arXiv subjects

Mingda Zhang

Publications and source records attributed to Mingda Zhang.

At least 19 recordsLinked to original sources

arXiv2026

DecoupAttn: Decoupled Sparse Attention for Training-Free Acceleration of Video Diffusion

Sparse attention accelerates few-step video diffusion transformers by computing a few key blocks exactly and compensating for the rest. However, existing sparse attention still faces key challenges, including identification overhead from finding critical blocks at every denoising step, suboptimal budget allocation that sets sparsity budgets uniformly or by local metrics, and cross-step error accumulation as reused masks or clusters go stale. Observing that the errors left on skipped blocks cancel each other once every key is compensated, we propose DecoupAttn, a new paradigm of decoupled sparse attention: semantic clusters compensate every key, and exact attention corrects their error on a few blocks fixed once offline by position. To support this paradigm, we introduce Distortion Budget, a rate-distortion objective for budget allocation that places exact computation across timesteps, layers, and heads where it removes the most distortion that compensation leaves in the final latent. Built on this budget, we further propose Reassign Cache, in which each step reassigns the drifting keys to cached centroids and refreshes compensation while the budgeted blocks stay fixed. Moreover, DecoupAttn is training-free: a few calibration clips compile one plan offline, and inference runs standard block-sparse kernels with one fused compensation kernel. Our code will be released after submission.

cs.CV↗

arXiv2026

LADDER: Diagnosing Domain-Level Self-Assessment Under Noisy Feedback

Knowing when they are reliable is essential for language models to support human-AI collaboration. Yet interaction feedback is not always complete or correct, and may mislead models' understanding of their own capabilities. We introduce LADDER to study how language models assess themselves under unreliable feedback. Our findings show that models can accurately report supplied performance information while still struggling to form trustworthy self-assessments from interaction experience. Apparently accurate assessments can obscure capability differences and do not necessarily predict future task performance. These findings suggest that trustworthy self-assessment should be evaluated against observed task performance and its predictive value. Our study provides empirical evidence of the limits of model self-knowledge and a basis for evaluating capability awareness in reliable human-AI collaboration.

cs.CL↗

arXiv2026

From Belief Inference to Task Action: A Multi-Level Evaluation of Theory of Mind on Verifiable Tasks

Effective collaboration requires language models to understand what others know and believe, and to act accordingly. Models that assume others share their information may omit necessary communication and undermine coordination. Many theory-of-mind (ToM) benchmarks assess mental-state inference through story questions, but correct answers do not establish that models can use these judgments in subsequent tasks. We introduce a multi-level evaluation built on existing, verifiable tasks. We extend household collaboration and calendar scheduling with asymmetric information while retaining their original constraints and validators. Independent interfaces assess world understanding, information access, beliefs, decisions, and execution. Controls supply personal evidence or correct beliefs to distinguish state recovery from state use. Across six model configurations, these abilities diverge. On matched scheduling cases, Claude recovers all agent beliefs correctly but produces subjectively optimal plans in only 67.9% of cases. In household collaboration, supplying beliefs raises Qwen-27B success from 55.0% to 75.0%, yet Qwen-4B answers every fixed supplied-belief source question correctly while completing only 20.0% of the corresponding continuation condition. These results show why ToM evaluation must connect perspective inference to independently verified decisions and task consequences.

cs.CL↗

arXiv2026

SkillFlow: Flow-Driven Recursive Skill Evolution for Agentic Orchestration

In recent years, a variety of powerful LLM-based agentic systems have been applied to automate complex tasks through task orchestration. However, existing orchestration methods still face key challenges, including strategy collapse under reward maximization, high gradient variance with opaque credit assignment, and unguided skill evolution whose decisions are typically made by directly prompting an LLM to judge rather than derived from principled training signals. To address these challenges, we propose SkillFlow, a flow-based framework that takes a trainable Supervisor as the agent and a structured environment with dynamic skill library and frozen executor, automating task orchestration through multi-turn interaction. SkillFlow employs Tempered Trajectory Balance (TTB), a regression-based flow-matching loss that samples trajectories proportional to reward, preserving diverse orchestration strategies rather than collapsing to a single mode. The same flow objective yields a jointly learned backward policy that provides transparent per-step credit assignment at zero additional inference cost. Building on these flow diagnostics, a recursive skill evolution mechanism determines when to evolve, what skills to create or prune, and where decision gaps lie -- closing the loop from training signal to autonomous capability growth. Experimental results on 14 datasets show that SkillFlow significantly outperforms baselines across question answering, mathematical reasoning, code generation, and real-world interactive decision making tasks. Our code is available at https://github.com/beita6969/SkillFlow.

