arXiv subjects

Lijun Wang

Publications and source records attributed to Lijun Wang.

At least 19 recordsLinked to original sources

arXiv2026

MA-VLA: Multi-Arm Vision-Language-Action Model for Collaboration and Compositional Generalization

Multi-arm collaboration is becoming a core capability in embodied manipulation. Recent vision-language-action (VLA) models integrate perception, language, and control, but most represent language as a single global instruction and do not provide an explicit mechanism for assigning and composing arm-specific behaviors. This design limits transfer to collaboration patterns that differ from those observed during training. We present MA-VLA, a unified framework for multi-arm collaboration via atomic action assignment. MA-VLA decomposes cooperative behavior into mid-level atomic prompts and allocates them to individual arms, enabling explicit subgoal specification and compositional reuse across tasks. To reduce reliance on fixed execution roles, we introduce Arm Shuffle, a training-time permutation of the observation, state, and assigned atomic prompts for each arm. This permutation enforces role-agnostic instruction following and supports recomposition into unseen coordination patterns, which we term multi-arm compositional generalization. We also construct a benchmark in which test-time collaboration patterns are absent in training set. Across simulation and real-world evaluations, prior state-of-the-art VLAs largely fail under these unseen collaborations, while MA-VLA consistently succeeds. These results indicate that structured, per-arm atomic action assignment offers a practical route to scalable generalization in multi-arm embodied systems. Code, models, and data are available at https://github.com/zhangzaibin/future-robots

cs.RO↗

arXiv2026

SAGE: An LLM-driven Self Reflective Agentic Framework for Fraud Detection

Fraud detection in payment, e-commerce, and telecommunications systems requires accuracy at the individual level, robustness under severe class imbalance, and ease of understanding for risk managers. Existing methods fall at least one of these requirements: automated machine learning systems search a fixed numerical space without semantic awareness of the dataset; graph neural network-based methods require pre-defined relational graphs and remain opaque at the individual-decision level; and the design of general-purpose large language model (LLM) agents does not consider the recall and precision constraints specific to real-world fraud detection. In this paper, we propose SAGE, the first end-to-end LLM-driven multi-agent framework for fraud detection. SAGE coordinates three dedicated agents that make decisions based on a six-layer Data Diagnostic Tree (DDT) and a Markov decision process guided by natural-language gradients, automatically optimizing the model under a fraud-specific reward. On five fraud datasets and five LLM backbones, SAGE wins $96.00\%$ of method--dataset comparisons and improves F1 by an average of $40.86\%$ over baselines. The code is available at https://github.com/yichenC1c/SAGE.

cs.AI↗

arXiv2026

Revisiting Salient Object Detection from an Observer-Centric Perspective

Salient object detection is inherently a subjective problem, as observers with different priors may perceive different objects as salient. However, existing methods predominantly formulate it as an objective prediction task with a single groundtruth segmentation map for each image, which renders the problem under-determined and fundamentally ill-posed. To address this issue, we propose Observer-Centric Salient Object Detection (OC-SOD), where salient regions are predicted by considering not only the visual cues but also the observer-specific factors such as their preferences or intents. As a result, this formulation captures the intrinsic ambiguity and diversity of human perception, enabling personalized and context-aware saliency prediction. By leveraging multi-modal large language models, we develop an efficient data annotation pipeline and construct the first OC-SOD dataset named OC-SODBench, comprising 33k training, validation and test images with 152k textual prompts and object pairs. Built upon this new dataset, we further design OC-SODAgent, an agentic baseline which performs OC-SOD via a human-like "Perceive-Reflect-Adjust" process. Extensive experiments on our proposed OC-SODBench have justified the effectiveness of our contribution. Through this observer-centric perspective, we aim to bridge the gap between human perception and computational modeling, offering a more realistic and flexible understanding of what makes an object truly "salient." Code and dataset are publicly available at: https://github.com/Dustzx/OC_SOD

cs.CV↗

arXiv2026

Think3D: Thinking with Space for Spatial Reasoning

While Vision-Language Models (VLMs) excel at 2D visual understanding, they remain constrained by 2D-centric paradigm that severely limits genuine 3D spatial reasoning. To bridge this gap, we introduce Think3D, a novel framework that equips VLM agents with interactive, 3D chain-of-thought reasoning capabilities. By integrating a suite of 3D manipulation tools, Think3D transforms perception into active spatial exploration, mirroring human geometric reasoning. Think3D consistently improves proprietary models, including GPT-4.1 and Gemini 2.5 Pro, across BLINK Multi-view, MindCube-1K, and VSI-Bench-Tiny. We further propose Think3D-RL to teach smaller open-weight models how to manipulate 3D space effectively. Using only final-answer rewards, without process supervision or handcrafted exploration trajectories, Think3D-RL enables Qwen3-VL-4B to autonomously learn effective 3D exploration strategies. After training, the model exhibits tool-use patterns similar to those of stronger proprietary models, while shifting the effect of 3D tool use on MindCube-1K from a performance drop to a substantial improvement. These results show that active exploration in 3D space provides an effective and general paradigm for improving spatial reasoning in multimodal agents. Code, models, and data are available at https://github.com/zhangzaibin/spagent.

