arXiv subjects

Xin Shen

Publications and source records attributed to Xin Shen.

At least 37 records · Page 2Linked to original sources

arXiv2026

Stay in Character, Stay Safe: Dual-Cycle Adversarial Self-Evolution for Safety Role-Playing Agents

LLM-based role-playing has rapidly improved in fidelity, yet stronger adherence to persona constraints commonly increases vulnerability to jailbreak attacks, especially for risky or negative personas. Most prior work mitigates this issue with training-time solutions (e.g., data curation or alignment-oriented regularization). However, these approaches are costly to maintain as personas and attack strategies evolve, can degrade in-character behavior, and are typically infeasible for frontier closed-weight LLMs. We propose a training-free Dual-Cycle Adversarial Self-Evolution framework with two coupled cycles. A Persona-Targeted Attacker Cycle synthesizes progressively stronger jailbreak prompts, while a Role-Playing Defender Cycle distills observed failures into a hierarchical knowledge base of (i) global safety rules, (ii) persona-grounded constraints, and (iii) safe in-character exemplars. At inference time, the Defender retrieves and composes structured knowledge from this hierarchy to guide generation, producing responses that remain faithful to the target persona while satisfying safety constraints. Extensive experiments across multiple proprietary LLMs show consistent gains over strong baselines on both role fidelity and jailbreak resistance, and robust generalization to unseen personas and attack prompts.

cs.AI↗

arXiv2026

$\mathrm{D}^\mathrm{3}$-Predictor: Noise-Free Deterministic Diffusion for Dense Prediction

Although diffusion models with strong visual priors have emerged as powerful dense prediction backbones, they overlook a core limitation: the stochastic noise at the core of diffusion sampling is inherently misaligned with dense prediction that requires a deterministic mapping from image to geometry. In this paper, we show that this stochastic noise corrupts fine-grained spatial cues and pushes the model toward timestep-specific noise objectives, consequently destroying meaningful geometric structure mappings. To address this, we introduce $\mathrm{D}^\mathrm{3}$-Predictor, a noise-free deterministic diffusion-based dense prediction model built by reformulating a pretrained diffusion model without stochasticity noise. Instead of relying on noisy inputs to leverage diffusion priors, $\mathrm{D}^\mathrm{3}$-Predictor views the pretrained diffusion network as an ensemble of timestep-dependent visual experts and self-supervisedly aggregates their heterogeneous priors into a single, clean, and complete geometric prior. Meanwhile, we utilize task-specific supervision to seamlessly adapt this noise-free prior to dense prediction tasks. Extensive experiments on various dense prediction tasks demonstrate that $\mathrm{D}^\mathrm{3}$-Predictor achieves competitive or state-of-the-art performance in diverse scenarios. In addition, it requires less than half the training data previously used and efficiently performs inference in a single step. Our code, data, and checkpoints are publicly available at https://x-gengroup.github.io/HomePage_D3-Predictor/.

cs.CV↗

arXiv2026

Video-MSR: Benchmarking Multi-hop Spatial Reasoning Capabilities of MLLMs

Spatial reasoning has emerged as a critical capability for Multimodal Large Language Models (MLLMs), drawing increasing attention and rapid advancement. However, existing benchmarks primarily focus on single-step perception-to-judgment tasks, leaving scenarios requiring complex visual-spatial logical chains significantly underexplored. To bridge this gap, we introduce Video-MSR, the first benchmark specifically designed to evaluate Multi-hop Spatial Reasoning (MSR) in dynamic video scenarios. Video-MSR systematically probes MSR capabilities through four distinct tasks: Constrained Localization, Chain-based Reference Retrieval, Route Planning, and Counterfactual Physical Deduction. Our benchmark comprises 3,052 high-quality video instances with 4,993 question-answer pairs, constructed via a scalable, visually-grounded pipeline combining advanced model generation with rigorous human verification. Through a comprehensive evaluation of 20 state-of-the-art MLLMs, we uncover significant limitations, revealing that while models demonstrate proficiency in surface-level perception, they exhibit distinct performance drops in MSR tasks, frequently suffering from spatial disorientation and hallucination during multi-step deductions. To mitigate these shortcomings and empower models with stronger MSR capabilities, we further curate MSR-9K, a specialized instruction-tuning dataset, and fine-tune Qwen-VL, achieving a +7.82% absolute improvement on Video-MSR. Our results underscore the efficacy of multi-hop spatial instruction data and establish Video-MSR as a vital foundation for future research. The code and data will be available at https://github.com/ruiz-nju/Video-MSR.

