arXiv subjects

Yan Zhou

Publications and source records attributed to Yan Zhou.

At least 19 recordsLinked to original sources

arXiv2026

Gnomon4D: Learning Where and When to Focus for Geometric Motion-Grounded World-Action Modeling

World-action models (WAMs) use video-generation priors to jointly predict future observations and robot actions. Yet RGB futures entangle physical state changes with appearance, while frame-wise depth and point maps lack temporal correspondence. Existing 4D approaches either keep motion implicit in visual features or translate predicted traces into actions through a separate interface. We present Gnomon4D, a WAM that jointly generates future video, robot actions, and GeoDelta, a video-aligned 4D geometric motion field encoding the cross-frame image-plane displacement and depth change of embodiment and object points. By placing embodiment motion and object responses in a shared space, GeoDelta exposes their motion relations to the action expert at inference. Interaction-Salient Attention (InSAttn) uses these relations to focus spatial readout on task-relevant interaction regions, while Transition-Anchored Aggregation (TransAgg) aggregates evidence around changes in their relative motion, such as grasp and release. GeoDelta describes surface motion independently of embodiment-specific controls, providing a shared human--robot quasi-action for action-free pretraining on EgoManip-5000H followed by robot mid-training on Robot-30H. Gnomon4D leads on RoboTwin~2.0, VLABench, and RoboDojo, and outperforms the strongest real-robot baseline by 7.0 and 16.0 percentage points under in-distribution and out-of-distribution settings, respectively.

cs.CV↗

arXiv2026

EvoMemBench: Benchmarking Agent Memory from a Self-Evolving Perspective

Recent benchmarks for Large Language Model (LLM) agents mainly evaluate reasoning, planning, and execution. However, memory is also essential for agents, as it enables them to store, update, and retrieve information over time. This ability remains under-evaluated, largely because existing benchmarks do not provide a systematic way to assess memory mechanisms. In this paper, we study agent memory from a self-evolving perspective and introduce EvoMemBench, a unified benchmark organized along two axes: memory scope (in-episode vs. cross-episode) and memory content (knowledge-oriented vs. execution-oriented). We compare 15 representative memory methods with strong long-context baselines under a standardized protocol. Results show that current memory systems are still far from a general solution: long-context baselines remain highly competitive, memory helps most when the current context is insufficient or tasks are difficult, and no single memory form works consistently across all settings. Retrieval-based methods remain strong for knowledge-intensive settings, whereas procedural and long-term memory methods are more effective for execution-oriented tasks when their stored experience matches the task structure. We hope EvoMemBench facilitates future research on more effective memory systems for LLM-based agents. Our dataset and code are available at https://github.com/DSAIL-Memory/EvoMemBench.

cs.CL↗

arXiv2026

Bidirectional motion of antiferromagnetic skyrmions driven by competing spin torques

Antiferromagnetic skyrmions are swirling topological spin textures with rich dynamics and intriguing transport properties, yet their bidirectional dynamics remain largely unexplored. Here, we investigate the dynamics of antiferromagnetic skyrmions driven by current-induced spin-transfer and spin-orbit torques. We computationally demonstrate that antiferromagnetic skyrmions moving in one direction at low current densities can reverse their motion direction when the driving current is above a threshold. Based on the Thiele approach analysis, we show that this bidirectional motion originates from a change in the relative strengths of two effective forces arising from spin-transfer and spin-orbit torques. Furthermore, exploiting this bidirectional motion on a single racetrack, we design programmable logic gates. Our results not only uncover a hidden mechanism for bidirectional skyrmion motion but also facilitate the development of antiferromagnet-based logic devices.

cond-mat.mes-hall↗

arXiv2026

Structural stability of cylindrical transonic shocks to the three-dimensional Zeldovich-von Neumann-Döring combustion equations

We address the structural stability of transonic shock solutions to the Zeldovich-von Neumann-Döring (ZND) combustion equations in a cylindrical domain under three-dimensional perturbations of the incoming supersonic flow and exit pressure. The perturbed transonic shock is completely free and its strength and position are uniquely determined by the incoming flow and the exit pressure. The solution shares the same regularity for the velocity, the pressure, the entropy and the fraction of unburned gas. The main difficulties are that the shock front is a free boundary to be determined simultaneously with the subsonic flow and the equations for all states are coupled with each other due to the combustion process. One of the key ingredients of the analysis is to reformulate the three-dimensional ZND combustion equations into a deformation-curl system and four transport equations in terms of the deformation tensor and vorticity. Another one is an elaborate decomposition of the Rankine-Hugoniot conditions so that the shock front is uniquely determined by an algebraic equation and an unusual second order differential boundary condition is obtained on the shock front for the first order nonlocal deformation-curl system, from which an oblique boundary condition can be derived after homogenizing the curl system and introducing the potential function.

