arXiv subjects

Haodong Zhu

Publications and source records attributed to Haodong Zhu.

At least 19 recordsLinked to original sources

arXiv2026

Kalman Meets Curriculum: Efficient Dynamic Prompt Selection for Adaptive RL Finetuning

Reinforcement learning (RL) finetuning significantly enhances the reasoning capabilities of large language models (LLMs), yet its effectiveness critically depends on selecting prompts of appropriate difficulty for the current policy. This is challenging because prompt difficulty evolves throughout training. Existing online methods therefore face a trade-off: evaluation-based approaches are accurate but expensive, while prediction-based approaches are efficient but typically assume stationary difficulty, making them ill-suited to RL's non-stationary training dynamics. To address these issues, we propose a Kalman-Guided Prompt Selection method (KGPS), which reformulates prompt selection as a dynamic state estimation problem rather than static difficulty prediction. KGPS models each prompt's latent success rate in logit space using a linear-Gaussian state-space model, with process noise coupled to the magnitude of policy updates so that uncertainty increases when the policy changes more substantially. A Kalman filter then maintains a calibrated Gaussian posterior over prompt difficulty, and prompts are selected by maximizing a posterior-expected training utility that favors intermediate-difficulty prompts while naturally revisiting uncertain ones. The resulting procedure is adaptive to policy drift and requires no additional rollouts beyond standard policy training. Extensive experiments across mathematics, planning, and geometry reasoning benchmarks, as well as multiple RL algorithms, show that KGPS consistently improves both final accuracy and rollout efficiency over strong baselines, establishing state-of-the-art performance among online prompt selection methods. For example, on DeepSeek-R1-Distill-7B, KGPS uses 83% fewer rollouts than DS while even improving the average performance by 0.12 point across six math reasoning benchmarks.

cs.LG↗

arXiv2026

LDFE: Laplacian Decoupled Feature Enhancement Block for Dual-Stream CNN-based RGB-IR Object Detection

The complementary information between RGB and IR images can significantly enhance object detection performance under extreme conditions. Existing methods prefer dual-stream CNN backbones built upon YOLO for feature extraction and focus on the design of feature fusion. In this paper, we introduce the Laplacian Decoupled Feature Enhancement block (LDFE) to fuse features from different stages of the dual-stream CNN backbone. By design, LDFE simultaneously considers the characteristics of modalities and structures for feature fusion by employing global-local decomposition, denoising, fusion, and reconstruction, sequentially. The LDFE first separates features into global and local components based on Laplacian Pyramid, and then performs denoising and fusion based on Global State Space Enhancement module (GS2E) and Local Convolutional Correlation Enhancement module (LC2E) separately. Specifically, the GS2E conducts a two-branch architecture for the main and auxiliary modalities. It dynamically suppresses noise in the main modality through cross-modal attention derived from the auxiliary modality, while employing a State Space Model to capture long-range dependencies within the global feature representations of the main modality. To obtain bidirectional interaction, the two modalities systematically alternate their main/auxiliary roles. Moreover, the LC2E suppresses noise in local features and leverages spatial and channel dimension along with triple convolution to extract fine-grained details for fusion. These innovative designs achieve a significant performance improvement, with mAP surpassing the SOTA methods 6.2%, 3.7%, 4.7%, 2.3%, 4.1% and 2.0% on M3FD, DroneVehicle, LLVIP, FLIR-Aligned, KAIST and VEDAI datasets,respectively.