cs.CV↗

arXiv2026

AR-MOT: Autoregressive Multi-object Tracking

As multi-object tracking (MOT) tasks continue to evolve toward more general and multi-modal scenarios, the rigid and task-specific architectures of existing MOT methods increasingly hinder their applicability across diverse tasks and limit flexibility in adapting to new tracking formulations. Most approaches rely on fixed output heads and bespoke tracking pipelines, making them difficult to extend to more complex or instruction-driven tasks. To address these limitations, we propose AR-MOT, a novel autoregressive paradigm that formulates MOT as a sequence generation task within a large language model (LLM) framework. This design enables the model to output structured results through flexible sequence construction, without requiring any task-specific heads. To enhance region-level visual perception, we introduce an Object Tokenizer based on a pretrained detector. To mitigate the misalignment between global and regional features, we propose a Region-Aware Alignment (RAA) module, and to support long-term tracking, we design a Temporal Memory Fusion (TMF) module that caches historical object tokens. AR-MOT offers strong potential for extensibility, as new modalities or instructions can be integrated by simply modifying the output sequence format without altering the model architecture. Extensive experiments on MOT17 and DanceTrack validate the feasibility of our approach, achieving performance comparable to state-of-the-art methods while laying the foundation for more general and flexible MOT systems.

cs.CV↗

arXiv2026

Improve Power of Knockoffs with Annotation Information of Covariates

Genome-wide association studies (GWAS) often find association signals between many genetic variants and traits of interest in a genomic region. Functional annotations of these variants provide valuable prior information that helps prioritize biologically relevant variants and enhances the power to detect causal variants. However, due to substantial correlations among these variants, a critical question is how to rigorously control the false discovery rate while effectively leveraging prior knowledge. We introduce annotation-informed knockoffs (AnnoKn), a knockoff-based method that performs annotation-informed variable selection with strict control of the false discovery rate. AnnoKn integrates the knockoff procedure with adaptive Lasso regression to evaluate the importance of multiple covariates while incorporating functional annotation information within a unified Bayesian framework. To facilitate real-world applications where individual-level data are not accessible, we further extend AnnoKn to operate on summary statistics. Through simulations and real-world applications to GTEx and GWAS datasets, we show that AnnoKn achieves superior power in detecting causal genetic variants compared with existing annotation-informed variable selection methods, while maintaining valid control over false discoveries.

stat.ME↗

arXiv2025

Optimal Labeler Assignment and Sampling for Active Learning in the Presence of Imperfect Labels

Active Learning (AL) has garnered significant interest across various application domains where labeling training data is costly. AL provides a framework that helps practitioners query informative samples for annotation by oracles (labelers). However, these labels often contain noise due to varying levels of labeler accuracy. Additionally, uncertain samples are more prone to receiving incorrect labels because of their complexity. Learning from imperfectly labeled data leads to an inaccurate classifier. We propose a novel AL framework to construct a robust classification model by minimizing noise levels. Our approach includes an assignment model that optimally assigns query points to labelers, aiming to minimize the maximum possible noise within each cycle. Additionally, we introduce a new sampling method to identify the best query points, reducing the impact of label noise on classifier performance. Our experiments demonstrate that our approach significantly improves classification performance compared to several benchmark methods.

cs.LG↗

arXiv2025

Suspended waveguide-enhanced near-infrared photothermal spectroscopy for ppb-level molecular gas sensing on a chalcogenide chip