math.AP↗

arXiv2026

Stride: Scale-Aware Multi-Agent Evolutionary Refinement for Infographic Generation

Infographic generation challenges text-to-image models because a single image must satisfy numerous visual, textual, and semantic constraints. Iterative agentic refinement with a multimodal verifier is a natural remedy, but existing methods face a scale mismatch: local edits cannot correct global composition errors, while full regeneration may destroy already-correct regions. We introduce Stride, a constraint-grounded multi-agent framework that treats refinement over any frozen generator as a history-aware evolution strategy. Stride separates constraint evaluation - whether each requirement is satisfied - from visual grounding - where and how it is violated. It then routes global revisions to a composer subagent and local corrections to a retoucher subagent. Each subagent conducts internal planning and generates multiple candidates, while Stride accepts a candidate only when it improves the verifier score, reducing regression across iterations. Because stochastic generators and imperfect verifiers make any single step unreliable, it is this combination of scale-aware routing, per-step redundancy, and monotonic acceptance that yields stable improvement across iterations. With fully open-weight generator-verifier systems, Stride outperforms proprietary systems including Nano-Banana-2.0 and Seedream-5.0-Pro on BizGenEval, matches GPT-Image-2.0 and surpasses Seedream-5.0-Pro on IGenBench, and achieves a new state-of-the-art score on WISE, while substantially outperforming existing agentic methods across all evaluated benchmarks.

cs.CV↗

arXiv2026

EVISKILL: Grounding Skill Evolution in Replayable Evidence

Continual skill evolution enables LLM agents to accumulate and refine reusable procedural knowledge from interaction experience without updating model parameters. Its effectiveness depends on determining not only what to change, but also why a change is justified and when it should become persistent guidance. However, existing experience-driven methods can lose the behavioral evidence and task contexts supporting edits. Moreover, a global validation outcome provides an incomplete judgment of its constituent changes: locally supported corrections may be discarded with a rejected revision, while evidence may require further experience to inform useful updates. To this end, we introduce EVISKILL, an evidence-driven framework that organizes execution observations into Replayable Evidence Cards and synthesizes edits with explicit links to their supporting contexts. Targeted replay verifies these edits through re-execution and provides feedback for correction. Across epochs, EVISKILL preserves evidence and provisionally retains supported edits for further refinement, while global validation governs their incorporation into the final skill. Experiments on three interactive benchmarks across six LLM backbones demonstrate the effectiveness of this approach.

cs.AI↗

arXiv2026

SubjectAnchor: Subject-Aware Memory-to-Video for Multi-Shot Storytelling

We present SubjectAnchor, a Subject-Aware Memory-to-Video paradigm for multi-shot storytelling in which the current shot is generated by conditioning on explicit visual memories extracted from previous shots. The objective is to preserve subject identity and scene consistency across cuts while retaining the controllability of shot-wise prompting. Built on Wan2.2-I2V-A14B, SubjectAnchor contains three key components: subject-related memory construction, subject-aware temporal rotary position encoding, and memory-aware attention partition. For each target shot, the method constructs a compact memory bank by tracing each required subject to its historical appearance and retrieving the most relevant precomputed keyframes. These memory frames are encoded into the model input as explicit visual conditions, while different subjects are assigned to separated negative temporal slots to reduce identity interference. In addition, memory-aware attention partition regulates the interaction between memory tokens and generated content within a shared backbone. This formulation preserves the appearance anchoring of explicit visual memory while remaining compatible with script-driven shot-by-shot generation. Experiments show that SubjectAnchor improves cross-shot identity consistency over representative memory-based and holistic baselines while maintaining competitive visual quality.