cs.CV↗

arXiv2025

Fast and Robust T1 Mapping Based on a 3D Dual-Echo UTE Sequence (PETALUTE) for SPION Biodistribution Assessment

Superparamagnetic iron oxide nanoparticles (SPIONs) such as ferumoxytol are promising theranostic agents detectable with MRI. Relaxation time mapping offers reproducible, quantitative biomarkers of SPION distribution, but conventional methods suffer from susceptibility artifacts, long echo times, and extended scan durations, limiting accurate quantification. This study developed a fast, B1-corrected T1-mapping protocol using PETALUTE, a 3D dual-echo ultrashort-echo MRI sequence with a rosette k-space trajectory and variable flip-angle acquisition for quantitative ferumoxytol imaging. Agarose phantoms containing 0-5000 ppm ferumoxytol were scanned at 7T with PETALUTE and vendor-supplied RARE-VTR. PETALUTE T1 maps were derived from two flip angles (4 deg and 20 deg), and mean R1 values were correlated with ferumoxytol concentration. For in vivo feasibility, mice bearing 4T1 mammary and flank tumors were scanned 24 h post-injection (ferumoxytol: n=2, 40 mg/kg; control: n=1). Regions of interest in muscle and tumors were analyzed to compare T1 and R1 values obtained with both methods. PETALUTE produced positive contrast for all phantom concentrations except 5000 ppm, whereas RARE-VTR did not. PETALUTE demonstrated a significant linear correlation between R1 and ferumoxytol concentration (R=0.975, p<0.01), in contrast to RARE-VTR (R=0.672, p=0.144). In vivo, PETALUTE enabled high-resolution, whole-abdominal imaging in 4 min 19 s. Ferumoxytol-injected mice showed T1 shortening in flank tumors, consistent with iron uptake, and PETALUTE revealed elevated T1 value with preserved T2*-weighted signal in one mammary tumor. PETALUTE-based T1 mapping provides fast, quantitative, positive-contrast ferumoxytol imaging with greater spatial coverage and a wider usable concentration range than conventional RARE-VTR.

physics.med-ph↗

arXiv2025

FingerCap: Fine-grained Finger-level Hand Motion Captioning

Understanding fine-grained human hand motion is fundamental to visual perception, embodied intelligence, and multimodal communication. In this work, we propose Fine-grained Finger-level Hand Motion Captioning (FingerCap), which aims to generate textual descriptions that capture detailed finger-level semantics of hand actions. To support this task, we curate FingerCap-40K, a large-scale corpus of 40K paired hand-motion videos and captions spanning two complementary sources: concise instruction-style finger motions and diverse, naturalistic hand-object interactions. To enable effective evaluation, we employ HandJudge, a LLM-based rubric that measures finger-level correctness and motion completeness. Temporal sparsity remains a fundamental bottleneck for current Video-MLLMs, since sparse RGB sampling is insufficient to capture the subtle, high-frequency dynamics underlying fine finger motions. As a simple and compute-friendly remedy, we introduce FiGOP (Finger Group-of-Pictures), which pairs each RGB keyframe with subsequent hand keypoints until the next keyframe. A lightweight temporal encoder converts the keypoints into motion embeddings and integrates them with RGB features. FiGOP adapts the classic GOP concept to finger motion, recovering fine temporal cues without increasing RGB density. Experiments on FingerCap-40K show that strong open- and closed-source Video-MLLMs still struggle with finger-level reasoning, while our FiGOP-augmented model yield consistent gains under HandJudge and human studies.

