arXiv subjects

Can Cui

Publications and source records attributed to Can Cui.

At least 19 recordsLinked to original sources

arXiv2026

Designing Versatile Samples for Learned Trajectory Scoring

Many current end-to-end driving policies emit a pool of candidate trajectories and select one, which makes selection a separable component: a scorer can be retrained while the planner, its backbone, and its trajectory generator all stay frozen. However, many strong planners concentrate their proposals around safe mode, providing limited supervision near decision boundaries. In this work, we design a training dataset that provides more informative supervision for the scorer. In particular, we construct two generators that perturb the logged human trajectory along the two axes a vehicle can be displaced: laterally toward the drivable boundary and longitudinally toward a leading vehicle. The designed dataset produces more informative positive and negative samples than the base planner's proposal pool. We attach a transformer-based scorer to two frozen generative planners, DiffusionDrive and MeanFuser, and train it on the NAVSIM navtrain dataset. The results of the experiments show that we achieve 90.1 EPDMS on DiffusionDrive and 90.4 EPDMS on MeanFuser when using ResNet-34, with 0.4 and 0.3 EPDMS respectively, from the designed training dataset.

cs.RO↗

arXiv2026

CoRE: Weakly Supervised Coarse-to-Fine Risk Evidence Learning in Driving Videos

Perceived risk in driving evolves over time and may be supported by specific scene entities, yet supervision is typically limited to coarse video-level judgments. Learning \emph{when} supporting evidence emerges and \emph{which entities} support a risk predictor would ordinarily require costly temporal- and entity-level annotations. We introduce \textbf{CoRE}, a weakly supervised coarse-to-fine framework that learns fine-grained prediction support from coarse video supervision. CoRE first trains a video-level predictor and then freezes it. Structured interventions over candidate temporal regions or entity tracks measure how each candidate changes the coarse prediction, producing graded prediction-effect targets. These targets are distilled into a student that directly predicts temporal and entity support from the original video, without requiring interventions at inference. We evaluate this learning principle across three complementary settings: RISEE tests perceived-risk support from subjective clip-level judgments without temporal or entity-level risk annotations; DoTA provides independent temporal event annotations for evaluating weakly supervised traffic-anomaly localization; and UCF-Crime tests whether the same coarse-to-fine mechanism extends to a standard non-driving anomaly-detection benchmark. Across these settings, CoRE learns informative fine-grained support from coarse supervision, with strong temporal localization on DoTA and competitive performance on UCF-Crime. These results show that coarse video predictions can provide useful supervision for recovering the fine-grained evidence supporting them, without requiring corresponding fine-grained labels.

cs.CV↗

arXiv2026

GlanceWAM: Sparse Test-Time Imagination for World-Action Models

Video generative models provide rich physical priors for robot learning, yet existing world-action models (WAMs) face a fundamental trade-off: synchronous video generation at control rate is latency-prohibitive, while abandoning test-time visual imagination sacrifices task success. We show that visual imagination achieves both real-time inference and superior success rates when generated asynchronously off the critical path and consumed directly in latent space. We introduce GlanceWAM, which decouples imagination from control within a single video DiT: an asynchronous proposer glances ahead on a slow clock to imagine a single lookahead frame seconds into the future in the background, while an action head decodes action chunks at control rate (48 ms) purely in latent space without blocking. Enabled by a non-interfering attention mask that isolates video representations and staleness-robust horizon training that accommodates asynchronous lookahead aging, GlanceWAM breaks the speed-success dilemma. Trained purely on demonstrations, it attains 72.2% on the 24-task RoboCasa kitchen benchmark (surpassing synchronous Cosmos Policy at 67.1% and imagination-free co-training at 64.4%) and 99.0% on LIBERO, executing at 48 ms per chunk on an NVIDIA A100 GPU (24x faster than synchronous baselines). Code is available at https://github.com/linhanwang/GlanceWAM.

cs.CV↗

arXiv2026

How Can Driving World Models Do Counterfactual Prediction?

