arXiv subjects

Xin Luo

Publications and source records attributed to Xin Luo.

At least 37 records · Page 2Linked to original sources

arXiv2026

Learning Thermal-Aware Locomotion Policies for an Electrically-Actuated Quadruped Robot

Electrically-actuated quadrupedal robots possess high mobility on complex terrains, but their motors tend to accumulate heat under high-torque cyclic loads, potentially triggering overheat protection and limiting long-duration tasks. This work proposes a thermal-aware control method that incorporates motor temperatures into reinforcement learning locomotion policies and introduces thermal-constraint rewards to prevent temperature exceedance. Real-world experiments on the Unitree A1 demonstrate that, under a fixed 3 kg payload, the baseline policy triggers overheat protection and stops within approximately 7 minutes, whereas the proposed method can operate continuously for over 27 minutes without thermal interruptions while maintaining comparable command-tracking performance, thereby enhancing sustainable operational capability.

cs.RO↗

arXiv2026

Progressively Exploring and Exploiting Inference Data to Break Fine-Grained Classification Barrier

Current fine-grained classification research primarily focuses on fine-grained feature learning. However, in real-world scenarios, fine-grained data annotation is challenging, and the features and semantics are highly diverse and frequently changing. These issues create inherent barriers between traditional experimental settings and real-world applications, limiting the effectiveness of conventional fine-grained classification methods. Although some recent studies have provided potential solutions to these issues, most of them still rely on limited supervised information and thus fail to offer effective solutions. In this paper, based on theoretical analysis, we propose a novel learning paradigm to break the barriers in fine-grained classification. This paradigm enables the model to progressively learn during inference, thereby leveraging cost-free data at inference time to more accurately represent fine-grained categories and adapt to dynamic semantic changes. On this basis, an efficient EXPloring and EXPloiting strategy and method (EXP2) is designed. Thereinto, useful inference data samples are explored according to class representations and exploited to optimize classifiers. Experimental results demonstrate the general effectiveness of our method, providing guidance for future in-depth understanding and exploration of real-world fine-grained classification.

cs.CV↗

arXiv2026

Quantized-Tinyllava: a new multimodal foundation model enables efficient split learning

Multimodal foundation models are increasingly trained on sensitive data across domains such as finance, biomedicine, and personal identifiers. However, this distributed setup raises serious privacy concerns due to the need for cross-partition data sharing. Split learning addresses these concerns by enabling collaborative model training without raw data exchange between partitions, yet it introduces a significant challenge: transmitting high-dimensional intermediate feature representations between partitions leads to substantial communication costs. To address this challenge, we propose Quantized-TinyLLaVA, a multimodal foundation model with an integrated communication-efficient split learning framework. Our approach adopts a compression module that quantizes intermediate feature into discrete representations before transmission, substantially reducing communication overhead. Besides, we derive a principled quantization strategy grounded in entropy coding theory to determine the optimal number of discrete representation levels. We deploy our framework in a two-partition setting, with one partition operating as the client and the other as the server, to realistically simulate distributed training. Under this setup, Quantized-TinyLLaVA achieves an approximate \textbf{87.5\%} reduction in communication overhead with 2-bit quantization, while maintaining performance of the original 16-bit model across five benchmark datasets. Furthermore, our compressed representations exhibit enhanced resilience against feature inversion attacks, validating the privacy of transmission. The code is available at https://github.com/anonymous-1742/Quantized-TinyLLaVA.

cs.LG↗

arXiv2026

EditScore: Unlocking Online RL for Image Editing via High-Fidelity Reward Modeling