cs.AI↗

arXiv2026

R$^2$ Flow: Recursive Self-Improvement via Recursive Skill Evolution

LLM-based agents can improve themselves across tasks by reusing and revising the skills they orchestrate into executable procedures. Flow-based training fits this loop: it samples procedures in proportion to reward, and the flow through each skill credits it for the next library revision. Three obstacles stand in the way of making this self-improvement reliable: flow training suffers strategy collapse over tree-structured histories; nonnegative flow-based credit rewards frequent use as if it were benefit; and library edits rest on the task reward the policy optimizes. We introduce R$^2$ Flow, a recursive self-improvement framework that alternates policy learning, independent verification, and versioned skill-library updates on a shared-state orchestration graph. The graph merges histories that differ only in the order of independent steps, allowing flow training to pool evidence across equivalent executions. A flow-share readout of the trained flow, invariant to the backward policy, and a separate signed utility rank which skills to change, verifier evidence decides whether an edit is warranted, and a residual-variance plateau sets when to update. Committed edits reshape the graph the next policy learns on, realizing recursive skill evolution. Across question answering, mathematical reasoning, interactive decision making, and code generation, R$^2$ Flow improves task accuracy and library-edit precision over heuristic orchestration, reinforcement learning, and skill-evolution baselines, and transfers across executors. Code is available at https://github.com/beita6969/r2flow.

cs.AI↗

arXiv2026

EvoSteer: Online Self-Evolving Graph Orchestration via Reference-Anchored Credit Assignment

In recent years, LLM-based multi-agent systems have been widely applied to orchestrate tool-using agents into executable communication graphs. However, existing self-evolving orchestration still faces key challenges, including post-hoc evolution that revises the team only after the trajectory ends, credit diffusion that gives every action the same terminal advantage under confounded baselines, and skill admission that is uncalibrated and never retired. To address these challenges, we propose EvoSteer, a new paradigm of Online Self-Evolving Graph Orchestration -- the orchestrator builds a running team and repairs its plausible but failing steps from execution features and a learned value estimate. To support this paradigm, we introduce Anchored Trajectory Balance (AnchorTB), a regression-style flow-matching loss that assigns each orchestration action a coefficient by balancing subtrajectories against a frozen reference. Built on the learned flow, we further propose Validated Skill Admission, in which a candidate skill is tried before promotion and promoted only if paired evidence passes a sequential test under a shared nominal testing budget. Moreover, AnchorTB combines measured task-level reference reward statistics with prefix-dependent corrections. Experimental results on twelve datasets show that EvoSteer significantly outperforms baselines across question answering, mathematical reasoning, code generation, and interactive decision making. Our code is available at https://github.com/beita6969/evosteer.

cs.AI↗

arXiv2026

CollabFlow: Recursive Self-Improvement of Agent Collaboration

Recursive self-improvement (RSI) lets a system improve from its own outcomes; in LLM-based multi-agent systems, Agents refine one another within a task, and outcomes improve how they collaborate across tasks. However, existing multi-agent collaboration leaves this loop open: collaboration is pre-defined at the operator level, topology-only learning keeps verbatim exchange that propagates errors, and reward maximization on a system's own outcomes concentrates on a few teams. To address these challenges, we propose CollabFlow, an RSI system of Learned Agent Collaboration: a trainable Collab-Director constructs teams of complete Agents, a frozen executor runs them, and each round's outcomes retrain the director. Within each round, the edges of a collaboration graph carry protocols of Evidence-Conditioned Communication: a receiver adopts a differing answer only when the sender's evidence is stronger by a margin, so the director learns who communicates and how. Across rounds, we further propose Collaborative Trajectory Balance (CTB), a flow-based objective that credits each team once across its construction orders and targets a reward-proportional distribution over teams, so several good teams stay in play. We also bound how far this self-generated target moves between rounds, which shrinks as records accumulate. On twelve datasets, CollabFlow outperforms all baselines and keeps improving across rounds. Code is available at https://anonymous.4open.science/r/CollabFlow-631E.

cs.MA↗

arXiv2026

JusticeAxis: Benchmarking Legal Judgment between Rigid Rule Application and Ungrounded Discretion