cs.CV↗

arXiv2025

ProofFlow: A Dependency Graph Approach to Faithful Proof Autoformalization

Proof autoformalization, the task of translating natural language theorems and proofs into machine-verifiable code, is a critical step for integrating large language models into rigorous mathematical workflows. Current approaches focus on producing executable code, but they frequently fail to preserve the semantic meaning and logical structure of the original human-written argument. To address this, we introduce ProofFlow, a novel pipeline that treats structural fidelity as a primary objective. ProofFlow first constructs a directed acyclic graph (DAG) to map the logical dependencies between proof steps. Then, it employs a novel lemma-based approach to systematically formalize each step as an intermediate lemma, preserving the logical structure of the original argument. To facilitate evaluation, we present a new benchmark of 184 undergraduate-level problems, manually annotated with step-by-step solutions and logical dependency graphs, and introduce ProofScore, a new composite metric to evaluate syntactic correctness, semantic faithfulness, and structural fidelity. Experimental results show our pipeline sets a new state-of-the-art for autoformalization, achieving a ProofScore of 0.545, substantially exceeding baselines like full-proof formalization (0.123), which processes the entire proof at once, and step-proof formalization (0.072), which handles each step independently. Our pipeline, benchmark, and score metric are open-sourced to encourage further progress at https://github.com/Huawei-AI4Math/ProofFlow.

cs.AI↗

arXiv2025

Learning Domain-Robust Bioacoustic Representations for Mosquito Species Classification with Contrastive Learning and Distribution Alignment

Mosquito Species Classification (MSC) is crucial for vector surveillance and disease control. The collection of mosquito bioacoustic data is often limited by mosquito activity seasons and fieldwork. Mosquito recordings across regions, habitats, and laboratories often show non-biological variations from the recording environment, which we refer to as domain features. This study finds that models directly trained on audio recordings with domain features tend to rely on domain information rather than the species' acoustic cues for identification, resulting in illusory good performance while actually performing poor cross-domain generalization. To this end, we propose a Domain-Robust Bioacoustic Learning (DR-BioL) framework that combines contrastive learning with distribution alignment. Contrastive learning aims to promote cohesion within the same species and mitigate inter-domain discrepancies, and species-conditional distribution alignment further enhances cross-domain species representation. Experiments on a multi-domain mosquito bioacoustic dataset from diverse environments show that the DR-BioL improves the accuracy and robustness of baselines, highlighting its potential for reliable cross-domain MSC in the real world.

eess.AS↗

arXiv2025

Why Settle for One? Text-to-ImageSet Generation and Evaluation

Despite remarkable progress in Text-to-Image models, many real-world applications require generating coherent image sets with diverse consistency requirements. Existing consistent methods often focus on a specific domain with specific aspects of consistency, which significantly constrains their generalizability to broader applications. In this paper, we propose a more challenging problem, Text-to-ImageSet (T2IS) generation, which aims to generate sets of images that meet various consistency requirements based on user instructions. To systematically study this problem, we first introduce $\textbf{T2IS-Bench}$ with 596 diverse instructions across 26 subcategories, providing comprehensive coverage for T2IS generation. Building on this, we propose $\textbf{T2IS-Eval}$, an evaluation framework that transforms user instructions into multifaceted assessment criteria and employs effective evaluators to adaptively assess consistency fulfillment between criteria and generated sets. Subsequently, we propose $\textbf{AutoT2IS}$, a training-free framework that maximally leverages pretrained Diffusion Transformers' in-context capabilities to harmonize visual elements to satisfy both image-level prompt alignment and set-level visual consistency. Extensive experiments on T2IS-Bench reveal that diverse consistency challenges all existing methods, while our AutoT2IS significantly outperforms current generalized and even specialized approaches. Our method also demonstrates the ability to enable numerous underexplored real-world applications, confirming its substantial practical value. Visit our project in https://chengyou-jia.github.io/T2IS-Home.