cs.CV↗

arXiv2026

InfraNet: Quality-Aware RGB Guidance for Efficient Infrared Object Detection

Robust object detection under adverse visual conditions remains a long-standing challenge for multi-modal perception systems. Existing fusion-based methods typically require both RGB and infrared (IR) inputs, and treat them equally during both training and inference, which compromises their robustness when the RGB modality becomes unreliable or unavailable. In this case, we propose \textbf{InfraNet}, an IR-centric quality-aware framework that regulates RGB guidance during training and supports flexible RGB--IR or IR-only deployment. InfraNet employs an asymmetric architecture where the primary IR pathway extracts multi-scale infrared features for predictions, while the auxiliary RGB pathway provides reliability-controlled supervisory signals. The core of InfraNet is \textbf{QualGate}, a quality-aware fusion module that learns a task-oriented control signal to suppress unreliable RGB guidance and compensate IR features during cross-modal training. Built upon InfraNet, we design two architectural variants: a lightweight IR-only architecture InfraNet-IR and an RGB--IR architecture InfraNet-RGB-IR. Our method is evaluated through extensive experiments on four benchmark datasets (LLVIP, FLIR-Aligned, M$^3$FD, and DroneVehicle), showing strong or competitive accuracy in challenging low-light and adverse weather conditions. Notably, InfraNet maintains high efficiency in IR-only inference, making it both accurate and computationally efficient.

cs.CV↗

arXiv2026

OpenHLM: An Empirical Recipe for Whole-Body Humanoid Loco-Manipulation

Whole-body humanoid loco-manipulation requires coordinating the robot's entire kinematic chain. However, most existing systems typically decouple the upper and lower bodies into separate controllers, limiting such coordination and yielding behaviors similar to those of a wheeled dual-arm platform. In this paper, we ask what it takes to build a whole-body native vision-language-action (VLA) model that maps language and pixels directly to all of the humanoid's degrees of freedom. We conduct a systematic empirical study organized as a roadmap of one-variable-at-a-time experiments across three phases: whole-body teleoperation, VLA model design, and heterogeneous co-training. Our study yields several intriguing findings: a joint-based whole-body teleoperation interface outperforms alternatives that only partially expose the humanoid's degrees of freedom; a VLA pretrained on static and wheeled dual-arm platforms transfers surprisingly well to a humanoid's full action space; and co-training with HuMI, the humanoid analog of UMI, extends the policy to new objects and instructions without additional whole-body teleoperation on those targets. Following this roadmap yields OpenHLM, an open-source recipe for whole-body humanoid loco-manipulation. In a challenging long-horizon task that spans a wide vertical range of the humanoid, OpenHLM outperforms two state-of-the-art humanoid VLA baselines (GR00T N1.6 and $\Psi_0$) using less than half the total demonstration time. Our code, training data, and model checkpoints are available at [https://openhlm-project.github.io/].

cs.RO↗

arXiv2026

Unbiased Dynamic Pruning for Efficient Group-Based Policy Optimization

Group Relative Policy Optimization (GRPO) effectively scales LLM reasoning but incurs prohibitive computational costs due to its extensive group-based sampling requirement. While recent selective data utilization methods can mitigate this overhead, they could induce estimation bias by altering the underlying sampling distribution, compromising theoretical rigor and convergence behavior. To address this limitation, we propose Dynamic Pruning Policy Optimization (DPPO), a framework that enables dynamic pruning while preserving unbiased gradient estimation through importance sampling-based correction. By incorporating mathematically derived rescaling factors, DPPO significantly accelerates GRPO training without altering the optimization objective of the full-batch baseline. Furthermore, to mitigate the data sparsity induced by pruning, we introduce Dense Prompt Packing, a window-based greedy strategy that maximizes valid token density and hardware utilization. Extensive experiments demonstrate that DPPO consistently accelerates training across diverse models and benchmarks. For instance, on Qwen3-4B trained on MATH, DPPO achieves 2.37$\times$ training speedup and outperforms GRPO by 3.36% in average accuracy across six mathematical reasoning benchmarks.

cs.LG↗

arXiv2026

UrbanGS: A Scalable and Efficient Architecture for Geometrically Accurate Large-Scene Reconstruction