A sound judgment applies the law to established facts and weighs the circumstances in which they arose. However, existing methods swing between rigid statute matching and ungrounded discretion, benchmarks score a label or a rubric, and the experience that would supply the balance stays unverified. We formalize legal judgment as a reference-anchored task, whose object is a single decision that stays tied to the statute and to the circumstances at once. We introduce JusticeAxis, 256 real-world criminal cases from 18 countries with audio, image, and text evidence, and three lawyer-written judgments for every case: the recorded one and one for each failure. We further propose JusticeAgent, a harness whose element agents establish the facts and whose judge agent applies the law under skills carrying experience of the circumstances. Skills are distilled from execution trajectories and admitted only under Bayesian credible bounds. Experiments show that failure turns direction with scale: open-weight backbones drift to unsupported grounds, frontier models to the statutory default. We further verify that JusticeAgent, as a simple yet effective plugin, carries a frozen open-weight backbone to commercial level. Project resources are available at https://github.com/beita6969/JusticeAxis.

cs.AI↗

arXiv2026

R$^2$Orch: Recursive Self-Improvement for Agent Orchestration via Self-Play within a Skill-Internalization Loop

In recent years, agentic orchestration has been widely applied to solve complex human tasks. However, existing orchestration learning still faces key challenges, including hand-designed curricula, advantage collapse when a task's rollouts all reach the same outcome, and unreliable evaluation signals that judge each skill by one noisy score. To address these challenges, we propose R$^2$Orch, a new paradigm of Recursive Self-Play Orchestration: an inner loop of self-play that improves the orchestrator on self-chosen tasks, and an outer loop that improves how the inner loop improves. To support this paradigm, we introduce Recursive Learnability, a training objective that rewards a Proposer for choosing tasks whose rollouts keep reaching different outcomes as skills are withdrawn. Built on this objective, we further propose Skill Withdrawal, in which the outer loop withdraws each skill once a controlled evaluation within self-play shows that the orchestrator performs as well without it. Moreover, the two loops form one cycle: failures are distilled into skills, skills into parameters, and withdrawals lead the Proposer to new failures, so the orchestrator moves toward zero-skill deployment. Experimental results on twelve datasets show that R$^2$Orch consistently outperforms baselines across question answering, math, code, and interactive tasks. Our code is available at https://github.com/beita6969/R2Orch.

cs.AI↗

arXiv2026

ScienceClaw: Benchmarking Continual Self-Evolution of AI-for-Science Agents Across the Natural and Social Sciences

Large language model agents are accelerating scientific automation, yet verified executions rarely become persistent program-level improvements, and existing evaluations do not examine this process across sequential tasks in both the natural and social sciences. We formalize ScienceClaw as fixed-parameter program self-evolution that unifies task solving, scientific verification, and program updates. ScienceClaw-Eval spans 23 disciplines and measures scientific correctness, evolutionary gain, retention, cross-dataset transfer, and evolution cost through sequential streams and independent reset evaluation. Our framework repairs executable workflows through multi-turn interaction, converts re-execution-verified failure--success trajectories into linked Skill and Operator candidates, and retains an update only when source-task replay reproduces the repair and independent scientific tasks improve. Code is available at https://github.com/beita6969/ScienceClaw.

cs.AI↗

arXiv2026

Topological-Charge-Enabled Photonic Doping in ENZ Media

Conventional photonic doping schemes predominantly employ circular or rectangular dielectric dopants with zero topological charge, where the effective permeability can only be tuned through material selection and geometric scaling, resulting in limited design flexibility. In this work, topological structures are introduced into dielectric dopants by embedding internal holes to generate nonzero topological charge. Based on this concept, a theoretical model is established to describe the effective permeability of photonic doping systems with nonzero topological charge, and the underlying mechanisms governing topological-charge-dependent transmission are systematically elucidated. The results demonstrate that engineering nonzero topological charge through the number, shape, size and position of internal holes within dielectric dopants enables flexible manipulation of the internal magnetic field distributions, thereby providing precisely control over the effective permeability, as well as the resonance frequency and spectral linewidth of the transmission spectrum. The proposed multi-dimensional photonic doping strategy, integrating topological-charge engineering with geometric design, substantially enriches the available degrees of freedom for dispersion engineering and provides a versatile platform for advanced functional photonic devices.

physics.optics↗

arXiv2026

FlowSteer: Towards Agents Designing Agentic Workflows via Reinforced Progressive Canvas Editing

