arXiv subjects

Haoran Xu

Publications and source records attributed to Haoran Xu.

At least 19 recordsLinked to original sources

arXiv2026

Aligned Delta-Triplane Transformers as Occupancy World Models

Occupancy World Models (OWMs) aim to predict future 3D occupancy scenes from historical observations and future ego motions, providing a useful world simulation tool for autonomous driving. Existing methods usually rely on large networks to implicitly align multi-frame historical states and predict future occupancy in a full-state manner, which is costly and redundant because most scene regions remain unchanged over short time intervals. In this paper, we propose Aligned Delta-Triplane Transformer (ADTT), a compact 4D OWM that explicitly aligns historical triplanes with ego-motion compensation and predicts only future scene changes. Based on the aligned triplane prior, a query-conditioned Transformer predicts residual changes, where ego-motion queries guide controllable future motion and learnable external queries capture scene changes beyond ego motion. The predicted changes are added to the aligned prior and decoded into future occupancy scenes. Experiments on Occ3D-nus show that ADTT achieves state-of-the-art forecasting performance with fewer parameters and higher FPS. Our code is publicly available online: https://github.com/kyoran/ADTT.

cs.RO↗

arXiv2026

UniRA: Unified Representation Alignment and Auto-Encoding for Efficient Diffusion Generation

Diffusion transformers (DiTs) have become the dominant architecture for high-fidelity image synthesis, yet they suffer from slow training convergence and high computational cost. Two recent lines of work address this from complementary angles: Representation Alignment (REPA) accelerates DiT training by aligning intermediate features with a pretrained external representation model (e.g., DINOv2), while Representation Auto-Encoders (RAE) replace the pixel-space VAE with a compact representation-space tokenizer learned from self-supervised features. However, these approaches are developed and applied in isolation: REPA still relies on a pixel-space VAE for latent diffusion, and RAE does not exploit the alignment signal that accelerates the diffusion backbone. We present \textbf{UniRA}, a unified framework that couples representation alignment and representation auto-encoding into a single objective. UniRA learns a representation-space tokenizer whose output serves simultaneously as the diffusion target and the alignment anchor, so the tokenizer, the diffusion backbone, and the external representation model are co-optimized. We give a theoretical result showing that this joint optimization preserves the representation quality of the external model while accelerating diffusion convergence, and we derive a principled weighting that balances the reconstruction, alignment, and generation objectives. On ImageNet class-conditional generation at 256*256 and 512*512, UniRA matches or surpasses state-of-the-art DiTs (SD3, REPA, RAE) in FID while reducing training steps by up to 5*. We further show that the unified representation space supports improved zero-shot classification and linear probing, confirming that generation and understanding benefit from a shared, aligned latent space.

cs.CV↗

arXiv2026

Direction-Aware Token Credit Assignment with Reward-Model Gradients

GRPO shares a response-level advantage across tokens, leaving unresolved how strongly each token should contribute to learning. Updates induced by the shared advantage reinforce or suppress the sampled token at each position, but these updates do not always point toward increasing reward at the token level. This motivates a work-based view of credit: just as work depends on the component of force along a displacement, token credit should reflect the local reward change along the policy update direction. We introduce Direction-Aware Token Credit Assignment (DATC), which maps frozen reward-model gradients into policy logit space and allocates weights by evaluating the alignment between the reward direction and GRPO's signed sampled-action update direction, while preserving the sign and mean of the response-level advantage. DATC outperforms GRPO in all four settings across two model families.

cs.LG↗

arXiv2026

NeoWorld-3: From Pixels to Code-Defined Physical Worlds

Simulation-ready world reconstruction demands physical interactivity beyond visual realism, which existing black-box 3D generation fails to deliver. We present NeoWorld-3, a coding-agent framework that reconstructs interactive 3D worlds from visual observations via tunable, executable programs. At its core, NeoSDK provides the underlying computational engine for low-level geometries, part composition, differentiable parameter refinement, and physics validation. At the same time, NeoMCP exposes these capabilities as structured, tool-callable APIs for vision-language agents to visually inspect, edit code, and iteratively refine programs. Through a hierarchical coordinator-worker architecture, the framework scales seamlessly from standalone articulated objects to complex compositional scenes. Evaluations across object-level and scene-level reconstruction benchmarks demonstrate that NeoWorld-3 achieves state-of-the-art geometric fidelity and physical readiness. Crucially, manipulation trajectories planned within our reconstructed twins achieve successful zero-shot execution on a physical dual-arm robot, completing the Real-Sim-Real loop.