On-chip waveguide sensors have attracted significant attention recently due to their potential for high level integration. However, so far on-chip gas sensing based on traditional laser absorption spectroscopy has demonstrated low detection sensitivity, due to weak light-gas interaction over a limited interaction distance. On-chip photothermal spectroscopy (PTS) appears to be a powerful technique to achieve higher sensitivity, its performance is yet constrained to parts-per-million (ppm)-level due to small fraction of evanescent field in the light-gas interaction zone and fast thermal dissipation through the solid substrate. Herein, we demonstrated suspended chalcogenide glass waveguide (ChGW)-enhanced PTS that overcomes these limitations, enabling highly sensitive parts-per-billion (ppb)-level molecular gas sensing. We fabricated a nanoscale suspended ChGW with low loss of 2.6 dB/cm using CMOS-compatible two-step patterning process. By establishing an equivalent PTS model to guide the optimization of the ChGW geometry, we achieved a 4-fold increase in the absorption-induced heat source power and a 10.6-fold decrease in the equivalent heat conductivity, resulting in a 45-fold enhancement in photothermal phase modulation efficiency over the non-suspended waveguides. Combining with a high-contrast waveguide facet-formed Fabry-Perot interferometer, we achieved an unprecedented acetylene detection limit of 330 ppb, a large dynamic range close to 6 orders of magnitude, and a fast response of less than 1 s. The overall system exhibits a noise-equivalent absorption coefficient of 3.8x10-7 cm-1, setting a new benchmark for photonic waveguide gas sensors to the best of our knowledge. This work provides a key advancement towards prototyping an integrated sensor-on-a-chip for highly sensitive and background-free photonic sensing applications.

physics.optics↗

arXiv2025

From Forecasting to Planning: Policy World Model for Collaborative State-Action Prediction

Despite remarkable progress in driving world models, their potential for autonomous systems remains largely untapped: the world models are mostly learned for world simulation and decoupled from trajectory planning. While recent efforts aim to unify world modeling and planning in a single framework, the synergistic facilitation mechanism of world modeling for planning still requires further exploration. In this work, we introduce a new driving paradigm named Policy World Model (PWM), which not only integrates world modeling and trajectory planning within a unified architecture, but is also able to benefit planning using the learned world knowledge through the proposed action-free future state forecasting scheme. Through collaborative state-action prediction, PWM can mimic the human-like anticipatory perception, yielding more reliable planning performance. To facilitate the efficiency of video forecasting, we further introduce a dynamically enhanced parallel token generation mechanism, equipped with a context-guided tokenizer and an adaptive dynamic focal loss. Despite utilizing only front camera input, our method matches or exceeds state-of-the-art approaches that rely on multi-view and multi-modal inputs. Code and model weights will be released at https://github.com/6550Zhao/Policy-World-Model.

cs.CV↗

arXiv2025

How Far are VLMs from Visual Spatial Intelligence? A Benchmark-Driven Perspective

Visual Spatial Reasoning (VSR) is a core human cognitive ability and a critical requirement for advancing embodied intelligence and autonomous systems. Despite recent progress in Vision-Language Models (VLMs), achieving human-level VSR remains highly challenging due to the complexity of representing and reasoning over three-dimensional space. In this paper, we present a systematic investigation of VSR in VLMs, encompassing a review of existing methodologies across input modalities, model architectures, training strategies, and reasoning mechanisms. Furthermore, we categorize spatial intelligence into three levels of capability, ie, basic perception, spatial understanding, spatial planning, and curate SIBench, a spatial intelligence benchmark encompassing nearly 20 open-source datasets across 23 task settings. Experiments with state-of-the-art VLMs reveal a pronounced gap between perception and reasoning, as models show competence in basic perceptual tasks but consistently underperform in understanding and planning tasks, particularly in numerical estimation, multi-view reasoning, temporal dynamics, and spatial imagination. These findings underscore the substantial challenges that remain in achieving spatial intelligence, while providing both a systematic roadmap and a comprehensive benchmark to drive future research in the field. The related resources of this study are accessible at https://sibench.github.io/Awesome-Visual-Spatial-Reasoning/.

cs.AI↗

arXiv2025

Fiber-integrated NV Magnetometer with Microcontroller-based Software Lock-in Technique

Fiber-integrated nitrogen-vacancy (NV) magnetometers possess high sensitivity, integration, and flexibility, and thus have been explored extensively for industrial applications. While most studies have focused on the optimization of the quantum sensing head, less attention has been paid to the frequently employed professional, expensive, and bulky electronics, which hinder their practical applications. In this article, we fabricate a fiber-integrated NV magnetometer and develop a low-cost microcontroller-based software lock-in technique. In this technique, a microcontroller coordinates efficiently a microwave source chip and an analog-to-digital converter, and a program mimicking the lock-in mechanism realizes microwave frequency-modulated optically detected magnetic resonance of NV centers. As a result, with our setup and technique, we have realized the detection of weak magnetic field with a sensitivity of 93 nT/Hz^{1/2}, which is comparable to what obtained with bulky and professional devices. Furthermore, we demonstrated real-time magnetic field detection, achieving a standard deviation of 488 nT. Our work provides a novel and cost-effective technique for electronic miniaturization, thereby potentially accelerating the industrial application of NV magnetometers.