cs.CV↗

arXiv2026

Magnetic Skyrmion Interacting with Optical Skyrmion

Magnetic skyrmions (MSks) and optical skyrmions (OSks) embody topology in matter and in light, respectively. Here we investigate the interaction between a single MSk and an OSk beam. Three distinct nonlinear dynamical modes are identified: rotation, skipping, and trochoidal motion. By decomposing the optical driving force into gradient, orbital-angular-momentum, and spin-angular-momentum contributions, we clarify their respective roles of radial confinement, azimuthal drift, and precessional modulation. The skipping motion arises from the azimuthal asymmetry of the OSk beam and exhibits spatial selectivity originating from the magnetization-polarization coupling between the MSk and OSk. In three dimensions, the coupling acquires a propagation-dependent phase dominated by the differential Gouy phase, which yields $z$-asymmetric skipping trajectories. These results bridge topological particles and topological fields within a unified framework, offering helicity-selective and phase-programmable routes to optomagnonic control.

cond-mat.mes-hall↗

arXiv2026

Distributed Selective Inference for Quantile Regression

We propose a distributed selective inference framework tailored for high-dimensional quantile regression. To enable valid post-selection inference in this context, we address the computational challenge posed by the non-smooth quantile loss via a response-surrogation strategy. This strategy transforms the problem into a penalized least-squares formulation, thereby facilitating distributed selective inference. For valid post-selection inference, a randomized procedure is introduced, in which the Lasso selection event is characterized through the associated Karush-Kuhn-Tucker conditions and the conditional distribution of the aggregated estimator is derived given the selection event. The resulting algorithm requires only three rounds of communication between local machines and the central server. Under standard regularity conditions, we establish the asymptotic validity of the proposed procedure and develop a large-deviation approximation to the selective likelihood for computationally tractable implementation. Simulation studies and a real-data application demonstrate the satisfactory finite-sample performance of the proposed method.

stat.ME↗

arXiv2026

Three-dimensional steady transonic shocks for the relativistic Euler equations

We investigates the transonic shock problem for steady relativistic Euler equations in hemispherical shells. We show the existence and uniqueness of spherically symmetric transonic shock solutions via the shooting method and the monotonicity between the exit pressure and shock position. Furthermore, without any restrictions on background transonic shocks, we prove the existence and stability of transonic shocks under three dimensional perturbations of the exit pressure. The key techniques involve the decoupling the hyperbolic and elliptic components in steady relativistic Euler equations using the deformation tensor and vorticity. To address the coordinate singularities, the ``spherical projection coordinates" combining the spherical coordinates and the stereographic projection is employed. Subsequently, the Rankine-Hugoniot conditions are appropriately reformulated to determine the shock front and derive the boundary conditions on the shock front for a first-order nonlocal deformation-curl system.

math.AP↗

arXiv2026

Sequential Trajectories and Simultaneous Blending: Multi-Emotion Modeling for Instruction-Following TTS

Natural-language instructions enable flexible control of synthesized speech, yet emotional TTS systems primarily model a single utterance-level affect, leaving multi-emotion control underexplored. We study two complementary multi-emotion TTS tasks: emotion trajectory, which spans several ordered affective stages, and emotion blending, in which multiple emotions coexist throughout an utterance. These tasks expose a supervision mismatch: supervised fine-tuning (SFT) does not explicitly evaluate emotion features, while single-emotion rewards provide neither structure-aware feedback for trajectory completion nor pair-aware feedback for blending. We introduce HybridEmo, a post-training framework that initializes both tasks with SFT and then aligns the speech-token policy through Group Relative Policy Optimization using a sample-aware hybrid reward. For trajectory samples, segment-aligned consistency combines average and weakest-stage evidence to preserve the correctness and completeness of prescribed stages. For blending samples, a GMM-based reward combines frame-level support from the union of target-emotion anchors in an offline emotion space with an utterance-level weaker-target margin. Both branches share an ASR reward and are routed within a unified policy. On MultiEmo-Test, HybridEmo significantly improves trajectory correctness and blending intensity, without a noticeable degradation in speaker similarity. Human evaluation prefers HybridEmo to CosyVoice 3 and EmoVoice-0.5B, with nearly balanced preferences against Qwen3-TTS.