cs.CV↗

arXiv2026

SEDM: Selective Self-Evolving Distributed Memory for Agents

Agents in a society can reuse each other's experience, but indiscriminate memory use increases processing cost and can reduce task success. We present SEDM (Selective Self-Evolving Distributed Memory), a framework that selects both what agents share and when recipients use it. On the sender side, SEDM admits memories through paired execution tests, ranks them by relevance and measured utility, and abstracts them for cross-domain reuse. On the recipient side, Calibrated Memory Gating (CMG) predicts success with and without memory in a single forward pass, learning from paired outcomes including ties. It injects memory only when the calibrated success gain outweighs a context-cost penalty. SEDM exceeds the strongest evaluated baseline on LoCoMo open-domain questions by 15.1 F1 points and outperforms G-Memory on FEVER and HotpotQA with 16-32% fewer prompt tokens. In a controlled HotpotQA study, CMG achieves success comparable to always using memory with 18% fewer agent input tokens. Within the full SEDM pipeline on LoCoMo, CMG outperforms both always-use and no-memory policies for two SEDM memories while using 19-31% fewer input tokens than always-use. These results show that selective use can recover value from shared memories that hurt performance when injected indiscriminately.

cs.AI↗

arXiv2026

SMI: Ontology-Grounded Semantic IDs for Hierarchy-Aware Medical Concept Representation

Biomedical language models (LMs) provide strong representations for clinical tasks, but they often fail to capture the hierarchical structure among medical concepts. This limitation is especially pronounced for rare and fine-grained clinical codes in long-tailed vocabularies, where hierarchy-agnostic LM representations become less informative as concept frequency decreases. We introduce Semantic Medical Identifier (SMI), a training-free framework for constructing ontology-grounded Semantic IDs and hierarchy-aware medical concept representations. For each diagnosis, procedure, or medication concept, SMI defines its Semantic ID as the coarse-to-fine path through an expert-defined medical ontology. SMI then aggregates frozen biomedical LM embeddings across the corresponding ontology levels to form hierarchy-aware medical concept representations that preserve coarse-to-fine clinical semantics. The base SMI uses uniform aggregation across ontology levels, while a level-weighted variant selects global level weights on validation data. Our theoretical analysis shows that SMI yields more informative medical concept representations than hierarchy-agnostic LM embeddings, particularly for rare and fine-grained clinical concepts. Experiments show that SMI captures meaningful hierarchical structure and improves clinical relation prediction. On MIMIC-IV, SMI improves mortality prediction, readmission prediction, drug recommendation, and rare drug prediction. On eICU, SMI yields more robust patient representations under cross-hospital distribution shift.

cs.CL↗

arXiv2026

Trajectory-Refined Distillation

On-policy distillation (OPD) has become a central post-training tool for large language models (LLMs), providing dense per-token teacher supervision along the student's own rollouts. In this work, we identify a common structural cause underlying OPD, which we call prefix failure. Under prefix failure, dense per-token supervision induces a bimodal teacher mixture and fragmented gradients that token-level loss truncation or reweighting fail to address. We thus propose Trajectory-Refined Distillation (TRD), a trajectory-level correction method that revises the student's rollout under the teacher guidance while within on-policy support. By correcting problematic prefixes before distillation, TRD mitigates prefix failure at its source. Moreover, TRD improves exploration by exposing the student to alternative valid derivations under teacher guidance, even when the original rollouts are already correct. TRD can also be applied to on-policy self-distillation (OPSD), a parameter-sharing variant that uses the student model conditioned on privileged information as the teacher. TRD improves average math accuracy while remaining competitive on code. Under OPSD, it solves $9/23$ AMOBench questions left unsolved by the Qwen3-8B base model, nearly doubling the strongest dense-KL baseline. Code is available at https://github.com/louieworth/trd.

cs.AI↗

arXiv2026

SIMPLE: Semantic Structure-Aligned Prompt Learning for Open-Vocabulary Audio-Visual Event Localization

