lifetimes to identify risky pointers. It then selectively instruments risky pointers to enforce only the necessary spatial or temporal memory safety checks. Consequently, LiteRSan introduces significantly lower runtime overhead (18.84% versus 152.05% and 183.50%) and negligible memory overhead (0.81% versus 739.27% and 861.98%) compared with existing ASan-based sanitizers while being capable of detecting memory safety bugs that prior techniques miss.

cs.CR↗

arXiv2025

Mitigating Strategy Preference Bias in Emotional Support Conversation via Uncertainty Estimations

Emotional support conversation (ESC) aims to alleviate distress through empathetic dialogue, yet large language models (LLMs) face persistent challenges in delivering effective ESC due to low accuracy in strategy planning. Moreover, there is a considerable preference bias towards specific strategies. Prior methods using fine-tuned strategy planners have shown potential in reducing such bias, while the underlying causes of the preference bias in LLMs have not well been studied. To address these issues, we first reveal the fundamental causes of the bias by identifying the knowledge boundaries of LLMs in strategy planning. Then, we propose an approach to mitigate the bias by reinforcement learning with a dual reward function, which optimizes strategy planning via both accuracy and entropy-based confidence for each region according to the knowledge boundaries. Experiments on the ESCov and ExTES datasets with multiple LLM backbones show that our approach outperforms the baselines, confirming the effectiveness of our approach.

cs.CL↗

arXiv2025

Black-box Model Merging for Language-Model-as-a-Service with Massive Model Repositories

Model merging refers to the process of integrating multiple distinct models into a unified model that preserves and combines the strengths and capabilities of the individual models. Most existing approaches rely on task vectors to combine models, typically under the assumption that model parameters are accessible. However, for extremely large language models (LLMs) such as GPT-4, which are often provided solely as black-box services through API interfaces (Language-Model-as-a-Service), model weights are not available to end users. This presents a significant challenge, which we refer to as black-box model merging (BMM) with massive LLMs. To address this challenge, we propose a derivative-free optimization framework based on the evolutionary algorithm (Evo-Merging) that enables effective model merging using only inference-time API queries. Our method consists of two key components: (1) sparsity-based denoising, designed to identify and filter out irrelevant or redundant information across models, and (2) sign-aware scaling, which dynamically computes optimal combination weights for the relevant models based on their performance. We also provide a formal justification, along with a theoretical analysis, for our asymmetric sparsification. Extensive experimental evaluations demonstrate that our approach achieves state-of-the-art results on a range of tasks, significantly outperforming existing strong baselines.

cs.AI↗

arXiv2025

MiniCPM-V 4.5: Cooking Efficient MLLMs via Architecture, Data, and Training Recipe

Multimodal Large Language Models (MLLMs) are undergoing rapid progress and represent the frontier of AI development. However, their training and inference efficiency have emerged as a core bottleneck in making MLLMs more accessible and scalable. To address the challenges, we present MiniCPM-V 4.5, an 8B parameter model designed for high efficiency and strong performance. We introduce three core improvements in model architecture, data strategy and training method: a unified 3D-Resampler model architecture for highly compact encoding over images and videos, a unified learning paradigm for document knowledge and text recognition without heavy data engineering, and a hybrid reinforcement learning strategy for proficiency in both short and long reasoning modes. Comprehensive experimental results in OpenCompass evaluation show that MiniCPM-V 4.5 surpasses widely used proprietary models such as GPT-4o-latest, and significantly larger open-source models such as Qwen2.5-VL 72B. Notably, the strong performance is achieved with remarkable efficiency. For example, on the widely adopted VideoMME benchmark, MiniCPM-V 4.5 achieves state-of-the-art performance among models under 30B size, using just 46.7\% GPU memory cost and 8.7\% inference time of Qwen2.5-VL 7B.

cs.LG↗

arXiv2025

CM-Align: Consistency-based Multilingual Alignment for Large Language Models

Current large language models (LLMs) generally show a significant performance gap in alignment between English and other languages. To bridge this gap, existing research typically leverages the model's responses in English as a reference to select the best/worst responses in other languages, which are then used for Direct Preference Optimization (DPO) training. However, we argue that there are two limitations in the current methods that result in noisy multilingual preference data and further limited alignment performance: 1) Not all English responses are of high quality, and using a response with low quality may mislead the alignment for other languages. 2) Current methods usually use biased or heuristic approaches to construct multilingual preference pairs. To address these limitations, we design a consistency-based data selection method to construct high-quality multilingual preference data for improving multilingual alignment (CM-Align). Specifically, our method includes two parts: consistency-guided English reference selection and cross-lingual consistency-based multilingual preference data construction. Experimental results on three LLMs and three common tasks demonstrate the effectiveness and superiority of our method, which further indicates the necessity of constructing high-quality preference data.

cs.CL↗

arXiv2025

GC-VLN: Instruction as Graph Constraints for Training-free Vision-and-Language Navigation