Instruction-guided image editing has achieved remarkable progress, yet current models still face challenges with complex instructions and often require multiple samples to produce a desired result. Reinforcement Learning (RL) offers a promising solution, but its adoption in image editing has been severely hindered by the lack of a high-fidelity, efficient reward signal. In this work, we present a comprehensive methodology to overcome this barrier, centered on the development of a state-of-the-art, specialized reward model. We first introduce EditReward-Bench, a comprehensive benchmark to systematically evaluate reward models on editing quality. Building on this benchmark, we develop EditScore, a series of reward models (7B-72B) for evaluating the quality of instruction-guided image editing. Through meticulous data curation and filtering, EditScore effectively matches the performance of learning proprietary VLMs. Furthermore, coupled with an effective self-ensemble strategy tailored for the generative nature of EditScore, our largest variant even surpasses GPT-5 in the benchmark. We then demonstrate that a high-fidelity reward model is the key to unlocking online RL for image editing. Our experiments show that, while even the largest open-source VLMs fail to provide an effective learning signal, EditScore enables efficient and robust policy optimization. Applying our framework to a strong base model, OmniGen2, results in a final model that shows a substantial and consistent performance uplift. Overall, this work provides the first systematic path from benchmarking to reward modeling to RL training in image editing, showing that a high-fidelity, domain-specialized reward model is the key to unlocking the full potential of RL in this domain.

cs.CV↗

arXiv2026

Multi-granularity Interactive Attention Framework for Residual Hierarchical Pronunciation Assessment

Automatic pronunciation assessment plays a crucial role in computer-assisted pronunciation training systems. Due to the ability to perform multiple pronunciation tasks simultaneously, multi-aspect multi-granularity pronunciation assessment methods are gradually receiving more attention and achieving better performance than single-level modeling tasks. However, existing methods only consider unidirectional dependencies between adjacent granularity levels, lacking bidirectional interaction among phoneme, word, and utterance levels and thus insufficiently capturing the acoustic structural correlations. To address this issue, we propose a novel residual hierarchical interactive method, HIA for short, that enables bidirectional modeling across granularities. As the core of HIA, the Interactive Attention Module leverages an attention mechanism to achieve dynamic bidirectional interaction, effectively capturing linguistic features at each granularity while integrating correlations between different granularity levels. We also propose a residual hierarchical structure to alleviate the feature forgetting problem when modeling acoustic hierarchies. In addition, we use 1-D convolutional layers to enhance the extraction of local contextual cues at each granularity. Extensive experiments on the speechocean762 dataset show that our model is comprehensively ahead of the existing state-of-the-art methods.

cs.CL↗

arXiv2026

CSCBench: A PVC Diagnostic Benchmark for Commodity Supply Chain Reasoning

Large Language Models (LLMs) have achieved remarkable success in general benchmarks, yet their competence in commodity supply chains (CSCs) -- a domain governed by institutional rule systems and feasibility constraints -- remains under-explored. CSC decisions are shaped jointly by process stages (e.g., planning, procurement, delivery), variety-specific rules (e.g., contract specifications and delivery grades), and reasoning depth (from retrieval to multi-step analysis and decision selection). We introduce CSCBench, a 2.3K+ single-choice benchmark for CSC reasoning, instantiated through our PVC 3D Evaluation Framework (Process, Variety, and Cognition). The Process axis aligns tasks with SCOR+Enable; the Variety axis operationalizes commodity-specific rule systems under coupled material-information-financial constraints, grounded in authoritative exchange guidebooks/rulebooks and industry reports; and the Cognition axis follows Bloom's revised taxonomy. Evaluating representative LLMs under a direct prompting setting, we observe strong performance on the Process and Cognition axes but substantial degradation on the Variety axis, especially on Freight Agreements. CSCBench provides a diagnostic yardstick for measuring and improving LLM capabilities in this high-stakes domain.