Driving world models are often interpreted as counterfactual simulators for observed driving episodes: given a factual driving log, they are asked what would have happened under an alternative ego action. In this paper, we identify a fundamental mismatch between this goal and direct action-conditioned prediction. The direct prediction uses the shared history and the alternative action but not the factual continuation observed after that history. It can therefore generate a plausible future without preserving what actually happened in this episode. We formalize this gap using the causal recipe of abduction, action, and prediction and study it in a setting with a short time horizon, where the alternative ego action does not alter how surrounding agents evolve. To make the gap measurable, we construct a controlled simulation benchmark with factual outcomes and matched counterfactual outcomes. Across two representative world models, direct predictions fail to match the counterfactual ground truth, supporting our analysis. As a constructive check of this analysis, we introduce a deliberately simple, training-free pipeline that moves observed evidence into the counterfactual view and lets the frozen model complete what remains unknown. Even this simple construction raises the overall recovered fraction substantially and reduces perceptual distance to the matched counterfactual on both models. We hope this work draws attention to this gap and motivates better counterfactual prediction methods for driving world models.

cs.CV↗

arXiv2026

Momentum-Selective Two-Component Excitations in Electron-Doped Mott Insulators

Experimental studies reveal a striking asymmetry in low-energy single-particle excitations between electron- and hole-doped cuprates. Electron-doped cuprates display a nontrivial dichotomy: Fermi-liquid-like behavior (suggesting weaker electronic correlations) coexists with correlation-driven features typical of hole-doped systems. This dual nature challenges a unified description within a doped Mott insulator framework. The present work addresses this issue by establishing that the ground-state wave function of the $t$-$t'$-$J$ model generically possesses a two-component structure, comprising a coherent quasiparticle and an incoherent composite component. The kinetic energy arises from both the intrinsic propagation of the coherent quasiparticle and the resonance between these components. Using variational Monte Carlo at the level of a single hole, we show that for hole doping ($t'<0$), this resonance between components dominates and concentrates in the nodal region at low energies. This emergent propagation induced by resonance can be physically interpreted as originating from the recombination of fractionalized degrees of freedom, which drives various phenomena associated with strong correlations. Conversely, for electron doping ($t'>0$), the coherent quasiparticle propagation, which exhibits conventional properties of a Fermi liquid, is selectively enhanced in the antinodal region at low energies. This produces a separation in momentum space for systems with electron doping: the antinodal spectral weight at low energies is governed by the coherent quasiparticle, fundamentally differing from the nodal region, which remains dominated by the incoherent composite component. Motivated by such a structure and guided by experimental observations, we propose a phenomenological Green's function at finite doping, yielding spectral features consistent with experiments.

cond-mat.str-el↗

arXiv2026

SAGEAgent: A Self-Evolving Agent for Cost-Aware Modality Acquisition in Multimodal Survival Prediction

Does every cancer patient truly need a complete diagnostic workup for accurate survival prediction? In multimodal clinical oncology, diagnostic modalities follow a clinically mandated order of escalating burden -- from demographics collected at intake to genomic profiling requiring specialized tissue analysis. Current multimodal survival methods either assume all modalities are available or passively handle missing data, but none actively reason about whether acquiring the next modality is justified for a given patient along this ordered workflow. We formulate this as a sequential decision problem and propose SAGEAgent (Sequential Acquisition Guided by Experience), a self-evolving LLM-based clinical agent that decides which diagnostic modalities to acquire for each patient, balancing predictive accuracy against clinical invasiveness. SAGEAgent reasons about each patient's evolving diagnostic state through clinical tools that translate numerical predictions into text, an episodic memory that retrieves similar past cases, and a semantic memory that accumulates reusable decision patterns from experience. Experiments on a glioma cohort combining TCGA-LGG, TCGA-GBM, and BraTS with four diagnostic modalities demonstrate that SAGEAgent achieves competitive survival prediction accuracy while reducing average acquisition burden by 55%.

cs.AI↗

arXiv2026

Post-Training in End-to-End Autonomous Driving

End-to-end models that map multimodal inputs directly to future trajectories/maneuvers have emerged as an increasingly prominent research paradigm in autonomous driving. This class of models includes both Vision-Language-Action models and trajectory-generative planners. Unlike classic machine learning applications, autonomous vehicles operate in safety-critical and interaction-intensive environments where traditional open-loop imitation of expert demonstrations is not sufficient to ensure reliability. In particular, small execution errors can accumulate over time, while recovery behaviors are scarce in training data. In addition, long-horizon objectives such as safety and driving comfort are not captured by pointwise labels either. These limitations have motivated a shift toward post-training techniques, which further refine driving policies beyond pure imitation. This survey presents a unified view of post-training for autonomous driving by defining its scope and organizing the existing literature into four major families based on the form of supervision they use. For each family, we discuss its capabilities, limitations, and open challenges. We aim to facilitate a systematic understanding of this emerging area and stimulate future research on reliable and efficient post-training for autonomous driving.A collection of related papers is available at https://github.com/RYNing/Awesome-Post-Training-In-Autonomous-Driving-Papers.