cs.CV↗

arXiv2025

On the accurate computation of expected modularity in probabilistic networks

Modularity is one of the most widely used measures for evaluating communities in networks. In probabilistic networks, where the existence of edges is uncertain and uncertainty is represented by probabilities, the expected value of modularity can be used instead. However, efficiently computing expected modularity is challenging. To address this challenge, we propose a novel and efficient technique (FPWP) for computing the probability distribution of modularity and its expected value. In this paper, we implement and compare our method and various general approaches for expected modularity computation in probabilistic networks. These include: (1) translating probabilistic networks into deterministic ones by removing low-probability edges or treating probabilities as weights, (2) using Monte Carlo sampling to approximate expected modularity, and (3) brute-force computation. We evaluate the accuracy and time efficiency of FPWP through comprehensive experiments on both real-world and synthetic networks with diverse characteristics. Our results demonstrate that removing low-probability edges or treating probabilities as weights produces inaccurate results, while the convergence of the sampling method varies with the parameters of the network. Brute-force computation, though accurate, is prohibitively slow. In contrast, our method is much faster than brute-force computation, but guarantees an accurate result.

cs.DS↗

arXiv2025

Trust-Aware Diversion for Data-Effective Distillation

Dataset distillation compresses a large dataset into a small synthetic subset that retains essential information. Existing methods assume that all samples are perfectly labeled, limiting their real-world applications where incorrect labels are ubiquitous. These mislabeled samples introduce untrustworthy information into the dataset, which misleads model optimization in dataset distillation. To tackle this issue, we propose a Trust-Aware Diversion (TAD) dataset distillation method. Our proposed TAD introduces an iterative dual-loop optimization framework for data-effective distillation. Specifically, the outer loop divides data into trusted and untrusted spaces, redirecting distillation toward trusted samples to guarantee trust in the distillation process. This step minimizes the impact of mislabeled samples on dataset distillation. The inner loop maximizes the distillation objective by recalibrating untrusted samples, thus transforming them into valuable ones for distillation. This dual-loop iteratively refines and compensates for each other, gradually expanding the trusted space and shrinking the untrusted space. Experiments demonstrate that our method can significantly improve the performance of existing dataset distillation methods on three widely used benchmarks (CIFAR10, CIFAR100, and Tiny ImageNet) in three challenging mislabeled settings (symmetric, asymmetric, and real-world).

cs.CV↗

arXiv2025

Exact Parent Hamiltonians for All Landau Level States in a Half-flux Lattice

Realizing topological flat bands with tailored single-particle Hilbert spaces is a critical step toward exploring many-body phases, such as those featuring anyonic excitations. One prominent example is the Kapit-Mueller model, a variant of the Harper-Hofstadter model that stabilizes lattice analogs of the lowest Landau level states. The Kapit-Mueller model is constructed based on the Poisson summation rule, an exact lattice sum rule for coherent states. In this work, we consider higher Landau-level generalizations of the Poisson summation rule, from which we derive families of parent Hamiltonians on a half-flux lattice which have exact flat bands whose flatband wavefunctions are lattice version of higher Landau level states. Focusing on generic Bravais lattices with only translation and inversion symmetries, we discuss how these symmetries enforced gaplessness and singular points for odd Landau level series, and how to achieve fully gapped parent Hamiltonians by mixing even and odd series. Our model points to a large class of tight-binding models with suitable energetic and quantum geometries that are potentially useful for realizing non-Abelian fractionalized states when interactions are included. The model exhibits fast decay hopping amplitudes, making it potentially realizable with neutral atoms in optical lattices.

cond-mat.mes-hall↗

arXiv2024

MM-WLAuslan: Multi-View Multi-Modal Word-Level Australian Sign Language Recognition Dataset