Open-vocabulary audio-visual event localization aims to infer explicit segment-level categories for both seen and unseen audio-visual events. Existing solutions rely heavily on hand-crafted prompt templates, whose predictions are highly sensitive to subtle wording variations. Learnable prompts remove this manual tuning burden, but when trained solely on seen events, they tend to overfit to seen categories and distort the inter-event semantic topology required for generalization to unseen categories, a failure we term semantic topology drift. In this paper, we propose SIMPLE, a semantic structure-aligned prompt learning framework that aligns cross-category semantic topology between learnable and hand-crafted prompts across asynchronous audio and visual streams. Specifically, our SIMPLE combines Learnable Continuous Prompt Tuning (LCPT) with Semantic Topology Consistency Regularization (STCR). The LCPT automates prompt engineering through data-driven optimization, while the STCR module aligns the category-wise similarity distributions of learnable and hand-crafted prompts under random category subset sampling to alleviate semantic topology drift. Comprehensive experiments and analyses on the OV-AVEBench demonstrate that our SIMPLE framework achieves state-of-the-art performance on the seen and unseen splits, while using roughly half the trainable parameters and roughly one-third the training time.

cs.MM↗

arXiv2026

MedMemoryBench: Benchmarking Agent Memory in Personalized Healthcare

The large-scale deployment of personalized healthcare agents demands memory mechanisms that are exceptionally precise, safe, and capable of long-term clinical tracking. However, existing benchmarks primarily focus on daily open-domain conversations, failing to capture the high-stakes complexity of real-world medical applications. Motivated by the stringent production requirements of an industry-leading health management agent serving tens of millions of active users, we introduce MedMemoryBench. We develop a human-agent collaborative pipeline to synthesize highly realistic, long-horizon medical trajectories based on clinically grounded, synthetic patient archetypes. This process yields a massive, expertly validated dataset comprising approximately 2,000 sessions and 16,000 interaction turns. Crucially, MedMemoryBench departs from traditional static evaluations by pioneering an "evaluate-while-constructing" streaming assessment protocol, which precisely mirrors dynamic memory accumulation in production environments. Furthermore, we formalize and systematically investigate the critical phenomenon of memory saturation, where sustained information influx actively degrades retrieval and reasoning robustness. Comprehensive benchmarking reveals severe bottlenecks in mainstream architectures, particularly concerning complex medical reasoning and noise resilience. By exposing these fundamental flaws, MedMemoryBench establishes a vital foundation for developing robust, production-ready medical agents.

cs.AI↗

arXiv2026

Outage-Aware Robust Sensing for UAV ISAC Systems

Integrated sensing and communications (ISAC) infrastructure can provide an external perception layer for aerial robots, but dynamic blockage makes the set of informative sensing links at the next update uncertain. A single predicted availability pattern can miss low-information outcomes, whereas protecting all patterns wastes power and may become infeasible when all links are blocked. We propose an outage-aware conformal robust power-allocation framework for multi-BS UAV tracking. A history-aware joint predictor maps radar and tracking histories to next-slot availability probabilities, and adaptive prediction sets retain plausible correlated patterns. Tracking constraints are enforced only for retained non-outage patterns, while the all-blocked pattern is treated separately as a physical sensing outage. With per-slot beam directions fixed, the resulting scalar-power allocation is a semidefinite program. A reliability bound relates serviceable-slot tracking failure to deployed-policy miscoverage and optimization infeasibility. Across paired closed-loop simulations over five random seeds, the method attains a 0.01% serviceable-slot constraint-violation rate and uses 47.0% less power than protection over all non-outage patterns.

eess.SP↗

arXiv2026

Online Linear Programming with Batching

We study Online Linear Programming (OLP) with batching. The planning horizon is cut into $K$ batches, and decisions on orders can be delayed to the end of their associated batch. The ability to delay decisions improves operational performance, as measured by regret. We study two questions: (1) What is a lower bound on the regret as a function of $K$ and the length of the planning horizon? (2) Which algorithms can achieve this regret lower bound? This paper analyzes these questions when the distribution of the reward has a continuous support. We provide an $Ω(\log K)$ regret lower bound in the single-resource case, and we provide pricing algorithms having an $O(\log K)$ regret in the setting with Poisson arrivals and multiple types of resources. All the algorithms update the prices at most $K$ times and only delay orders of the first and the last batches. All the regret bounds are independent of the length of the planning horizon. Finally, we study a more realistic large support setting where the number of distinct order types is finite but scales with the total number of orders. We prove that our $Θ(\log K)$ bounds still hold for the batching operation of a multisecretary problem with discretized uniform rewards in this large support setting, provided the support grows fast enough. This suggests that the continuous support setting that is the paper's focus serves as a useful theoretical surrogate for the more realistic large finite support setting.