In recent years, agentic workflows have been widely applied to solve complex human tasks. However, existing workflow construction still faces key challenges, including human-dependent workflow construction, the lack of graph-level execution feedback, and the inability to repair errors in-loop during long-horizon construction. To address these challenges, we propose FlowSteer, a new paradigm of Agent Designing Agentic Workflows - a single agent itself end-to-end designs the workflow that a downstream executor runs. To support this paradigm, we introduce the Workflow Canvas, a novel executable graph-state environment that returns syntax-checked execution feedback for every atomic edit. Built on the canvas, we further propose Reinforced Progressive Canvas Editing, in which a lightweight policy agent issues one atomic edit per turn conditioned on real canvas feedback, and is trained end-to-end via reinforcement learning. Moreover, FlowSteer provides a plug-and-play framework that supports diverse operator libraries and interchangeable LLM backends. Experimental results on twelve datasets show that FlowSteer significantly outperforms baselines across various tasks. Our code is available at https://anonymous.4open.science/r/FlowSteer-9B2E.

cs.AI↗

arXiv2026

CS-MUNet: A Channel-Spatial Dual-Stream Mamba Network for Multi-Organ Segmentation

Recently Mamba-based methods have shown promise in abdominal organ segmentation. However, existing approaches neglect cross-channel anatomical semantic collaboration and lack explicit boundary-aware feature fusion mechanisms. To address these limitations, we propose CS-MUNet with two purpose-built modules. The Boundary-Aware State Mamba module employs a Bayesian-attention framework to generate pixel-level boundary posterior maps, injected directly into Mamba's core scan parameters to embed boundary awareness into the SSM state transition mechanism, while dual-branch weight allocation enables complementary modulation between global and local structural representations. The Channel Mamba State Aggregation module redefines the channel dimension as the SSM sequence dimension to explicitly model cross-channel anatomical semantic collaboration in a data-driven manner. Experiments on two public benchmarks demonstrate that CS-MUNet consistently outperforms state-of-the-art methods across multiple metrics, establishing a new SSM modeling paradigm that jointly addresses channel semantic collaboration and boundary-aware feature fusion for abdominal multi-organ segmentation.

cs.CV↗

arXiv2025

Knowledge-Guided Brain Tumor Segmentation via Synchronized Visual-Semantic-Topological Prior Fusion

Background: Brain tumor segmentation requires precise delineation of hierarchical structures from multi-sequence MRI. However, existing deep learning methods primarily rely on visual features, showing insufficient discriminative power in ambiguous boundary regions. Moreover, they lack explicit integration of medical domain knowledge such as anatomical semantics and geometric topology. Methods: We propose a knowledge-guided framework, Synchronized Tri-modal Prior Fusion (STPF), that explicitly integrates three heterogeneous knowledge priors: pathology-driven differential features (T1ce-T1, T2-FLAIR, T1/T2) encoding contrast patterns; unsupervised semantic descriptions transformed into voxel-level guidance via spatialization operators; and geometric constraints extracted through persistent homology analysis. A dual-level fusion architecture dynamically allocates prior weights at the voxel level based on confidence and at the sample level through hypernetwork-generated conditional vectors. Furthermore, nested output heads structurally ensure the hierarchical constraint ET subset TC subset WT. Results: STPF achieves a mean Dice coefficient of 0.868 on the BraTS 2020 dataset, surpassing the best baseline by 2.6 percentage points (3.09% relative improvement). Notably, five-fold cross-validation yields coefficients of variation between 0.23% and 0.33%, demonstrating stable performance. Additionally, ablation experiments show that removing topological and semantic priors leads to performance degradation of 2.8% and 3.5%, respectively. Conclusions: By explicitly integrating medical knowledge priors - anatomical semantics and geometric constraints - STPF improves segmentation accuracy in ambiguous boundary regions while demonstrating generalization capability and clinical deployment potential.

cs.CV↗

arXiv2025

Multimodal Fusion at Three Tiers: Physics-Driven Data Generation and Vision-Language Guidance for Brain Tumor Segmentation

Accurate brain tumor segmentation is crucial for neuro-oncology diagnosis and treatment planning. Deep learning methods have made significant progress, but automatic segmentation still faces challenges, including tumor morphological heterogeneity and complex three-dimensional spatial relationships. This paper proposes a three-tier fusion architecture that achieves precise brain tumor segmentation. The method processes information progressively at the pixel, feature, and semantic levels. At the pixel level, physical modeling extends magnetic resonance imaging (MRI) to multimodal data, including simulated ultrasound and synthetic computed tomography (CT). At the feature level, the method performs Transformer-based cross-modal feature fusion through multi-teacher collaborative distillation, integrating three expert teachers (MRI, US, CT). At the semantic level, clinical textual knowledge generated by GPT-4V is transformed into spatial guidance signals using CLIP contrastive learning and Feature-wise Linear Modulation (FiLM). These three tiers together form a complete processing chain from data augmentation to feature extraction to semantic guidance. We validated the method on the Brain Tumor Segmentation (BraTS) 2020, 2021, and 2023 datasets. The model achieves average Dice coefficients of 0.8665, 0.9014, and 0.8912 on the three datasets, respectively, and reduces the 95% Hausdorff Distance (HD95) by an average of 6.57 millimeters compared with the baseline. This method provides a new paradigm for precise tumor segmentation and boundary localization.