Isolated Sign Language Recognition (ISLR) focuses on identifying individual sign language glosses. Considering the diversity of sign languages across geographical regions, developing region-specific ISLR datasets is crucial for supporting communication and research. Auslan, as a sign language specific to Australia, still lacks a dedicated large-scale word-level dataset for the ISLR task. To fill this gap, we curate \underline{\textbf{the first}} large-scale Multi-view Multi-modal Word-Level Australian Sign Language recognition dataset, dubbed MM-WLAuslan. Compared to other publicly available datasets, MM-WLAuslan exhibits three significant advantages: (1) the largest amount of data, (2) the most extensive vocabulary, and (3) the most diverse of multi-modal camera views. Specifically, we record 282K+ sign videos covering 3,215 commonly used Auslan glosses presented by 73 signers in a studio environment. Moreover, our filming system includes two different types of cameras, i.e., three Kinect-V2 cameras and a RealSense camera. We position cameras hemispherically around the front half of the model and simultaneously record videos using all four cameras. Furthermore, we benchmark results with state-of-the-art methods for various multi-modal ISLR settings on MM-WLAuslan, including multi-view, cross-camera, and cross-view. Experiment results indicate that MM-WLAuslan is a challenging ISLR dataset, and we hope this dataset will contribute to the development of Auslan and the advancement of sign languages worldwide. All datasets and benchmarks are available at MM-WLAuslan.

cs.CV↗

arXiv2024

Diverse Sign Language Translation

Like spoken languages, a single sign language expression could correspond to multiple valid textual interpretations. Hence, learning a rigid one-to-one mapping for sign language translation (SLT) models might be inadequate, particularly in the case of limited data. In this work, we introduce a Diverse Sign Language Translation (DivSLT) task, aiming to generate diverse yet accurate translations for sign language videos. Firstly, we employ large language models (LLM) to generate multiple references for the widely-used CSL-Daily and PHOENIX14T SLT datasets. Here, native speakers are only invited to touch up inaccurate references, thus significantly improving the annotation efficiency. Secondly, we provide a benchmark model to spur research in this task. Specifically, we investigate multi-reference training strategies to enable our DivSLT model to achieve diverse translations. Then, to enhance translation accuracy, we employ the max-reward-driven reinforcement learning objective that maximizes the reward of the translated result. Additionally, we utilize multiple metrics to assess the accuracy, diversity, and semantic precision of the DivSLT task. Experimental results on the enriched datasets demonstrate that our DivSLT method achieves not only better translation performance but also diverse translation results.

cs.MM↗

arXiv2023

Divide and Ensemble: Progressively Learning for the Unknown

In the wheat nutrient deficiencies classification challenge, we present the DividE and EnseMble (DEEM) method for progressive test data predictions. We find that (1) test images are provided in the challenge; (2) samples are equipped with their collection dates; (3) the samples of different dates show notable discrepancies. Based on the findings, we partition the dataset into discrete groups by the dates and train models on each divided group. We then adopt the pseudo-labeling approach to label the test data and incorporate those with high confidence into the training set. In pseudo-labeling, we leverage models ensemble with different architectures to enhance the reliability of predictions. The pseudo-labeling and ensembled model training are iteratively conducted until all test samples are labeled. Finally, the separated models for each group are unified to obtain the model for the whole dataset. Our method achieves an average of 93.6\% Top-1 test accuracy~(94.0\% on WW2020 and 93.2\% on WR2021) and wins the 1$st$ place in the Deep Nutrient Deficiency Challenge~\footnote{https://cvppa2023.github.io/challenges/}.

cs.AI↗

arXiv2023

Strongly interacting Bose-Fermi mixture: mediated interaction, phase diagram and sound propagation

