arXiv subjects

Yimeng Wang

Publications and source records attributed to Yimeng Wang.

At least 19 recordsLinked to original sources

arXiv2026

PROSE: Preservation-Constrained Selective Editing for Guilty Bias in Legal Judgment Prediction

Legal judgment prediction models may retain a guilty verdict after a legally decisive condition has been altered such that the specified charge is no longer established. We study this guilty bias via selective repair over matched charge-established and charge-not-established cases, where the unsupported verdict should change while its supported counterpart remains stable. We construct CAIL-MC, a legally reviewed dataset of 11,478 matched pairs covering 261 Chinese criminal charges. Held-out activation interventions show that contrasts learned from correctly resolved pairs transfer to false-guilty cases, but target-reaching interventions can also damage matched supported-guilty decisions. Motivated by this observation, we introduce PROSE, which reconstructs case-specific targets from a training-derived residual geometry, aligns them with a local margin-descent signal, attenuates supported-guilty-sensitive components, and realizes bounded targets through a preservation-regularized low-rank parameter update. Across Qwen3-8B and GLM4-9B, PROSE achieves stronger pairwise safe repair and evidence sensitivity, while preserving standard single-pass inference.

cs.CL↗

arXiv2026

Progressive Pseudo-Label Optimization for Point-Supervised Change Detection

Point-supervised change detection (PS-CD) aims to identify pixel-level changes between bi-temporal images using only sparsely annotated points. Although point annotations substantially reduce labeling costs, their limited spatial coverage often results in incomplete and noisy pseudo-labels. To address this issue, we propose a two-stage framework that introduces SAM2 priors into PS-CD and progressively adapts them to the target task. In Stage I, SAM2 generates object-aware candidate masks from point annotations on the bi-temporal images, and a bi-temporal mask selection strategy is designed to convert generic segmentation responses into more reliable change pseudo-labels. Subsequently, a lightweight CNN refinement module with an uncertainty-aware loss is employed to improve boundary quality and local structural consistency. In Stage II, we construct a teacher-student self-training framework in which the teacher is updated by exponential moving average and periodically refreshes the pseudo-labels. This design establishes a closed-loop optimization process that alternates between pseudo-label refinement and model re-optimization. Experiments on three benchmark datasets, including WHU-CD, LEVIR-CD, and SYSU-CD, demonstrate that the proposed method outperforms previous weakly supervised approaches on most benchmarks and remains competitive with several fully supervised methods.

cs.CV↗

arXiv2026

Lower Bounds for Parallel Diffusion Sampling

Standard diffusion samplers generate samples through sequential evaluations of a learned score function, with each query depending on previous answers. Parallel samplers seek to accelerate this process by making multiple score queries simultaneously. We ask how many adaptive rounds are necessary when each round allows polynomially many queries. We establish the first polynomial lower bounds on the round complexity of diffusion sampling with approximate scores satisfying a mean-square accuracy guarantee. Specifically, we prove (1) a $\widetildeΩ(d^{1/3})$-round lower bound for sampling smooth, near-isotropic Gaussian mixtures in $\mathbb{R}^d$, and (2) a tight $Ω(d)$-round lower bound for uniform sampling from anisotropic axis-aligned boxes contained in the unit ball. Both bounds hold for arbitrary randomized algorithms making polynomially many queries per round at arbitrary locations and noise levels, with inverse-polynomial score error and constant total variation accuracy. Our near-isotropic construction follows the blockwise approach of earlier lower bounds for parallel matroid search and discrete sampling \cite{KUW85,AGR24}. We use independent coordinate blocks containing random codebooks with well-separated rates and construct approximate score oracles whose responses in each block depend on consistency tests in earlier blocks. Together, the rate separation and these tests ensure that each round reveals information about at most one additional block with high probability.

cs.DS↗

arXiv2026

The weighted union-set conjecture is false

We disprove a weighted generalization of the union-set conjecture [Fra95] raised in [Gow16]. We construct nonnegative weights on subsets of a finite ground set satisfying $w(\varnothing)=0$ and $w(A\cup B)\ge\min\{w(A),w(B)\}$, while every coordinate belongs to an arbitrarily small fraction of the total weight.