eess.IV↗

arXiv2025

ELPG-DTFS: Prior-Guided Adaptive Time-Frequency Graph Neural Network for EEG Depression Diagnosis

Timely and objective screening of major depressive disorder (MDD) is vital, yet diagnosis still relies on subjective scales. Electroencephalography (EEG) provides a low-cost biomarker, but existing deep models treat spectra as static images, fix inter-channel graphs, and ignore prior knowledge, limiting accuracy and interpretability. We propose ELPG-DTFS, a prior-guided adaptive time-frequency graph neural network that introduces: (1) channel-band attention with cross-band mutual information, (2) a learnable adjacency matrix for dynamic functional links, and (3) a residual knowledge-graph pathway injecting neuroscience priors. On the 128-channel MODMA dataset (53 subjects), ELPG-DTFS achieves 97.63% accuracy and 97.33% F1, surpassing the 2025 state-of-the-art ACM-GNN. Ablation shows that removing any module lowers F1 by up to 4.35, confirming their complementary value. ELPG-DTFS thus offers a robust and interpretable framework for next-generation EEG-based MDD diagnostics.

cs.CV↗

arXiv2025

DCFFSNet: Deep Connectivity Feature Fusion Separation Network for Medical Image Segmentation

Medical image segmentation leverages topological connectivity theory to enhance edge precision and regional consistency. However, existing deep networks integrating connectivity often forcibly inject it as an additional feature module, resulting in coupled feature spaces with no standardized mechanism to quantify different feature strengths. To address these issues, we propose DCFFSNet (Dual-Connectivity Feature Fusion-Separation Network). It introduces an innovative feature space decoupling strategy. This strategy quantifies the relative strength between connectivity features and other features. It then builds a deep connectivity feature fusion-separation architecture. This architecture dynamically balances multi-scale feature expression. Experiments were conducted on the ISIC2018, DSB2018, and MoNuSeg datasets. On ISIC2018, DCFFSNet outperformed the next best model (CMUNet) by 1.3% (Dice) and 1.2% (IoU). On DSB2018, it surpassed TransUNet by 0.7% (Dice) and 0.9% (IoU). On MoNuSeg, it exceeded CSCAUNet by 0.8% (Dice) and 0.9% (IoU). The results demonstrate that DCFFSNet exceeds existing mainstream methods across all metrics. It effectively resolves segmentation fragmentation and achieves smooth edge transitions. This significantly enhances clinical usability.

cs.CV↗

arXiv2025

A Semantic Segmentation Algorithm for Pleural Effusion Based on DBIF-AUNet

Pleural effusion semantic segmentation can significantly enhance the accuracy and timeliness of clinical diagnosis and treatment by precisely identifying disease severity and lesion areas. Currently, semantic segmentation of pleural effusion CT images faces multiple challenges. These include similar gray levels between effusion and surrounding tissues, blurred edges, and variable morphology. Existing methods often struggle with diverse image variations and complex edges, primarily because direct feature concatenation causes semantic gaps. To address these challenges, we propose the Dual-Branch Interactive Fusion Attention model (DBIF-AUNet). This model constructs a densely nested skip-connection network and innovatively refines the Dual-Domain Feature Disentanglement module (DDFD). The DDFD module orthogonally decouples the functions of dual-domain modules to achieve multi-scale feature complementarity and enhance characteristics at different levels. Concurrently, we design a Branch Interaction Attention Fusion module (BIAF) that works synergistically with the DDFD. This module dynamically weights and fuses global, local, and frequency band features, thereby improving segmentation robustness. Furthermore, we implement a nested deep supervision mechanism with hierarchical adaptive hybrid loss to effectively address class imbalance. Through validation on 1,622 pleural effusion CT images from Southwest Hospital, DBIF-AUNet achieved IoU and Dice scores of 80.1% and 89.0% respectively. These results outperform state-of-the-art medical image segmentation models U-Net++ and Swin-UNet by 5.7%/2.7% and 2.2%/1.5% respectively, demonstrating significant optimization in segmentation accuracy for complex pleural effusion CT images.

cs.CV↗