cs.CV↗

arXiv2026

Ring Spacing from a Fourth-Order Radial Feedback Green Function in a Keplerian Accretion Disk

ALMA observations of protoplanetary disks reveal ubiquitous concentric ring structures whose origin remains debated. We present an exactly solvable local model for ring spacing in a Keplerian disk. The passive advection--diffusion problem with a general vertical diffusivity profile separates into a vertical Sturm--Liouville spectrum and a radial modified-Bessel Green function. This passive kernel is smooth and does not by itself generate a periodic ring train. We therefore introduce a minimal fourth-order radial feedback closure for the ring-averaged surface density. For a localized steady ring source, the resulting Green function has a damped oscillatory exterior branch whenever the decaying spatial roots are complex. Under an observationally motivated AU scaling, the same dimensionless solution gives an illustrative spacing of order $12$~AU, distinct from the fastest-growing temporal wavelength. The model separates the passive transport kernel from the feedback mechanism that selects the observable radial scale. Because this mechanism is internal to the disk, it does not require pre-existing planets. The vertical diffusivity profile affects the background kernel but is not the source of the Green-function oscillation.

astro-ph.EP↗

arXiv2026

OmniV2X: A Generative Foundation Planner for Efficient End-to-End Cooperative Driving

We present OmniV2X, a generative foundation model for vehicle-to-everything (V2X) cooperative driving. The model directly interprets independent context sequences comprising multi-modal and multi-agent observations. The new design mitigates the computational cost of dense 3D perception, the vulnerability to data scarcity in cooperative scenarios, and the poor compliance with standardized messaging in existing methods that fuse multi-modal inputs into a shared representation. For training, we present an end-to-end supervised pipeline using a downstream trajectory generation loss, in which a high-capacity generative sequence planner implicitly learns to steer the model and leverage multi-modal inputs via cross-attention injection. As a foundation model, we demonstrate that OmniV2X pre-trained on large-scale single-agent planning datasets can efficiently adapt to cooperative environments by integrating the conditioning context with lightweight, standard-compliant V2X tokens. Evaluated on the DAIR-V2X-Seq dataset, OmniV2X outperforms existing end-to-end cooperative driving baselines, achieving state-of-the-art performance with less than 10% of the fine-tune V2X dataset and less than 1% of the communication bandwidth. We conduct comprehensive evaluations to demonstrate its computational efficiency and robustness under real-world constraints.

cs.RO↗

arXiv2026

LoopFM: Learning frOm HistOrical RePresentations of Foundation Model for Recommendation

Knowledge distillation (KD) transfers a single scalar prediction from a large foundation model (FM) to compact vertical models (VMs), suffering from diminishing transfer ratio -- the fraction of FM improvement captured by the VM -- as a single scalar cannot convey the rich intermediate knowledge that larger FMs learn. To address this bottleneck, we propose LoopFM (Learning frOm HistOrical RePresentations of FM), a framework that opens a high-bandwidth transfer channel by structuring FM intermediate embeddings as input features (e.g., user history sequence) for downstream VMs, without requiring real-time FM inference at serving and architectural coupling between FM and VM. We provide a theoretical framework for LoopFM with a gain decomposition and transfer-ratio analysis. On three public benchmarks, LoopFM demonstrates strong AUC improvements (e.g., 6%+ on TaobaoAd) and complementary knowledge transfer capability with KD. On industrial-scale systems (billions of examples, trillion-parameter FMs), LoopFM approximately doubles the knowledge transfer ratio on top of KD, delivering a +0.5% conversion improvement in the first half after its initial launch, and +1.03% and +1.22% conversion improvement from two individual launches in the subsequent half.

cs.LG↗

arXiv2026

ICR-Drive: Instruction Counterfactual Robustness for End-to-End Language-Driven Autonomous Driving

Recent progress in vision-language-action (VLA) models has enabled language-conditioned driving agents to execute natural-language navigation commands in closed-loop simulation, yet standard evaluations largely assume instructions are precise and well-formed. In deployment, instructions vary in phrasing and specificity, may omit critical qualifiers, and can occasionally include misleading, authority-framed text, leaving instruction-level robustness under-measured. We introduce ICR-Drive, a diagnostic framework for instruction counterfactual robustness in end-to-end language-conditioned autonomous driving. ICR-Drive generates controlled instruction variants spanning four perturbation families: Paraphrase, Ambiguity, Noise, and Misleading, where Misleading variants conflict with the navigation goal and attempt to override intent. We replay identical CARLA routes under matched simulator configurations and seeds to isolate performance changes attributable to instruction language. Robustness is quantified using standard CARLA Leaderboard metrics and per-family performance degradation relative to the baseline instruction. Experiments on LMDrive and BEVDriver show that minor instruction changes can induce substantial performance drops and distinct failure modes, revealing a reliability gap for deploying embodied foundation models in safety-critical driving.