math.CO↗

arXiv2026

GANDR: Claim Auditing for Verifiable Legal Answer Generation

In high-stakes domains such as legal practice, a language-model answer is only useful to the extent that a reader can verify each claim against the source the system cites. Current grounded-generation pipelines score the answer as a whole, so a correct conclusion can rest on fabricated or loosely matched citations and still score well. Closing this gap requires both a system built for per-claim verification and an evaluation that measures it. We introduce GANDR (Grounded ANswer DRafter), a two-agent system in which a Drafter writes an answer in a structured legal-reasoning format and a separate Critic, with the same view as a human verifier, audits each claim against its cited source and emits a per-claim audit trace on every round. We pair it with a strict correctness criterion requiring every citation to resolve to a passage the retriever returned. On a 185-item legal benchmark where all six systems share one backbone, one retrieval surface, and one citation instruction, GANDR ranks first on every primary metric, reaching 70.8% strict accuracy and leading the strongest baseline by 11.3 points (p<0.01). Reverting the protocol-anchored commit rule lowers strict accuracy by 22.7 points, and the strict lead stays positive on three further backbones, at +3.2 to +6.5 points. This lead traces to the Drafter configuration and the protocol-anchored commit, not to rewriting. Against two law-trained annotators the audit flags under-supported claims at F1 0.84 as a binary detector, while its four-way verdict labels agree only weakly and are advisory. Code is available upon request.

cs.CL↗

arXiv2026

AtomCite: Verification and Correction of Supplied Page-Level Citations in Multi-Page Documents

Large language models answering questions over multi-page documents are expected to cite the supporting pages, yet supplied citations are sometimes inaccurate, and current evaluations score citations at generation time or against text passages: no existing benchmark evaluates whether a system can verify and correct a page-level citation already attached to an answer. We propose AtomCite, an agentic framework that parses an answer into claims, checks each claim against the image of its cited page, and applies a deterministic repair policy. To evaluate it, we introduce DocCite, to our knowledge the first benchmark for systems that verify and correct page-level citations in document images. Built on MP-DocVQA and DUDE, it combines 928 validated injected instances with 2,468 candidate natural errors harvested from frontier- and efficiency-tier models, of which a two-annotator audit confirms 1,909 as genuine errors. Primary labels are assigned deterministically, not by LLM judges, with the human audit as a separate validation layer. Across three model families (Gemini, Claude, and GPT), AtomCite reaches around 93% binary verification accuracy on the injected benchmark, significantly outperforming every OCR-only condition, including a compute-matched control, and exceeding every prior text-based baseline given the same OCR text. Its repair policy lifts citation precision on the injected mix from a constructed 34% to 87-90% while retaining over 90% of correct claims. AtomCite also transfers: with frozen prompts and zero training, it raises the hallucination-detection scores of two open 7-8B models on five public benchmarks above the same models prompted as direct judges. Finally, the audit shows that noise in automatic labels biases measured verifier accuracy and can reverse system rankings, so evaluations relying only on synthetic or automatic labels risk mismeasuring verification capability.

cs.CL↗

arXiv2026

Moonflowers and efficient code sparsification

We introduce \emph{moonflowers}, a weaker analogue of sunflowers. A family of sets $S_1,\ldots,S_k$ is a $k$-moonflower if each set $S_i$ contains at least one element that is absent from all the others. We study the extremal problem of determining the largest possible size of a family of sets of size at most $w$ that avoids a $k$-moonflower, and obtain near-optimal bounds. As an application, we revisit the code sparsification problem studied by Brakensiek and Guruswami (STOC 2025) and improve the bounds to near optimal. Concretely, we improve the dependence on the block length from poly-logarithmic to logarithmic, and show that such a dependence is necessary.

math.CO↗

arXiv2026

Chronos: A Physics-Informed Full-History Framework for Non-Markovian Long-Horizon Manipulation