cs.LG↗

arXiv2026

Test-time Multi-agent Coordination by Decomposed Value Gradient Flow

Offline multi-agent reinforcement learning (MARL) faces a persistent trade-off. Expressive generative policies can represent multi-modal coordination in the data, but cannot distinguish high-value regions, while value-optimized policies exploit the learned Q-function but collapse the multi-modal into a single dominant mode. A single agent's mode collapse can break joint coordination, and simultaneous drift across agents can push the joint policy into unseen regions of the action space. We propose scalable coordination via optimal unified transport (SCOUT), the first offline MARL framework to combine a generative foundation model with a learned value function through test-time action refinement. SCOUT trains two decoupled components: a flow-matching behavioral prior and a decomposed value function. At test-time, it transports behavioral samples toward high-value regions via Stein variational gradient descent. The number of transport steps controls adaptive test-time scaling, replacing a fixed regularization coefficient. Under the individual-global-max (IGM) principle, we prove a single-term KL bound on the joint soft-value gap that vanishes as transport converges, with an irreducible additive residual proportional to the IGM violation. Empirically, SCOUT achieves the best average performance across discrete and continuous offline MARL benchmarks and yields performance improvements in all offline-to-online configurations.

cs.LG↗

arXiv2026

NowcastDiT: Diffusion Transformers are Effective Precipitation Nowcasters

Precipitation nowcasting demands accurate short-term forecasts under strong spatiotemporal variability. Diffusion models are well suited to modeling complex precipitation distributions, yet existing approaches often introduce increasingly specialized designs, leaving the capability of a standard diffusion architecture underexplored. We show that a standard Diffusion Transformer already provides a simple and scalable foundation for precipitation nowcasting, with domain-specific requirements accommodated naturally within its design space. Based on this principle, we develop NowcastDiT and instantiate this flexibility through two complementary adaptations: a dynamics-aware noise prior for temporally coherent forecasts, and end-to-end reinforcement learning with timestep-aware rewards for meteorological skill. Experiments on SEVIR and MRMS benchmarks show that NowcastDiT achieves state-of-the-art performance in both perceptual quality and meteorological skill. These results suggest that standard DiT can serve as an effective foundation for precipitation nowcasting.

cs.LG↗

arXiv2026

Error Bounds for Boundary-Stopped Positive Cubature Schemes for Hamilton--Jacobi--Bellman Equations

We derive error bounds for a boundary-stopped positive cubature scheme for degenerate parabolic Hamilton--Jacobi--Bellman equations with Dirichlet data on bounded domains. The scheme balances antipodal branches at their first boundary contacts and uses a control-dependent effective time. Its stopped-barrier remainder is $O(τΔt^{γ/2})$, and the total mass of interpolated branches is at most $2τ/Δt$, where $τ$ is the effective time. After division by the effective time, these estimates give a consistency bound that is uniform as $τ\to0$. Under a common strict boundary barrier and the stated mesh conditions, interior consistency, coefficient shaking, switching, and localized comparison yield \[ \|(u-u_h)^+\|_\infty\le C(Δt^{1/4}+hΔt^{-1/2}), \qquad \|(u_h-u)^+\|_\infty\le C(Δt^{1/10}+h^{1/2}Δt^{-1/4}). \] Thus $Δt\asymp h^{10/7}$ gives an $O(h^{1/7})$ maximum-norm bound. The analysis applies to any fixed centrally symmetric positive degree-two cubature and permits rank-deficient diffusion. Numerical examples examine parallel heat loss to a cold wall and control-dependent, rank-deficient diffusion.

math.NA↗

arXiv2026

GT-Space: Enhancing Heterogeneous Collaborative Perception with Ground Truth Feature Space

In autonomous driving, multi-agent collaborative perception enhances sensing capabilities by enabling agents to share perceptual data. A key challenge lies in handling {\em heterogeneous} features from agents equipped with different sensing modalities or model architectures, which complicates data fusion. Existing approaches often require retraining encoders or designing interpreter modules for pairwise feature alignment, but these solutions are not scalable in practice. To address this, we propose {\em GT-Space}, a flexible and scalable collaborative perception framework for heterogeneous agents. GT-Space constructs a common feature space from ground-truth labels, providing a unified reference for feature alignment. With this shared space, agents only need a single adapter module to project their features, eliminating the need for pairwise interactions with other agents. Furthermore, we design a fusion network trained with contrastive losses across diverse modality combinations. Extensive experiments on simulation datasets (OPV2V and V2XSet) and a real-world dataset (RCooper) demonstrate that GT-Space consistently outperforms baselines in detection accuracy while delivering robust performance. Our code will be released at https://github.com/KingScar/GT-Space.