cs.CL↗

arXiv2025

NCSAC: Effective Neural Community Search via Attribute-augmented Conductance

Identifying locally dense communities closely connected to the user-initiated query node is crucial for a wide range of applications. Existing approaches either solely depend on rule-based constraints or exclusively utilize deep learning technologies to identify target communities. Therefore, an important question is proposed: can deep learning be integrated with rule-based constraints to elevate the quality of community search? In this paper, we affirmatively address this question by introducing a novel approach called Neural Community Search via Attribute-augmented Conductance, abbreviated as NCSAC. Specifically, NCSAC first proposes a novel concept of attribute-augmented conductance, which harmoniously blends the (internal and external) structural proximity and the attribute similarity. Then, NCSAC extracts a coarse candidate community of satisfactory quality using the proposed attribute-augmented conductance. Subsequently, NCSAC frames the community search as a graph optimization task, refining the candidate community through sophisticated reinforcement learning techniques, thereby producing high-quality results. Extensive experiments on six real-world graphs and ten competitors demonstrate the superiority of our solutions in terms of accuracy, efficiency, and scalability. Notably, the proposed solution outperforms state-of-the-art methods, achieving an impressive F1-score improvement ranging from 5.3\% to 42.4\%. For reproducibility purposes, the source code is available at https://github.com/longlonglin/ncsac.

cs.SI↗

arXiv2025

Revisiting the epitaxial Si$_3$N$_4$ crystalline cap on AlGaN/GaN via evolutionary structure search

In our recent experimental work (Appl. Phys. Lett. 125, 122109 (2024)), we observed that crystalline Si$_3$N$_4$ cap layers a few nanometers thick can form in situ on GaN surfaces. Compared with amorphous SiO$_2$ and Al$_2$O$_3$ caps, these crystalline caps yield cleaner GaN/Si$_3$N$_4$ interfaces with fewer defects and improved device performance. These observations raise two questions: why does Si$_3$N$_4$ away from the interface become amorphous as the cap thickens, and what is the actual crystal structure of the interfacial Si$_3$N$_4$? Previous work proposed a defect-wurtzite (DW) model constructed heuristically from $β$-Si$_3$N$_4$ and the AlGaN lattice constants, but this model is significantly higher in energy than $β$-Si$_3$N$_4$ and disagrees with experiment in both interlayer spacings and electronic gap. Using a systematic structure-search approach under in-plane lattice constraints commensurate with AlGaN, we identify a lower-energy configuration, denoted Lam-Si$_3$N$_4$, with quasi-two-dimensional (laminar) stacking normal to the interface. Under AlGaN-matched metrics, Lam-Si$_3$N$_4$ is about 60 meV/atom more stable than DW-Si$_3$N$_4$ and reproduces the experimentally observed interlayer spacings more closely. The substantial lattice mismatch explains amorphization when the crystalline cap grows far from the interface. Upon full relaxation, both DW- and Lam-Si$_3$N$_4$ exhibit wide $\sim$4 eV band gaps. Under AlGaN constraints, the DW gap collapses to $\sim$1.88 eV whereas Lam-Si$_3$N$_4$ maintains a larger $\sim$2.70 eV gap (for reference, PBE gaps: GaN 1.73 eV, AlN 4.05 eV). The wider gap and improved structural match of Lam-Si$_3$N$_4$ rationalize the superior capping performance and provide guidance for optimizing AlGaN/GaN device encapsulation.

cond-mat.mtrl-sci↗

arXiv2025

Testing the Cosmic Distance Duality Relation with Neural Kernel Gaussian Process Regression

In this work, we test the cosmic distance duality relation (CDDR) by combining Pantheon+ Type Ia supernova (SNe Ia) data and DESI DR2 baryon acoustic oscillation (BAO) measurements. To resolve the redshift mismatch between the two datasets, we develop a new method called Neural Kernel Gaussian Process Regression (NKGPR), which uses two neural networks to simultaneously learn the mean and kernel functions of a Gaussian process. This approach improves upon traditional Gaussian process regression by mitigating trend mismatches and removing the need for manual kernel selection. We investigate possible deviations from the CDDR by adopting three parameterizations of the deviation function and constrain the model-independent parameter $η_0$ through a marginalized likelihood analysis. Our results show no significant departure from the expected relation, confirming the consistency of the CDDR within current observational uncertainties.