cs.CL↗

arXiv2026

Efficient and Explainable End-to-End Autonomous Driving via Masked Vision-Language-Action Diffusion

Large Language Models (LLMs) and Vision-Language Models (VLMs) have emerged as promising candidates for end-to-end autonomous driving. However, these models typically face challenges in inference latency, action precision, and explainability. Existing autoregressive approaches struggle with slow token-by-token generation, while prior diffusion-based planners often rely on verbose, general-purpose language tokens that lack explicit geometric structure. In this work, we propose Masked Vision-Language-Action Diffusion for Autonomous Driving (MVLAD-AD), a novel framework designed to bridge the gap between efficient planning and semantic explainability via a masked vision-language-action diffusion model. Unlike methods that force actions into the language space, we introduce a discrete action tokenization strategy that constructs a compact codebook of kinematically feasible waypoints from real-world driving distributions. Moreover, we propose geometry-aware embedding learning to ensure that embeddings in the latent space approximate physical geometric metrics. Finally, an action-priority decoding strategy is introduced to prioritize trajectory generation. Extensive experiments on nuScenes and derived benchmarks demonstrate that MVLAD-AD achieves superior efficiency and outperforms state-of-the-art autoregressive and diffusion baselines in planning precision, while providing high-fidelity and explainable reasoning.

cs.CV↗

arXiv2026

Minimal loop currents in doped Mott insulators

For the $t$-$J$ model, variational wave functions can generally be constructed based on an accurate description of antiferromagnetism (AFM) at half-filling and an exact phase-string sign structure under doping. The single-hole-doped and two-hole-doped states, as determined by variational Monte Carlo (VMC) simulations, display sharply contrasting behaviors. The single-hole state constitutes a ``cat state'' that resonates strongly between a quasiparticle component and a local loop-current component, with approximately equal weights. In the ground state, the quasiparticle spectral weight $Z_{\mathbf{k}}$ peaks at momenta $\mathbf{k}_0 \equiv (\pm\frac{\pi}{2},\pm\frac{\pi}{2})$. The total-energy dispersion versus $\mathbf{k}$ agrees remarkably well with the Green function Monte Carlo results. However, Landau's one-to-one correspondence hypothesis for quasiparticles breaks down here with the incoherent component exhibiting intrinsic magnetization originating from a minimal $2\times2$ loop current that forms a $4\times4$ pattern on the square lattice--a finding in excellent agreement with density matrix renormalization group (DMRG) calculations. In the two-hole ground state, a new pairing mechanism is revealed: the two holes are automatically fused into a tightly bound object consisting of an incoherent $d_{xy}$ pairing along the diagonal direction by compensating the local loop currents. This hole pair is again a ``cat state'' that resonates strongly between the incoherent $d_{xy}$ and a coherent $d_{x^2-y^2}$ Cooper channel to gain substantial hopping energy. Its size extends over an area of about $4\times 4$ lattice spacings, much smaller than the divergent AFM correlation length, implying that it should survive as a minimal superconducting building block even in the dilute doping regime. Experimental implications and the generalization to the finite-doping case are briefly addressed.

cond-mat.str-el↗

arXiv2025

Equation of state for the hard sphere fluids

Based on the survey of the literatures on the new improvements on the equation of state (EOS) for the hard sphere fluids, we here compare lots of different EOSs and present a very accurate equation of state for this kind of fluids. The new equation is built up on the basis of (1) the best estimated virial coefficients B5-B11 by Tian et al. [ Phys. Chem. Chem. Phys., 2019, 21, 13070] and (2) the newest numerical simulation data of the compressibility factor versus the density by Pieprzyk et al. [Phys. Chem. Chem. Phys., 2019, 21, 6886]. Our results show that this equation is accurate in not only the stable density range but also the metastable density range with the proper closest packing fraction pole, and well derives the predictive values of the high order virial coefficients B13-B16.

cond-mat.soft↗

arXiv2025

Hierarchical Task Offloading and Trajectory Optimization in Low-Altitude Intelligent Networks Via Auction and Diffusion-based MARL