quant-ph↗

arXiv2025

SR-SLAM: Scene-reliability Based RGB-D SLAM in Diverse Environments

Visual simultaneous localization and mapping (SLAM) plays a critical role in autonomous robotic systems, especially where accurate and reliable measurements are essential for navigation and sensing. In feature-based SLAM, the quantityand quality of extracted features significantly influence system performance. Due to the variations in feature quantity and quality across diverse environments, current approaches face two major challenges: (1) limited adaptability in dynamic feature culling and pose estimation, and (2) insufficient environmental awareness in assessment and optimization strategies. To address these issues, we propose SRR-SLAM, a scene-reliability based framework that enhances feature-based SLAM through environment-aware processing. Our method introduces a unified scene reliability assessment mechanism that incorporates multiple metrics and historical observations to guide system behavior. Based on this assessment, we develop: (i) adaptive dynamic region selection with flexible geometric constraints, (ii) depth-assisted self-adjusting clustering for efficient dynamic feature removal in high-dimensional settings, and (iii) reliability-aware pose refinement that dynamically integrates direct methods when features are insufficient. Furthermore, we propose (iv) reliability-based keyframe selection and a weighted optimization scheme to reduce computational overhead while improving estimation accuracy. Extensive experiments on public datasets and real world scenarios show that SRR-SLAM outperforms state-of-the-art dynamic SLAM methods, achieving up to 90% improvement in accuracy and robustness across diverse environments. These improvements directly contribute to enhanced measurement precision and reliability in autonomous robotic sensing systems.

cs.RO↗

arXiv2025

GLDesigner: Leveraging Multi-Modal LLMs as Designer for Enhanced Aesthetic Text Glyph Layouts

Text logo design heavily relies on the creativity and expertise of professional designers, in which arranging element layouts is one of the most important procedures. However, this specific task has received limited attention, often overshadowed by broader layout generation tasks such as document or poster design. In this paper, we propose a Vision-Language Model (VLM)-based framework that generates content-aware text logo layouts by integrating multi-modal inputs with user-defined constraints, enabling more flexible and robust layout generation for real-world applications. We introduce two model techniques that reduce the computational cost for processing multiple glyph images simultaneously, without compromising performance. To support instruction tuning of our model, we construct two extensive text logo datasets that are five times larger than existing public datasets. In addition to geometric annotations (\textit{e.g.}, text masks and character recognition), our datasets include detailed layout descriptions in natural language, enabling the model to reason more effectively in handling complex designs and custom user inputs. Experimental results demonstrate the effectiveness of our proposed framework and datasets, outperforming existing methods on various benchmarks that assess geometric aesthetics and human preferences.

cs.CV↗

arXiv2025

Online 3D Bin Packing with Fast Stability Validation and Stable Rearrangement Planning

The Online Bin Packing Problem (OBPP) is a sequential decision-making task in which each item must be placed immediately upon arrival, with no knowledge of future arrivals. Although recent deep-reinforcement-learning methods achieve superior volume utilization compared with classical heuristics, the learned policies cannot ensure the structural stability of the bin and lack mechanisms for safely reconfiguring the bin when a new item cannot be placed directly. In this work, we propose a novel framework that integrates packing policy with structural stability validation and heuristic planning to overcome these limitations. Specifically, we introduce the concept of Load Bearable Convex Polygon (LBCP), which provides a computationally efficient way to identify stable loading positions that guarantee no bin collapse. Additionally, we present Stable Rearrangement Planning (SRP), a module that rearranges existing items to accommodate new ones while maintaining overall stability. Extensive experiments on standard OBPP benchmarks demonstrate the efficiency and generalizability of our LBCP-based stability validation, as well as the superiority of SRP in finding the effort-saving rearrangement plans. Our method offers a robust and practical solution for automated packing in real-world industrial and logistics applications.

cs.RO↗

arXiv2025

Efficiently Building a Domain-Specific Large Language Model from Scratch: A Case Study of a Classical Chinese Large Language Model

General-purpose large language models demonstrate notable capabilities in language comprehension and generation, achieving results that are comparable to, or even surpass, human performance in many natural language processing tasks. Nevertheless, when general models are applied to some specific domains, e.g., Classical Chinese texts, their effectiveness is often unsatisfactory, and fine-tuning open-source foundational models similarly struggles to adequately incorporate domain-specific knowledge. To address this challenge, this study developed a large language model, AI Taiyan, specifically designed for understanding and generating Classical Chinese. Experiments show that with a reasonable model design, data processing, foundational training, and fine-tuning, satisfactory results can be achieved with only 1.8 billion parameters. In key tasks related to language processing of Classical Chinese such as punctuation, identification of allusions, explanation of word meanings, and translation between ancient and modern Chinese, this model exhibits a clear advantage over both general-purpose large models and domain-specific traditional models, achieving levels close to or surpassing human baselines. This research provides a reference for the efficient construction of specialized domain-specific large language models. Furthermore, the paper discusses the application of this model in fields such as the collation of ancient texts, dictionary editing, and language research, combined with case studies.