While 3D Gaussian Splatting (3DGS) enables high-quality, real-time rendering for bounded scenes, its extension to large-scale urban environments gives rise to critical challenges in terms of geometric consistency, memory efficiency, and computational scalability. To address these issues, we present UrbanGS, a scalable reconstruction framework that effectively tackles these challenges for city-scale applications. First, we propose a Depth-Consistent D-Normal Regularization module. Unlike existing approaches that rely solely on monocular normal estimators, which can effectively update rotation parameters yet struggle to update position parameters, our method integrates D-Normal constraints with external depth supervision. This allows for comprehensive updates of all geometric parameters. By further incorporating an adaptive confidence weighting mechanism based on gradient consistency and inverse depth deviation, our approach significantly enhances multi-view depth alignment and geometric coherence, which effectively resolves the issue of geometric accuracy in complex large-scale scenes. To improve scalability, we introduce a Spatially Adaptive Gaussian Pruning (SAGP) strategy, which dynamically adjusts Gaussian density based on local geometric complexity and visibility to reduce redundancy. Additionally, a unified partitioning and view assignment scheme is designed to eliminate boundary artifacts and optimize computational load. Extensive experiments on multiple urban datasets demonstrate that UrbanGS achieves superior performance in rendering quality, geometric accuracy, and memory efficiency, providing a systematic solution for high-fidelity large-scale scene reconstruction.

cs.CV↗

arXiv2026

Humanoid Manipulation Interface: Humanoid Whole-Body Manipulation from Robot-Free Demonstrations

Current approaches for humanoid whole-body manipulation, primarily relying on teleoperation or visual sim-to-real reinforcement learning, are hindered by hardware logistics and complex reward engineering. Consequently, demonstrated autonomous skills remain limited and are typically restricted to controlled environments. In this paper, we present the Humanoid Manipulation Interface (HuMI), a portable and efficient framework for learning diverse whole-body manipulation tasks across various environments. HuMI enables robot-free data collection by capturing rich whole-body motion using portable hardware. This data drives a hierarchical learning pipeline that translates human motions into dexterous and feasible humanoid skills. Extensive experiments across five whole-body tasks--including kneeling, squatting, tossing, walking, and bimanual manipulation--demonstrate that HuMI achieves a 3x increase in data collection efficiency compared to teleoperation and attains a 70% success rate in unseen environments.

cs.RO↗

arXiv2026

Minimum Number of Monochromatic Subgraphs of a Random Graph

We consider the problem of minimizing the number of monochromatic subgraphs of a random graph, when each node of the host graph is assigned one of the two colors. Using a recently discovered contiguity between appearance of strictly balanced subgraphs $F$ in a random graph, and random hypergraphs where copies of $F$ are generated independently, we show that the minimum value converges to a limit, when the expected number of copies of $F$ is linear in the number of nodes $|V|$. Furthermore, using the connections with mean field spin glass models, we obtain an asymptotic expression for this limit as the normalized expected number of copies of $F$ and the size of $F$ diverge to infinity.

math.CO↗

arXiv2025

PartDiffuser: Part-wise 3D Mesh Generation via Discrete Diffusion

Existing autoregressive (AR) methods for generating artist-designed meshes struggle to balance global structural consistency with high-fidelity local details, and are susceptible to error accumulation. To address this, we propose PartDiffuser, a novel semi-autoregressive diffusion framework for point-cloud-to-mesh generation. The method first performs semantic segmentation on the mesh and then operates in a "part-wise" manner: it employs autoregression between parts to ensure global topology, while utilizing a parallel discrete diffusion process within each semantic part to precisely reconstruct high-frequency geometric features. PartDiffuser is based on the DiT architecture and introduces a part-aware cross-attention mechanism, using point clouds as hierarchical geometric conditioning to dynamically control the generation process, thereby effectively decoupling the global and local generation tasks. Experiments demonstrate that this method significantly outperforms state-of-the-art (SOTA) models in generating 3D meshes with rich detail, exhibiting exceptional detail representation suitable for real-world applications.

cs.CV↗

arXiv2025

Asymptotic size of the Karp-Sipser Core in Configuration Model

We study the asymptotic size of the Karp-Sipser core in the configuration model with arbitrary degree distributions. The Karp-Sipser core is the induced subgraph obtained by iteratively removing all leaves and their neighbors through the leaf-removal process, and finally discarding any isolated vertices \cite{BCC}. Our main result establishes the convergence of the Karp-Sipser core size to an explicit fixed-point equation under general degree assumptions.The approach is based on analyzing the corresponding local weak limit of the configuration model - a unimodular Galton-Watson tree and tracing the evolution process of all vertex states under leaf-removal dynamics by use of the working mechanism of an enhanced version of Warning Propagation along with Node Labeling Propagation.