astro-ph.CO↗

arXiv2025

Suppressing Gradient Conflict for Generalizable Deepfake Detection

Robust deepfake detection models must be capable of generalizing to ever-evolving manipulation techniques beyond training data. A promising strategy is to augment the training data with online synthesized fake images containing broadly generalizable artifacts. However, in the context of deepfake detection, it is surprising that jointly training on both original and online synthesized forgeries may result in degraded performance. This contradicts the common belief that incorporating more source-domain data should enhance detection accuracy. Through empirical analysis, we trace this degradation to gradient conflicts during backpropagation which force a trade-off between source domain accuracy and target domain generalization. To overcome this issue, we propose a Conflict-Suppressed Deepfake Detection (CS-DFD) framework that explicitly mitigates the gradient conflict via two synergistic modules. First, an Update Vector Search (UVS) module searches for an alternative update vector near the initial gradient vector to reconcile the disparities of the original and online synthesized forgeries. By further transforming the search process into an extremum optimization problem, UVS yields the uniquely update vector, which maximizes the simultaneous loss reductions for each data type. Second, a Conflict Gradient Reduction (CGR) module enforces a low-conflict feature embedding space through a novel Conflict Descent Loss. This loss penalizes misaligned gradient directions and guides the learning of representations with aligned, non-conflicting gradients. The synergy of UVS and CGR alleviates gradient interference in both parameter optimization and representation learning. Experiments on multiple deepfake benchmarks demonstrate that CS-DFD achieves state-of-the-art performance in both in-domain detection accuracy and cross-domain generalization.

cs.CV↗

arXiv2025

Gamma-Ray Bursts Calibrated by Using Artificial Neural Networks from the Pantheon+ Sample

In this paper, we calibrate the luminosity relation of gamma-ray bursts (GRBs) by Artificial Neural Networks (ANN) which is employed to analyze the Pantheon+ sample of type Ia supernovae (SNe Ia) in a manner independent of cosmological assumptions. The A219 GRB dataset are used to calibrate the Amati relation ($E_{\rm p}$-$E_{\rm iso}$) at low redshift with the ANN framework, facilitating the construction of the Hubble diagram at higher redshifts. Cosmological models are constrained with GRBs at high-redshift and the latest observational Hubble data (OHD) via a Markov Chain Monte Carlo numerical approach. For the Chevallier-Polarski-Linder (CPL) model within a flat universe, we obtain $Ω_{\rm m} = 0.321^{+0.078}_{-0.069}$, $h = 0.654^{+0.053}_{-0.071}$, $w_0 = -1.02^{+0.67}_{-0.50}$, and $w_a = -0.98^{+0.58}_{-0.58}$ at the 1-$σ$ confidence level, which indicating a preference for dark energy with potential redshift evolution ($w_a \neq 0$). These findings by using ANN align closely with those derived from GRBs calibrated by using Gaussian Processes.

astro-ph.CO↗

arXiv2025

Dyna3DGR: 4D Cardiac Motion Tracking with Dynamic 3D Gaussian Representation

Accurate analysis of cardiac motion is crucial for evaluating cardiac function. While dynamic cardiac magnetic resonance imaging (CMR) can capture detailed tissue motion throughout the cardiac cycle, the fine-grained 4D cardiac motion tracking remains challenging due to the homogeneous nature of myocardial tissue and the lack of distinctive features. Existing approaches can be broadly categorized into image based and representation-based, each with its limitations. Image-based methods, including both raditional and deep learning-based registration approaches, either struggle with topological consistency or rely heavily on extensive training data. Representation-based methods, while promising, often suffer from loss of image-level details. To address these limitations, we propose Dynamic 3D Gaussian Representation (Dyna3DGR), a novel framework that combines explicit 3D Gaussian representation with implicit neural motion field modeling. Our method simultaneously optimizes cardiac structure and motion in a self-supervised manner, eliminating the need for extensive training data or point-to-point correspondences. Through differentiable volumetric rendering, Dyna3DGR efficiently bridges continuous motion representation with image-space alignment while preserving both topological and temporal consistency. Comprehensive evaluations on the ACDC dataset demonstrate that our approach surpasses state-of-the-art deep learning-based diffeomorphic registration methods in tracking accuracy. The code will be available in https://github.com/windrise/Dyna3DGR.