cs.CL↗

arXiv2025

BERP: A Blind Estimator of Room Parameters for Single-Channel Noisy Speech Signals

Room acoustical parameters (RAPs), room geometrical parameters (RGPs) and instantaneous occupancy level are essential metrics for parameterizing the room acoustical characteristics (RACs) of a sound field around a listener's local environment, offering comprehensive indications for various applications. Current blind estimation methods either fail to cover a broad range of real-world acoustic environments in the context of real background noise or estimate only a few RAPs and RGPs from noisy single-channel speech signals. In addition, they are limited in their ability to estimate the instantaneous occupancy level. In this paper, we propose a new universal blind estimation framework called the blind estimator of room parameters (BERP) to estimate RAPs, RGPs and occupancy level via a unified methodology. It consists of two modules: a unified room feature encoder that combines attention mechanisms with convolutional layers to learn common features across room parameters, and multiple separate parametric predictors for continuous estimation of each parameter in parallel. The combination of attention and convolutions enables the model to capture acoustic features locally and globally from speech, yielding more robust and multitask generalizable common features. Separate predictors allow the model to independently optimize for each room parameter to reduce task learning conflict and improve per-task performance. This estimation framework enables universal and efficient estimation of room parameters while maintaining satisfactory performance. To evaluate the effectiveness of the proposed framework, we compile a task-specific dataset from several publicly available datasets, including synthetic and real reverberant recordings. The results reveal that BERP achieves state-of-the-art (SOTA) performance and excellent adaptability to real-world scenarios. The code and weights are available on GitHub.

eess.AS↗

arXiv2025

Learning Universal Features for Generalizable Image Forgery Localization

In recent years, advanced image editing and generation methods have rapidly evolved, making detecting and locating forged image content increasingly challenging. Most existing image forgery detection methods rely on identifying the edited traces left in the image. However, because the traces of different forgeries are distinct, these methods can identify familiar forgeries included in the training data but struggle to handle unseen ones. In response, we present an approach for Generalizable Image Forgery Localization (GIFL). Once trained, our model can detect both seen and unseen forgeries, providing a more practical and efficient solution to counter false information in the era of generative AI. Our method focuses on learning general features from the pristine content rather than traces of specific forgeries, which are relatively consistent across different types of forgeries and therefore can be used as universal features to locate unseen forgeries. Additionally, as existing image forgery datasets are still dominated by traditional hand-crafted forgeries, we construct a new dataset consisting of images edited by various popular deep generative image editing methods to further encourage research in detecting images manipulated by deep generative models. Extensive experimental results show that the proposed approach outperforms state-of-the-art methods in the detection of unseen forgeries and also demonstrates competitive results for seen forgeries. The code and dataset are available at https://github.com/ZhaoHengrun/GIFL.

cs.CV↗

arXiv2025

Mono2Stereo: A Benchmark and Empirical Study for Stereo Conversion

With the rapid proliferation of 3D devices and the shortage of 3D content, stereo conversion is attracting increasing attention. Recent works introduce pretrained Diffusion Models (DMs) into this task. However, due to the scarcity of large-scale training data and comprehensive benchmarks, the optimal methodologies for employing DMs in stereo conversion and the accurate evaluation of stereo effects remain largely unexplored. In this work, we introduce the Mono2Stereo dataset, providing high-quality training data and benchmark to support in-depth exploration of stereo conversion. With this dataset, we conduct an empirical study that yields two primary findings. 1) The differences between the left and right views are subtle, yet existing metrics consider overall pixels, failing to concentrate on regions critical to stereo effects. 2) Mainstream methods adopt either one-stage left-to-right generation or warp-and-inpaint pipeline, facing challenges of degraded stereo effect and image distortion respectively. Based on these findings, we introduce a new evaluation metric, Stereo Intersection-over-Union, which prioritizes disparity and achieves a high correlation with human judgments on stereo effect. Moreover, we propose a strong baseline model, harmonizing the stereo effect and image quality simultaneously, and notably surpassing current mainstream methods. Our code and data will be open-sourced to promote further research in stereo conversion. Our models are available at mono2stereo-bench.github.io.

cs.CV↗