math.CO↗

arXiv2025

Surf3R: Rapid Surface Reconstruction from Sparse RGB Views in Seconds

Current multi-view 3D reconstruction methods rely on accurate camera calibration and pose estimation, requiring complex and time-intensive pre-processing that hinders their practical deployment. To address this challenge, we introduce Surf3R, an end-to-end feedforward approach that reconstructs 3D surfaces from sparse views without estimating camera poses and completes an entire scene in under 10 seconds. Our method employs a multi-branch and multi-view decoding architecture in which multiple reference views jointly guide the reconstruction process. Through the proposed branch-wise processing, cross-view attention, and inter-branch fusion, the model effectively captures complementary geometric cues without requiring camera calibration. Moreover, we introduce a D-Normal regularizer based on an explicit 3D Gaussian representation for surface reconstruction. It couples surface normals with other geometric parameters to jointly optimize the 3D geometry, significantly improving 3D consistency and surface detail accuracy. Experimental results demonstrate that Surf3R achieves state-of-the-art performance on multiple surface reconstruction metrics on ScanNet++ and Replica datasets, exhibiting excellent generalization and efficiency.

cs.GR↗

arXiv2025

The asymptotic rank of adjacency matrices of weighted configuration models over arbitrary fields

We study the asymptotic rank of adjacency matrices of a large class of edge-weighted configuration models. Here, the weight of a (multi-)edge can be any fixed non-zero element from an arbitrary field, as long as it is independent of the (multi-)graph. Our main result demonstrates that the asymptotic behavior of the normalized rank of the adjacency matrix neither depends on the fixed edge-weights, nor on which field they are chosen from. Our approach relies on a novel adaptation of the component exploration method of \cite{janson2009new}, which enables the application of combinatorial techniques from \cite{coja2022rank, HofMul25}.

math.CO↗

arXiv2025

Squeeze10-LLM: Squeezing LLMs' Weights by 10 Times via a Staged Mixed-Precision Quantization Method

Deploying large language models (LLMs) is challenging due to their massive parameters and high computational costs. Ultra low-bit quantization can significantly reduce storage and accelerate inference, but extreme compression (i.e., mean bit-width <= 2) often leads to severe performance degradation. To address this, we propose Squeeze10-LLM, effectively "squeezing" 16-bit LLMs' weights by 10 times. Specifically, Squeeze10-LLM is a staged mixed-precision post-training quantization (PTQ) framework and achieves an average of 1.6 bits per weight by quantizing 80% of the weights to 1 bit and 20% to 4 bits. We introduce Squeeze10LLM with two key innovations: Post-Binarization Activation Robustness (PBAR) and Full Information Activation Supervision (FIAS). PBAR is a refined weight significance metric that accounts for the impact of quantization on activations, improving accuracy in low-bit settings. FIAS is a strategy that preserves full activation information during quantization to mitigate cumulative error propagation across layers. Experiments on LLaMA and LLaMA2 show that Squeeze10-LLM achieves state-of-the-art performance for sub-2bit weight-only quantization, improving average accuracy from 43% to 56% on six zero-shot classification tasks--a significant boost over existing PTQ methods.

cs.LG↗

arXiv2025

WaveMamba: Wavelet-Driven Mamba Fusion for RGB-Infrared Object Detection

Leveraging the complementary characteristics of visible (RGB) and infrared (IR) imagery offers significant potential for improving object detection. In this paper, we propose WaveMamba, a cross-modality fusion method that efficiently integrates the unique and complementary frequency features of RGB and IR decomposed by Discrete Wavelet Transform (DWT). An improved detection head incorporating the Inverse Discrete Wavelet Transform (IDWT) is also proposed to reduce information loss and produce the final detection results. The core of our approach is the introduction of WaveMamba Fusion Block (WMFB), which facilitates comprehensive fusion across low-/high-frequency sub-bands. Within WMFB, the Low-frequency Mamba Fusion Block (LMFB), built upon the Mamba framework, first performs initial low-frequency feature fusion with channel swapping, followed by deep fusion with an advanced gated attention mechanism for enhanced integration. High-frequency features are enhanced using a strategy that applies an ``absolute maximum" fusion approach. These advancements lead to significant performance gains, with our method surpassing state-of-the-art approaches and achieving average mAP improvements of 4.5% on four benchmarks.