In this paper, we propose a training-free framework for vision-and-language navigation (VLN). Existing zero-shot VLN methods are mainly designed for discrete environments or involve unsupervised training in continuous simulator environments, which makes it challenging to generalize and deploy them in real-world scenarios. To achieve a training-free framework in continuous environments, our framework formulates navigation guidance as graph constraint optimization by decomposing instructions into explicit spatial constraints. The constraint-driven paradigm decodes spatial semantics through constraint solving, enabling zero-shot adaptation to unseen environments. Specifically, we construct a spatial constraint library covering all types of spatial relationship mentioned in VLN instructions. The human instruction is decomposed into a directed acyclic graph, with waypoint nodes, object nodes and edges, which are used as queries to retrieve the library to build the graph constraints. The graph constraint optimization is solved by the constraint solver to determine the positions of waypoints, obtaining the robot's navigation path and final goal. To handle cases of no solution or multiple solutions, we construct a navigation tree and the backtracking mechanism. Extensive experiments on standard benchmarks demonstrate significant improvements in success rate and navigation efficiency compared to state-of-the-art zero-shot VLN methods. We further conduct real-world experiments to show that our framework can effectively generalize to new environments and instruction sets, paving the way for a more robust and autonomous navigation framework.

cs.RO↗

arXiv2025

From Rigging to Waving: 3D-Guided Diffusion for Natural Animation of Hand-Drawn Characters

Hand-drawn character animation is a vibrant field in computer graphics, presenting challenges in achieving geometric consistency while conveying expressive motion. Traditional skeletal animation methods maintain geometric consistency but struggle with complex non-rigid elements like flowing hair and skirts, leading to unnatural deformation. Conversely, video diffusion models synthesize realistic dynamics but often create geometric distortions in stylized drawings due to domain gaps. This work proposes a hybrid animation system that combines skeletal animation and video diffusion. Initially, coarse images are generated from characters retargeted with skeletal animations for geometric guidance. These images are then enhanced in texture and secondary dynamics using video diffusion priors, framing this enhancement as an inpainting task. A domain-adapted diffusion model refines user-masked regions needing improvement, especially for secondary dynamics. To enhance motion realism further, we introduce a Secondary Dynamics Injection (SDI) strategy in the denoising process, incorporating features from a pre-trained diffusion model enriched with human motion priors. Additionally, to tackle unnatural deformations from low-poly single-mesh character modeling, we present a Hair Layering Modeling (HLM) technique that uses segmentation maps to separate hair from the body, allowing for more natural animation of long-haired characters. Extensive experiments show that our system outperforms state-of-the-art methods in both quantitative and qualitative evaluations.

cs.GR↗

arXiv2025

MiniCPM4: Ultra-Efficient LLMs on End Devices

This paper introduces MiniCPM4, a highly efficient large language model (LLM) designed explicitly for end-side devices. We achieve this efficiency through systematic innovation in four key dimensions: model architecture, training data, training algorithms, and inference systems. Specifically, in terms of model architecture, we propose InfLLM v2, a trainable sparse attention mechanism that accelerates both prefilling and decoding phases for long-context processing. Regarding training data, we propose UltraClean, an efficient and accurate pre-training data filtering and generation strategy, and UltraChat v2, a comprehensive supervised fine-tuning dataset. These datasets enable satisfactory model performance to be achieved using just 8 trillion training tokens. Regarding training algorithms, we propose ModelTunnel v2 for efficient pre-training strategy search, and improve existing post-training methods by introducing chunk-wise rollout for load-balanced reinforcement learning and data-efficient tenary LLM, BitCPM. Regarding inference systems, we propose CPM.cu that integrates sparse attention, model quantization, and speculative sampling to achieve efficient prefilling and decoding. To meet diverse on-device requirements, MiniCPM4 is available in two versions, with 0.5B and 8B parameters, respectively. Furthermore, we construct a hybrid reasoning model, MiniCPM4.1, which can be used in both deep reasoning mode and non-reasoning mode. Evaluation results demonstrate that MiniCPM4 and MiniCPM4.1 outperform similar-sized open-source models across benchmarks, with the 8B variants showing significant speed improvements on long sequence understanding and generation.

cs.CL↗

arXiv2025

A Synthetic-to-Real Dehazing Method based on Domain Unification

Due to distribution shift, the performance of deep learning-based method for image dehazing is adversely affected when applied to real-world hazy images. In this paper, we find that such deviation in dehazing task between real and synthetic domains may come from the imperfect collection of clean data. Owing to the complexity of the scene and the effect of depth, the collected clean data cannot strictly meet the ideal conditions, which makes the atmospheric physics model in the real domain inconsistent with that in the synthetic domain. For this reason, we come up with a synthetic-to-real dehazing method based on domain unification, which attempts to unify the relationship between the real and synthetic domain, thus to let the dehazing model more in line with the actual situation. Extensive experiments qualitatively and quantitatively demonstrate that the proposed dehazing method significantly outperforms state-of-the-art methods on real-world images.

eess.IV↗