General-purpose robot policies should be modeled as dynamical systems, yet many VLA and generative imitation policies still rely on present observations or short windows. This Markovian shortcut fails in memory-dependent manipulation: identical observations can demand different actions after different histories. We present Chronos, a physics-informed full-history framework for non-Markovian long-horizon manipulation. The key idea is to elevate observation history from auxiliary context to the latent state of the policy dynamics. At each physical control step, Chronos forms one state-representative token by fusing observation and proprioception, so the token sequence is aligned one-to-one with physical time. A selective state space model propagates this causal historical state, which conditions a multimodal coarse action prior through implicit maximum likelihood estimation (IMLE). This prior is then refined by a second-order Schrodinger-inspired bridge that predicts acceleration fields, yielding smoother and more physically grounded robot motion. Across 16 simulated tasks and 4 real-world experiments, Chronos is evaluated on precision insertion, general manipulation, and memory-dependent long-horizon control. On RMBench, where success requires remembering task phase, Chronos achieves 73.6% average success, outperforming Markovian VLA baseline pi0.5 by +62.4 percentage points, a 6.6x relative gain, while using 10x fewer parameters. It also surpasses the memory VLA Mem-0 by 22.8 points while using over 30x fewer parameters. In real-world dual-arm experiments using a single RGB camera, Chronos achieves 78% average success over four tasks, including 72% on the three memory-dependent tasks, whereas pi0.5 achieves 7% overall and 0% on the memory-dependent subset. These results suggest that history should not be treated as auxiliary context, but as the latent state of the manipulation policy.

cs.RO↗

arXiv2026

MIMO-OTFS-Based Semantic Communication for High-Mobility Scenarios

In high-mobility scenarios with time-frequency doubly-selective channels, existing semantic communication systems suffer significant performance degradation. To address this issue, we propose a semantic communication framework that synergistically integrates multiple-input multiple-output orthogonal time frequency space (MIMO-OTFS) with semantic-aware sub-channel allocation. First, an entropy module is employed to evaluate importance of different semantic features, and the Kendall correlation coefficient is used to quantify the alignment between semantic importance and sub-channel conditions. Subsequently, joint optimization of the encoder and decoder is achieved through a comprehensive loss function that balances image classification accuracy, reconstruction quality, and sub-channel matching degree. Experimental results confirm the superior reconstruction quality of our proposed framework compared to conventional semantic communication systems based on orthogonal frequency division multiplexing in high-mobility channel environment.

eess.SP↗

arXiv2026

HiDream-O1-Image: A Natively Unified Image Generative Foundation Model with Pixel-level Unified Transformer

The evolution of visual generative models has long been constrained by fragmented architectures relying on disjoint text encoders and external VAEs. In this report, we present HiDream-O1-Image, a natively unified generative foundation model via pixel-space Diffusion Transformer, that pioneers a paradigm shift from modular architectures to an end-to-end in-context visual generation engine. By mapping raw image pixels, text tokens, and task-specific conditions into a single shared token space, HiDream-O1-Image achieves a structural unification of multimodal inputs within an Unified Transformer (UiT) architecture. This native encoding paradigm eliminates the need for separate VAEs or disjoint pre-trained text encoders, allowing the model to treat diverse generation and editing tasks as a consistent in-context reasoning process. Extensive experiments show that HiDream-O1-Image excels across various generation tasks, including text-to-image generation, instruction-based editing, and subject-driven personalization. Notably, with only 8B parameters, HiDream-O1-Image (8B) achieves performance parity with or even surpasses established state-of-the-art models with significantly larger parameters (e.g., 27B Qwen-Image). Crucially, to validate the immense scalability of this paradigm, we successfully scale the architecture up to over 200B parameters. Experimental results demonstrate that this massive-scale version HiDream-O1-Image-Pro (200B+) unlocks unprecedented generative capabilities and superior performance, establishing new state-of-the-art benchmarks. Ultimately, HiDream-O1-Image highlights the immense potential of natively unified architectures and charts a highly scalable path toward next-generation multimodal AI.