cs.CV↗

arXiv2025

HRGS: Hierarchical Gaussian Splatting for Memory-Efficient High-Resolution 3D Reconstruction

3D Gaussian Splatting (3DGS) has made significant strides in real-time 3D scene reconstruction, but faces memory scalability issues in high-resolution scenarios. To address this, we propose Hierarchical Gaussian Splatting (HRGS), a memory-efficient framework with hierarchical block-level optimization. First, we generate a global, coarse Gaussian representation from low-resolution data. Then, we partition the scene into multiple blocks, refining each block with high-resolution data. The partitioning involves two steps: Gaussian partitioning, where irregular scenes are normalized into a bounded cubic space with a uniform grid for task distribution, and training data partitioning, where only relevant observations are retained for each block. By guiding block refinement with the coarse Gaussian prior, we ensure seamless Gaussian fusion across adjacent blocks. To reduce computational demands, we introduce Importance-Driven Gaussian Pruning (IDGP), which computes importance scores for each Gaussian and removes those with minimal contribution, speeding up convergence and reducing memory usage. Additionally, we incorporate normal priors from a pretrained model to enhance surface reconstruction quality. Our method enables high-quality, high-resolution 3D scene reconstruction even under memory constraints. Extensive experiments on three benchmarks show that HRGS achieves state-of-the-art performance in high-resolution novel view synthesis (NVS) and surface reconstruction tasks.

cs.CV↗

arXiv2025

Asymptotic diameter of preferential attachment model

We study the asymptotic diameter of the preferential attachment model $\operatorname{PA}\!_n^{(m,δ)}$ with parameters $m \ge 2$ and $δ> 0$. Building on the recent work \cite{VZ25}, we prove that the diameter of $G_n \sim \operatorname{PA}\!_n^{(m,δ)}$ is $(1+o(1))\log_νn$ with high probability, where $ν$ is the exponential growth rate of the local weak limit of $G_n$. Our result confirms the conjecture in \cite{VZ25} and closes the remaining gap in understanding the asymptotic diameter of preferential attachment graphs with general parameters $m \ge 1$ and $δ>-m$. Our proof follows a general recipe that relates the diameter of a random graph to its typical distance, which we expect to have applicability in a broader range of models.

math.PR↗

arXiv2025

Logarithmic typical distances in preferential attachment models

We prove that the typical distances in a preferential attachment model with out-degree $m\geq 2$ and strictly positive fitness parameter are close to $\log_\nu{n}$, where $\nu$ is the exponential growth parameter of the local limit of the preferential attachment model. The proof relies on a path-counting technique, the first- and second-moment methods, as well as a novel proof of the convergence of the spectral radius of the offspring operator under a certain truncation.

math.PR↗

arXiv2024

Noisy Linear Group Testing: Exact Thresholds and Efficient Algorithms

In group testing, the task is to identify defective items by testing groups of them together using as few tests as possible. We consider the setting where each item is defective with a constant probability $α$, independent of all other items. In the (over-)idealized noiseless setting, tests are positive exactly if any of the tested items are defective. We study a more realistic model in which observed test results are subject to noise, i.e., tests can display false positive or false negative results with constant positive probabilities. We determine precise constants $c$ such that $cn\log n$ tests are required to recover the infection status of every individual for both adaptive and non-adaptive group testing: in the former, the selection of groups to test can depend on previously observed test results, whereas it cannot in the latter. Additionally, for both settings, we provide efficient algorithms that identify all defective items with the optimal amount of tests with high probability. Thus, we completely solve the problem of binary noisy group testing in the studied setting.

cs.DM↗