cs.CL↗

arXiv2026

Thinking at the Right Size: Amortized Distillation Across Post-Trained LLMs

Practical deployment of large language models (LLMs) requires families of post-trained variants---instruction-tuned, reasoning-tuned, and chat-style models---each at multiple sizes to meet diverse latency and memory budgets. Producing each (variant, size) pair independently is prohibitive, so model families typically span only a handful of coarse-grained sizes per post-trained variant. Boomerang distillation (Kangaslahti et al., 2026) reduces this cost along the size axis for base models. Through model size interpolation, it constructs models of intermediate sizes from a single teacher-student pair without additional training. However, it still treats each post-trained variant as a separate object of optimization. We introduce ADAPT---Amortized Distillation Across Post-Trained LLMs---a framework for amortizing distillation across both axes of a model family: size and post-training variant, producing $L \times K$ models for $L$ interpolated sizes across $K$ post-trained variants with a single distillation run. ADAPT combines two components. First, a two-phase distillation procedure constructs post-trained students through pre-training alignment and supervised fine-tuning distillation, enabling smooth size--performance interpolation on generation and reasoning tasks. Second, weight-delta initialization approximates this construction across post-trained variants by transferring the distillation-induced weight change from the base model to students initialized from different post-trained variants. The resulting continuum of interpolated models also enables adaptive model-size selection at inference time, improving the compute--accuracy trade-off for long-form reasoning tasks.

cs.LG↗

arXiv2026

EmotionDialogCN: A Spontaneous Multimodal Dataset for Mandarin Emotional Dialogue

Face-to-face audiovisual interaction is central to human communication, conveying rich emotional and social cues. However, existing multimodal dialogue datasets remain limited by inadequate emotion annotations, poor emotional diversity, and small scale. We introduce EmotionDialogCN, a large-scale audiovisual-emotional dataset designed to capture authentic face-to-face communication. It contains 21,880 dialogue sessions performed by 119 professional actors across 20 everyday scenarios, covering 18 emotion categories with over 400 hours of recordings, the largest and most comprehensive dataset of its kind. A novel data collection framework minimizes equipment interference, enabling natural and nuanced emotional expressions. EmotionDialogCN achieves an emotion distribution deviation of 0.64 from real human emotion statistics (versus 5.65 for prior datasets) and consistent subject framing (52-59% frame occupancy). Together, these properties translate into stable unimodal and multimodal performance across acoustic, lexical, and visual modalities, with fusion results further underscoring strong multimodal alignment and cross-modal complementarity.

cs.CV↗

arXiv2026

BMDS-Net:Deployment-aware multi-modal brain tumor segmentation with adaptive fusion,decoder regularization,and Bayesian calibration

Multi-modal MRI enables detailed brain tumor sub-region segmentation, but clinical deployment remains affected by missing sequences,boundary errors, and overconfident predictions. We present BMDS-Net, a two-stage Swin UNETR framework that combines adaptivemodality fusion, boundary-aware decoder regularization, and efficient Bayesian calibration. The deterministic stage adds zero-initializedmultimodal contextual fusion (MMCF) and residual-gated deep decoder supervision (DDS) to modulate modality contributions andregularize hierarchical boundary reconstruction. The second stage replaces only the final convolutional layer with a Bayesian layer,enabling Monte Carlo uncertainty estimates with limited additional training cost. On BraTS 2021 (1,251 cases), BMDS-Net achieved Dice scores of 0.9293, 0.9098, and 0.8675 for WT, TC, and ET, with HD95 values of 2.27, 2.22, and 3.27 mm. Configuration comparisons showed that auxiliary supervision favored complete-input segmentation, whereas the complete MMCF+gated-DDS configuration was most robust under single-modality removal. Bayesian BMDS-Net achieved an expected calibration error of 0.0037, close to a three-model ensemble (0.0035) at approximately 60% lower training cost. On BraTS 2020, the same protocol preserved the leading internal ranking of the complete configuration among five Swin UNETR-controlled variants. The implementation code is available at https://github.com/RyanZhou168/BMDS-Net.

cs.CV↗

arXiv2026

CoBa: Cost-Effective Test-Time Scaling via Compute-Balanced Routing

Test-time scaling is often implemented by spending more compute along one axis: sampling more solutions, extending a chain of thought, or applying a stronger evaluator. Under a fixed inference budget, these choices compete. This paper formulates test-time reasoning as a compute-allocation problem in which a system must decide whether the next unit of compute should be spent on generation, verification, or stopping. We introduce CoBa, a compute-balanced routing policy that first obtains a small set of candidates, applies cheap verification broadly, and routes uncertain or high-value candidates to stronger verification. On 3,129 example-generator evaluations spanning MATH-500, AIME 2024/2025, AMC 2023, and procedural symbolic reasoning, CoBa-Routed-Strong reaches 85.13% macro accuracy, statistically matching a self-evaluation weighted-voting proxy at 85.20% while using 49.1% fewer parameter-weighted tokens. It also matches best-of-16 majority voting within 0.01 macro-accuracy points while using 58.9% fewer parameter-weighted tokens; paired tests retain a small best-of-16 edge at substantially higher cost. Paired bootstrap tests show significant gains over single-sample decoding, while the remaining gap to the pool oracle exposes headroom for sharper routing. For local reasoning systems, test-time scaling becomes a question of where the next computation is most valuable.