cs.CV↗

arXiv2026

Explore LLM-enabled Tools to Facilitate Imaginal Exposure Exercises for Social Anxiety

Social anxiety (SA) is a prevalent mental health challenge that significantly impacts daily social interactions. Imaginal Exposure (IE), a Cognitive Behavioral Therapy (CBT) technique involving imagined anxiety-provoking scenarios, is effective but underutilized, in part because traditional IE homework requires clients to construct and sustain clinically relevant fear narratives. In this work, we explore the feasibility of an LLM-enabled tool that supports IE by generating vivid, personalized exposure scripts. We first co-designed ImaginalExpoBot with mental health professionals, followed by a formative evaluation with five therapists and a user study involving 19 individuals experiencing SA symptoms. Our findings show that LLM-enabled support can facilitate preparation for anxiety-inducing situations while enabling immediate, user-specific adaptation, with scenarios remaining within a therapeutically beneficial "window of tolerance". Our participants and MHPs also identified limitations in continuity and customization, pointing to the need for deeper adaptivity in future designs. These findings offer preliminary design insights for integrating LLMs into structured therapeutic practices in accessible, scalable ways.

cs.HC↗

arXiv2026

Designing AI Peers for Collaborative Mathematical Problem Solving with Middle School Students: A Participatory Design Study

Collaborative problem solving (CPS) is a fundamental practice in middle-school mathematics education; however, student groups frequently stall or struggle without ongoing teacher support. Recent work has explored how Generative AI tools can be designed to support one-on-one tutoring, but little is known about how AI can be designed as peer learning partners in collaborative learning contexts. We conducted a participatory design study with 24 middle school students, who first engaged in mathematics CPS tasks with AI peers in a technology probe, and then collaboratively designed their ideal AI peer. Our findings reveal that students envision an AI peer as competent in mathematics yet explicitly deferential, providing progressive scaffolds such as hints and checks under clear student control. Students preferred a tone of friendly expertise over exaggerated personas. We also discuss design recommendations and implications for AI peers in middle school mathematics CPS.

cs.HC↗

arXiv2026

MetaGen: Self-Evolving Roles and Topologies for Multi-Agent LLM Reasoning

Large language models are increasingly deployed as multi-agent systems, where specialized roles communicate and collaborate through structured interactions to solve complex tasks that often exceed the capacity of a single agent. However, most existing systems still rely on a fixed role library and an execution-frozen interaction topology, a rigid design choice that frequently leads to task mismatch, prevents timely adaptation when new evidence emerges during reasoning, and further inflates inference cost. We introduce MetaGen, a training-free framework that adapts both the role space and the collaboration topology at inference time, without updating base model weights. MetaGen generates and rewrites query-conditioned role specifications to maintain a controllable dynamic role pool, then instantiates a constrained execution graph around a minimal backbone. During execution, it iteratively updates role prompts and adjusts structural decisions using lightweight feedback signals. Experiments on code generation and multi-step reasoning benchmarks show that MetaGen improves the accuracy and cost tradeoff over strong multi-agent baselines.

cs.CL↗

arXiv2026

Exploring Customizable Interactive Tools for Therapeutic Homework Support in Mental Health Counseling

Therapeutic homework (i.e., tasks assigned by therapists for clients to complete between sessions) is essential for effective psychotherapy, yet therapists often interpret fragmented client logs, assessments, and reflections within limited preparation time. Our formative study with licensed therapists revealed three critical design requirements: support for interpreting unstructured client self-reports, customization aligned with clinical objectives, and seamless integration across multiple data sources. We then designed and developed TheraTrack, a customizable, therapist-facing tool that integrates multi-dimensional data and leverages large language models to generate traceable summaries and support natural-language queries, to streamline between-session homework tracking. Our pilot study with 14 therapists showed that TheraTrack reduced their cognitive load, enabled verification through direct navigation from AI summaries to original data entries, and was adapted differently for private analysis compared to in-session use, with dependence varying based on therapist experience and usage duration. We also discuss design implications for clinician-centered AI for mental health.