Motivated by recent surprising experimental findings, we develop a strong-coupling theory for Bose-Fermi mixtures capable of treating resonant inter-species interactions while satisfying the compressibility sum rule. We show that the mixture can be stable at large interaction strengths close to resonance, in agreement with the experiment but at odds with the widely used perturbation theory. We also calculate the sound velocity of the Bose gas in the $^{133}$Cs-$^6$Li mixture, again finding good agreement with the experimental observations both at weak and strong interactions. A central ingredient of our theory is the generalization of a fermion mediated interaction to strong Bose-Fermi scatterings and to finite frequencies. This further leads to a predicted hybridization of the sound modes of the Bose and Fermi gases, which can be directly observed using Bragg spectroscopy.

cond-mat.quant-gas↗

arXiv2023

FishRecGAN: An End to End GAN Based Network for Fisheye Rectification and Calibration

We propose an end-to-end deep learning approach to rectify fisheye images and simultaneously calibrate camera intrinsic and distortion parameters. Our method consists of two parts: a Quick Image Rectification Module developed with a Pix2Pix GAN and Wasserstein GAN (W-Pix2PixGAN), and a Calibration Module with a CNN architecture. Our Quick Rectification Network performs robust rectification with good resolution, making it suitable for constant calibration in camera-based surveillance equipment. To achieve high-quality calibration, we use the straightened output from the Quick Rectification Module as a guidance-like semantic feature map for the Calibration Module to learn the geometric relationship between the straightened feature and the distorted feature. We train and validate our method with a large synthesized dataset labeled with well-simulated parameters applied to a perspective image dataset. Our solution has achieved robust performance in high-resolution with a significant PSNR value of 22.343.

cs.CV↗

arXiv2023

G-STO: Sequential Main Shopping Intention Detection via Graph-Regularized Stochastic Transformer

Sequential recommendation requires understanding the dynamic patterns of users' behaviors, contexts, and preferences from their historical interactions. Most existing works focus on modeling user-item interactions only from the item level, ignoring that they are driven by latent shopping intentions (e.g., ballpoint pens, miniatures, etc). The detection of the underlying shopping intentions of users based on their historical interactions is a crucial aspect for e-commerce platforms, such as Amazon, to enhance the convenience and efficiency of their customers' shopping experiences. Despite its significance, the area of main shopping intention detection remains under-investigated in the academic literature. To fill this gap, we propose a graph-regularized stochastic Transformer method, G-STO. By considering intentions as sets of products and user preferences as compositions of intentions, we model both of them as stochastic Gaussian embeddings in the latent representation space. Instead of training the stochastic representations from scratch, we develop a global intention relational graph as prior knowledge for regularization, allowing relevant shopping intentions to be distributionally close. Finally, we feed the newly regularized stochastic embeddings into Transformer-based models to encode sequential information from the intention transitions. We evaluate our main shopping intention identification model on three different real-world datasets, where G-STO achieves significantly superior performances to the baselines by 18.08% in Hit@1, 7.01% in Hit@10, and 6.11% in NDCG@10 on average.

cs.IR↗

arXiv2023

Data Efficient Training with Imbalanced Label Sample Distribution for Fashion Detection

Multi-label classification models have a wide range of applications in E-commerce, including visual-based label predictions and language-based sentiment classifications. A major challenge in achieving satisfactory performance for these tasks in the real world is the notable imbalance in data distribution. For instance, in fashion attribute detection, there may be only six 'puff sleeve' clothes among 1000 products in most E-commerce fashion catalogs. To address this issue, we explore more data-efficient model training techniques rather than acquiring a huge amount of annotations to collect sufficient samples, which is neither economic nor scalable. In this paper, we propose a state-of-the-art weighted objective function to boost the performance of deep neural networks (DNNs) for multi-label classification with long-tailed data distribution. Our experiments involve image-based attribute classification of fashion apparels, and the results demonstrate favorable performance for the new weighting method compared to non-weighted and inverse-frequency-based weighting mechanisms. We further evaluate the robustness of the new weighting mechanism using two popular fashion attribute types in today's fashion industry: sleevetype and archetype.

cs.CV↗