The low-altitude intelligent networks (LAINs) emerge as a promising architecture for delivering low-latency and energy-efficient edge intelligence in dynamic and infrastructure-limited environments. By integrating unmanned aerial vehicles (UAVs), aerial base stations, and terrestrial base stations, LAINs can support mission-critical applications such as disaster response, environmental monitoring, and real-time sensing. However, these systems face key challenges, including energy-constrained UAVs, stochastic task arrivals, and heterogeneous computing resources. To address these issues, we propose an integrated air-ground collaborative network and formulate a time-dependent integer nonlinear programming problem that jointly optimizes UAV trajectory planning and task offloading decisions. The problem is challenging to solve due to temporal coupling among decision variables. Therefore, we design a hierarchical learning framework with two timescales. At the large timescale, a Vickrey-Clarke-Groves auction mechanism enables the energy-aware and incentive-compatible trajectory assignment. At the small timescale, we propose the diffusion-heterogeneous-agent proximal policy optimization, a generative multi-agent reinforcement learning algorithm that embeds latent diffusion models into actor networks. Each UAV samples actions from a Gaussian prior and refines them via observation-conditioned denoising, enhancing adaptability and policy diversity. Extensive simulations show that our framework outperforms baselines in energy efficiency, task success rate, and convergence performance.

cs.LG↗

arXiv2025

FSDAM: Few-Shot Driving Attention Modeling via Vision-Language Coupling

Understanding not only where drivers look but also why their attention shifts is essential for interpretable human-AI collaboration in autonomous driving. Driver attention is not purely perceptual but semantically structured. Thus, attention shifts can be learned through minimal semantic supervision rather than dense large-scale annotation. We present \textbf{FSDAM} (\textbf{F}ew-\textbf{S}hot \textbf{D}river \textbf{A}ttention \textbf{M}odeling), a framework that achieves joint spatial attention prediction and structured explanation generation using 90 annotated examples. Our key insight is to decompose attention into an explicit reasoning representation, including scene context, current focus, anticipated next focus, and causal explanation, and to learn next-focus anticipation through minimal-pair supervision. To address task conflict and large sample requirements of existing models, and to mitigate task interference under limited data, we introduce a novel dual-pathway architecture in which separate modules handle spatial prediction and caption generation. In addition, we use a training-only vision-language alignment mechanism that injects semantic priors into spatial learning without increasing inference complexity, mitigating task interference under few-shot training. Despite extreme data scarcity, FSDAM achieves competitive performance in gaze prediction, and generates coherent, context-aware structural reasoning for improved interpretability. The model further demonstrates strong zero-shot generalization across multiple driving benchmarks.

cs.CV↗

arXiv2025

DRO-Based Computation Offloading and Trajectory Design for Low-Altitude Networks

The low-altitude networks (LANs) integrating unmanned aerial vehicles (UAVs) and high-altitude platforms (HAPs) have become a promising solution for the rising computation demands. However, the uncertain task sizes and high mobility of UAVs pose great challenges to guarantee the quality of service. To address these issues, we propose an LAN architecture where UAVs and HAPs collaboratively provide computation offloading for ground users. Moreover, the uncertainty sets are constructed to characterize the uncertain task size, and a distributionally robust optimization problem is formulated to minimize the worst-case delay by jointly optimizing the offloading decisions and UAV trajectories. To solve the mixed-integer min-max optimization problem, we design the distributionally robust computation offloading and trajectories optimization algorithm. Specifically, the original problem is figured out by iteratively solving the outerlayer and inner-layer problems. The convex outer-layer problem with probability distributions is solved by the optimization toolkit. As for the inner-layer mixed-integer problem, we employ the Benders decomposition. The decoupled master problem concerning the binary offloading decisions is solved by the integer solver, and UAV trajectories in the sub-problem are optimized via the successive convex approximation. Simulation results show the proposed algorithm outperforms traditional optimization methods in balancing the worst-case delay and robustness.

cs.CE↗

arXiv2025

Two dimensional anisotropic mean curvature flow with contact angle condition

In this paper, we study surfaces which evolve by anisotropic mean curvature flow with contact angle boundary condition over a strictly convex domain in $\mathbb{R}^2$. We establish a prior gradient estimate for smooth solutions to this boundary value problem. The same approach can also handle Dirichlet boundary condition in $\mathbb{R}^n$, $n\geq 2$. For both problems, we prove that the solutions converge to one that is translation invariant in time.

math.AP↗