cs.CV↗

arXiv2025

Stronger, Steadier & Superior: Geometric Consistency in Depth VFM Forges Domain Generalized Semantic Segmentation

Vision Foundation Models (VFMs) have delivered remarkable performance in Domain Generalized Semantic Segmentation (DGSS). However, recent methods often overlook the fact that visual cues are susceptible, whereas the underlying geometry remains stable, rendering depth information more robust. In this paper, we investigate the potential of integrating depth information with features from VFMs, to improve the geometric consistency within an image and boost the generalization performance of VFMs. We propose a novel fine-tuning DGSS framework, named DepthForge, which integrates the visual cues from frozen DINOv2 or EVA02 and depth cues from frozen Depth Anything V2. In each layer of the VFMs, we incorporate depth-aware learnable tokens to continuously decouple domain-invariant visual and spatial information, thereby enhancing depth awareness and attention of the VFMs. Finally, we develop a depth refinement decoder and integrate it into the model architecture to adaptively refine multi-layer VFM features and depth-aware learnable tokens. Extensive experiments are conducted based on various DGSS settings and five different datsets as unseen target domains. The qualitative and quantitative results demonstrate that our method significantly outperforms alternative approaches with stronger performance, steadier visual-spatial attention, and superior generalization ability. In particular, DepthForge exhibits outstanding performance under extreme conditions (e.g., night and snow). Code is available at https://github.com/anonymouse-xzrptkvyqc/DepthForge.

cs.CV↗

arXiv2025

When Schrödinger Bridge Meets Real-World Image Dehazing with Unpaired Training

Recent advancements in unpaired dehazing, particularly those using GANs, show promising performance in processing real-world hazy images. However, these methods tend to face limitations due to the generator's limited transport mapping capability, which hinders the full exploitation of their effectiveness in unpaired training paradigms. To address these challenges, we propose DehazeSB, a novel unpaired dehazing framework based on the Schrödinger Bridge. By leveraging optimal transport (OT) theory, DehazeSB directly bridges the distributions between hazy and clear images. This enables optimal transport mappings from hazy to clear images in fewer steps, thereby generating high-quality results. To ensure the consistency of structural information and details in the restored images, we introduce detail-preserving regularization, which enforces pixel-level alignment between hazy inputs and dehazed outputs. Furthermore, we propose a novel prompt learning to leverage pre-trained CLIP models in distinguishing hazy images and clear ones, by learning a haze-aware vision-language alignment. Extensive experiments on multiple real-world datasets demonstrate our method's superiority. Code: https://github.com/ywxjm/DehazeSB.

cs.CV↗

arXiv2025

Latent Posterior-Mean Rectified Flow for Higher-Fidelity Perceptual Face Restoration

The Perception-Distortion tradeoff (PD-tradeoff) theory suggests that face restoration algorithms must balance perceptual quality and fidelity. To achieve minimal distortion while maintaining perfect perceptual quality, Posterior-Mean Rectified Flow (PMRF) proposes a flow based approach where source distribution is minimum distortion estimations. Although PMRF is shown to be effective, its pixel-space modeling approach limits its ability to align with human perception, where human perception is defined as how humans distinguish between two image distributions. In this work, we propose Latent-PMRF, which reformulates PMRF in the latent space of a variational autoencoder (VAE), facilitating better alignment with human perception during optimization. By defining the source distribution on latent representations of minimum distortion estimation, we bound the minimum distortion by the VAE's reconstruction error. Moreover, we reveal the design of VAE is crucial, and our proposed VAE significantly outperforms existing VAEs in both reconstruction and restoration. Extensive experiments on blind face restoration demonstrate the superiority of Latent-PMRF, offering an improved PD-tradeoff compared to existing methods, along with remarkable convergence efficiency, achieving a 5.79X speedup over PMRF in terms of FID. Our code will be available as open-source.