cs.AI↗

arXiv2026

TEPA: Revoking Stale Memories for Conflict-Robust Language Agents

Long-term memory enables language agents to reuse past facts, preferences, and task experience. Persistence also creates a central falsifiability problem: when the world changes, stale memories can remain retrievable and pollute the prompt. We characterize this failure mode as memory pollution: degradation caused by active memories that newer conflicting evidence has superseded. We introduce TEPA, a revocable evidence-memory mechanism that makes validity an explicit state of memory. TEPA represents observations as keyed precedents and revokes active precedents when fresh evidence contradicts them under the same key, allowing retrieval to draw from current evidence while preserving revoked history for audit. Across controlled hidden-regime drift, real file-backed executable drift, and preference-update streams, revocation prevents stale active memory from remaining in the retrieval set after reversal. In controlled drift over 50 seeds, append-only and last-write-wins memory fell below no memory during full reversal (append-only and last-write-wins both 0.210, no memory 0.309, TEPA 0.950), and the same pattern reproduced under real file execution (append-only 0.203, no memory 0.298, TEPA 0.950). On clean MemoryAgentBench SH-6k, TEPA matches a strong last-write-wins cache, confirming that current-key replacement is the decisive operation for single-hop fact consolidation. Boundary tests on multi-hop and very long-context MemoryAgentBench settings expose retrieval-chain and context-selection bottlenecks beyond fact-level validity tracking. Together, these results establish lifecycle revocation as a core memory operation for agents that must falsify, audit, and later re-promote evolving knowledge.

cs.AI↗

arXiv2026

The Impossibility Triangle of Long-Context Modeling

We identify and prove a fundamental trade-off governing long-sequence models: no model can simultaneously achieve (i) per-step computation independent of sequence length (Efficiency), (ii) state size independent of sequence length (Compactness), and (iii) the ability to recall a number of historical facts proportional to sequence length (Recall). We formalize this trade-off within an Online Sequence Processor abstraction that unifies Transformers, state space models, linear recurrent networks, and their hybrids. Using the Data Processing Inequality and Fano's Inequality, we prove that any model satisfying Efficiency and Compactness can recall at most O(poly(d)/log V) key-value pairs from a sequence of arbitrary length, where d is the model dimension and V is the vocabulary size. We classify 52 architectures published before March 2026 into the triangle, showing that each achieves at most two of the three properties and that hybrid architectures trace continuous trajectories in the interior. Experiments on synthetic associative recall tasks with five representative architectures validate the theoretical bound: empirical recall capacity lies strictly below the information-theoretic limit, and no architecture escapes the triangle.

cs.CL↗

arXiv2026

Motion Beyond Morphology: Bootstrapping Cross-Category Motion Transfer from Abstract Motion Representations

Video motion transfer aims to animate a target object using dynamics from a reference video. Existing formulations largely rely on fixed structural correspondence, which becomes ill-defined when reference and target objects differ substantially in morphology, articulation, or deformation mechanisms. We introduce Motion Beyond Morphology, a perspective that seeks to transfer motion beyond fixed structural correspondence, by preserving dynamics that remain meaningful across different target morphologies. To realize this, we propose a two-stage framework. Stage~I learns complementary multi-granularity abstract motion views and uses them to bootstrap cross-category video pairs that preserve transferable dynamics across diverse morphologies. Stage~II internalizes this supervision into direct reference-video-conditioned generation, removing the need for explicit motion extraction at inference. We further introduce OpenVMT-Dataset and OpenVMT-Bench for training and evaluating image- and text-conditioned motion transfer across Same, Near, and Far category gaps. Extensive experiments demonstrate state-of-the-art motion fidelity and target preservation. Project page: https://miniz233.github.io/MotionBeyondMorphology/

cs.CV↗