cs.HC↗

arXiv2026

From "Thinking" to "Justifying": Aligning High-Stakes Explainability with Professional Communication Standards

Explainable AI (XAI) in high-stakes domains should help stakeholders trust and verify system outputs. Yet Chain-of-Thought methods reason before concluding, and logical gaps or hallucinations can yield conclusions that do not reliably align with their rationale. Thus, we propose "Result -> Justify", which constrains the output communication to present a conclusion before its structured justification. We introduce SEF (Structured Explainability Framework), operationalizing professional conventions (e.g., CREAC, BLUF) via six metrics for structure and grounding. Experiments across four tasks in three domains validate this approach: all six metrics correlate with correctness (r=0.20-0.42; p<0.001), and SEF achieves 83.9% accuracy (+5.3 over CoT). These results suggest structured justification can improve verifiability and may also improve reliability.

cs.AI↗

arXiv2025

Motional entanglement in low-energy collisions near shape resonances

Einstein, Podolsky, and Rosen discussed their paradox in terms of measuring the positions or momenta of two particles. These degrees of freedom can become entangled upon scattering, but how much entanglement can be created in this process? Here we address this question using fully coherent calculations of bipartite scattering in three-dimensional space, quantifying entanglement by the inverse of the single particle purity. We show that the standard plane-wave description of scattering fails to capture the entanglement properties, due to the essential role of quantum uncertainty in the initial state. For a more realistic description of a scattering setup, we find that the entanglement scales linearly with the scattering cross section, including strong enhancement near shape resonances, for sufficiently narrow initial momentum dispersion. We highlight the differences between scattering in one and higher spatial dimensions and discuss how the generation of motional entanglement can be detected in experiments. Our results open the way to probing, controlling, and eventually using entanglement in quantum collisions.

quant-ph↗

arXiv2025

Self-sustained microcomb lasing in an integrated hybrid oscillator

Microcavity optical frequency combs (microcombs) are compact, coherent light sources whose chip-scale integrability is poised to drive advances in metrology, communications, and sensing. Among available microcomb generation methods, hybrid cavities uniquely co-locate gain and Kerr dynamics, where the lasing mode directly resonates in the nonlinear microcavity, simultaneously enabling self-sustained and highly efficient microcomb generation. However, their implementation is often limited by partial integration or the need for external injection, which complicates operation architecture, raises power and hampers system miniaturization. In this work, we present a fully integrated hybrid cavity for self-sustained microcomb generation, relying solely on the co-oscillation of lasing and Kerr nonlinearity without external driving. The system collapses the pump laser, nonlinear resonator and feedback loops into a minimalist on-chip two-element cavity, consisting of a high-Q microresonator with engineered intracavity reflection and a reflective semiconductor optical amplifier (RSOA). The scheme delivers self-starting operation and stable performance without active feedback. The generated coherent microcomb achieves intrinsic linewidths below 1 kHz and integrated linewidths around 100 kHz, with self-sustained operation exceeding 24 hours. This ultra-compact architecture provides a practical path toward scalable, coherent multi-wavelength sources for integrated photonic systems.

physics.optics↗

arXiv2025

Transparent and heat-insulation bionic hydrogel-based smart window system for long-term cooling and waste heat collection

With the energy crisis and climate warming, the position of a new generation of smart windows is becoming increasingly important, and materials or systems that can have high blocking of near-infrared (NIR) and ultraviolet (UV) and high transmittance of visible light (VIS) are needed. Currently, it is difficult for smart heat-insulation materials to achieve high transmittance of VIS, good UV isolation, outstanding cooling and thermal insulation, and excellent waste heat collection. Here, we design a novel composite hydrogel to achieve an average 92% VIS transmittance, efficient UV absorption , 11 Celsius degree of thermal insulation, and sensing properties. Interestingly, we designed a transparent heat insulation system with this composite hydrogel to obtain about 22 Celsius degree of the record-breaking insulation performance for 168 hours, waste heat collection and reutilization, and temperature sensing. Our findings provide new ideas and possibilities for designing transparent and heat-insulation smart window systems.

physics.chem-ph↗