cs.CV↗

arXiv2025

StableCodec: Taming One-Step Diffusion for Extreme Image Compression

Diffusion-based image compression has shown remarkable potential for achieving ultra-low bitrate coding (less than 0.05 bits per pixel) with high realism, by leveraging the generative priors of large pre-trained text-to-image diffusion models. However, current approaches require a large number of denoising steps at the decoder to generate realistic results under extreme bitrate constraints, limiting their application in real-time compression scenarios. Additionally, these methods often sacrifice reconstruction fidelity, as diffusion models typically fail to guarantee pixel-level consistency. To address these challenges, we introduce StableCodec, which enables one-step diffusion for high-fidelity and high-realism extreme image compression with improved coding efficiency. To achieve ultra-low bitrates, we first develop an efficient Deep Compression Latent Codec to transmit a noisy latent representation for a single-step denoising process. We then propose a Dual-Branch Coding Structure, consisting of a pair of auxiliary encoder and decoder, to enhance reconstruction fidelity. Furthermore, we adopt end-to-end optimization with joint bitrate and pixel-level constraints. Extensive experiments on the CLIC 2020, DIV2K, and Kodak dataset demonstrate that StableCodec outperforms existing methods in terms of FID, KID and DISTS by a significant margin, even at bitrates as low as 0.005 bits per pixel, while maintaining strong fidelity. Additionally, StableCodec achieves inference speeds comparable to mainstream transform coding schemes. All source code are available at https://github.com/LuizScarlet/StableCodec.

eess.IV↗

arXiv2025

SDDiff: Boost Radar Perception via Spatial-Doppler Diffusion

Point cloud extraction (PCE) and ego velocity estimation (EVE) are key capabilities gaining attention in 3D radar perception. However, existing work typically treats these two tasks independently, which may neglect the interplay between radar's spatial and Doppler domain features, potentially introducing additional bias. In this paper, we observe an underlying correlation between 3D points and ego velocity, which offers reciprocal benefits for PCE and EVE. To fully unlock such inspiring potential, we take the first step to design a Spatial-Doppler Diffusion (SDDiff) model for simultaneously dense PCE and accurate EVE. To seamlessly tailor it to radar perception, SDDiff improves the conventional latent diffusion process in three major aspects. First, we introduce a representation that embodies both spatial occupancy and Doppler features. Second, we design a directional diffusion with radar priors to streamline the sampling. Third, we propose Iterative Doppler Refinement to enhance the model's adaptability to density variations and ghosting effects. Extensive evaluations show that SDDiff significantly outperforms state-of-the-art baselines by achieving 59% higher in EVE accuracy, 4X greater in valid generation density while boosting PCE effectiveness and reliability.

cs.RO↗

arXiv2025

Cross-Modal Clustering-Guided Negative Sampling for Self-Supervised Joint Learning from Medical Images and Reports

Learning medical visual representations directly from paired images and reports through multimodal self-supervised learning has emerged as a novel and efficient approach to digital diagnosis in recent years. However, existing models suffer from several severe limitations. 1) neglecting the selection of negative samples, resulting in the scarcity of hard negatives and the inclusion of false negatives; 2) focusing on global feature extraction, but overlooking the fine-grained local details that are crucial for medical image recognition tasks; and 3) contrastive learning primarily targets high-level features but ignoring low-level details which are essential for accurate medical analysis. Motivated by these critical issues, this paper presents a Cross-Modal Cluster-Guided Negative Sampling (CM-CGNS) method with two-fold ideas. First, it extends the k-means clustering used for local text features in the single-modal domain to the multimodal domain through cross-modal attention. This improvement increases the number of negative samples and boosts the model representation capability. Second, it introduces a Cross-Modal Masked Image Reconstruction (CM-MIR) module that leverages local text-to-image features obtained via cross-modal attention to reconstruct masked local image regions. This module significantly strengthens the model's cross-modal information interaction capabilities and retains low-level image features essential for downstream tasks. By well handling the aforementioned limitations, the proposed CM-CGNS can learn effective and robust medical visual representations suitable for various recognition tasks. Extensive experimental results on classification, detection, and segmentation tasks across five downstream datasets show that our method outperforms state-of-the-art approaches on multiple metrics, verifying its superior performance.

cs.CV↗