cs.LG↗

arXiv2026

Cherry-pick Override: LLM Judges Under-use the Non-Directional Verdicts Their Contract Authorizes

Evidence-grounded fact verification increasingly uses LLM judges to turn evidence into terminal verdicts. Many task contracts deliberately include non-directional verdicts - Conflicting for materially mixed evidence, Not Enough Evidence for absent evidence - so that a system can decline to assert a direction. Returning Supports or Refutes on a claim whose authorized verdict is non-directional is directional overcommitment: a label error whose output is a direction the contract did not authorize. Its conflict instance - both strands present, one of them returned - is Cherry-pick Override (CPO); unlike the insufficiency instance it cannot be repaired by retrieving more, because nothing was missing. On AVeriTeC's gold-Conflicting subset (N_C=150) a four-option typed panel commits directionally on 18.7% of claims, and four frozen contemporary judges (two proprietary, two open-weight) on 21.3-48.0%, with 24.0-38.0% on VitaminC-Mixed, a constructed stress test on which the ranking reverses. These are conservative lower bounds: the shared prompt specifies Conflicting in detail, forbids collapsing to a direction, and demonstrates the label three times. The failure is not specific to conflict: overcommitment on gold-Insufficient cases runs at a comparable rate for every judge and exceeds CPO for three of four on VitaminC-Mixed. Asked in a fresh context, two judges identify the conflict on about two thirds of their own CPO claims, against under 0.12 where they committed correctly: the recognition is available and does not reach the terminal action. Typed vocabulary, panel voting, confidence thresholds, a structural validator and self-decomposition each leave a substantial residual, and the last buys part of its reduction by relabelling one-sided evidence as conflicting. Every rate is dataset-defined; we report no human-confirmed prevalence.

cs.SE↗

arXiv2026

NeoRed: A Knowledge-Logic-Alignment Multimodal Large Language Model for Neonatal Respiratory Disease Diagnosis

Neonatal respiratory diseases are a major cause of neonatal morbidity and mortality, posing substantial challenges in clinical practice. Despite recent advances, existing Multimodal Large Language Models (MLLMs) face two key limitations in neonatal diagnosis: (1) domain gap arising from predominantly adult training data; (2) insufficient integration of multidimensional clinical context for accurate diagnosis. To address these challenges, we collect two real-world clinical datasets (NeoCXR and NeoCXR-EV) and propose NeoRed, to the best of our knowledge, the first MLLM tailored for neonatal respiratory disease, filling the gap in neonatal diagnostic reports generation. To enhance joint diagnosis from heterogeneous clinical context and chest X-rays, we design a novel Knowledge-Logic-Alignment (KLA) framework which constrains model behavior from three perspectives: 1) Knowledge Prior Injection (KPI) incorporates neonatologist-inspired diagnostic priors into multimodal representations, guiding disease-specific attention across modalities; 2) Diagnostic Logic Constraint (DLC) aligns the semantics of generated reports with multimodal diagnostic logic; and 3) Visual Semantic Alignment (VSA) establishes semantic correspondence between visual features and imaging conclusions. Extensive experiments demonstrate that NeoRed enables accurate neonatal diagnostic reports generation, achieving ROUGE-L of 53.29% and Clinical Efficacy F1 score of 65.19% on NeoCXR, outperforming existing MLLMs. NeoRed also preserves competitive report generation performance on adult benchmarks (MIMIC-CXR and IU-Xray). Datasets will be available upon application.

cs.AI↗

arXiv2026

Deflickering Vision-Based Occupancy Networks through Lightweight Spatio-Temporal Correlation

Vision-based occupancy networks (VONs) provide an end-to-end solution for reconstructing 3D environments in autonomous driving. However, existing methods often suffer from temporal inconsistencies, manifesting as flickering effects that degrade temporal coherence and adversely affect downstream decision-making. While recent approaches incorporate historical information to alleviate this issue, they often incur high computational costs and may introduce misaligned or redundant features that interfere with object detection. We propose OccLinker, a novel plugin framework that can be easily integrated into existing VONs to improve performance. Our method efficiently consolidates historical static and motion cues, learns sparse latent correlations with current features through a dual cross-attention mechanism, and generates correction occupancy components to refine the base network predictions. In addition, we introduce a new temporal consistency metric to quantitatively measure flickering effects. Extensive experiments on two benchmark datasets demonstrate that our method achieves superior performance with minimal computational overhead while effectively reducing flickering artifacts.

cs.CV↗