arXiv subjects

Chi Zhang

Publications and source records attributed to Chi Zhang.

At least 325 records · Page 18Linked to original sources

arXiv2025

NFIG: Multi-Scale Autoregressive Image Generation via Frequency Ordering

Autoregressive models have achieved significant success in image generation. However, unlike the inherent hierarchical structure of image information in the spectral domain, standard autoregressive methods typically generate pixels sequentially in a fixed spatial order. To better leverage this spectral hierarchy, we introduce NextFrequency Image Generation (NFIG). NFIG is a novel framework that decomposes the image generation process into multiple frequency-guided stages. NFIG aligns the generation process with the natural image structure. It does this by first generating low-frequency components, which efficiently capture global structure with significantly fewer tokens, and then progressively adding higher-frequency details. This frequency-aware paradigm offers substantial advantages: it not only improves the quality of generated images but crucially reduces inference cost by efficiently establishing global structure early on. Extensive experiments on the ImageNet-256 benchmark validate NFIG's effectiveness, demonstrating superior performance (FID: 2.81) and a notable 1.25x speedup compared to the strong baseline VAR-d20. The source code is available at https://github.com/Pride-Huang/NFIG.

cs.CV↗

arXiv2025

Dolphin v1.0 Technical Report

Ultrasound is crucial in modern medicine but faces challenges like operator dependence, image noise, and real-time scanning, hindering AI integration. While large multimodal models excel in other medical imaging areas, they struggle with ultrasound's complexities. To address this, we introduce Dolphin v1.0 (V1) and its reasoning-augmented version, Dolphin R1-the first large-scale multimodal ultrasound foundation models unifying diverse clinical tasks in a single vision-language framework.To tackle ultrasound variability and noise, we curated a 2-million-scale multimodal dataset, combining textbook knowledge, public data, synthetic samples, and general corpora. This ensures robust perception, generalization, and clinical adaptability.The Dolphin series employs a three-stage training strategy: domain-specialized pretraining, instruction-driven alignment, and reinforcement-based refinement. Dolphin v1.0 delivers reliable performance in classification, detection, regression, and report generation. Dolphin R1 enhances diagnostic inference, reasoning transparency, and interpretability through reinforcement learning with ultrasound-specific rewards.Evaluated on U2-Bench across eight ultrasound tasks, Dolphin R1 achieves a U2-score of 0.5835-over twice the second-best model (0.2968) setting a new state of the art. Dolphin v1.0 also performs competitively, validating the unified framework. Comparisons show reasoning-enhanced training significantly improves diagnostic accuracy, consistency, and interpretability, highlighting its importance for high-stakes medical AI.

cs.CV↗

arXiv2025

A Survey of Scientific Large Language Models: From Data Foundations to Agent Frontiers

Scientific Large Language Models (Sci-LLMs) are transforming how knowledge is represented, integrated, and applied in scientific research, yet their progress is shaped by the complex nature of scientific data. This survey presents a comprehensive, data-centric synthesis that reframes the development of Sci-LLMs as a co-evolution between models and their underlying data substrate. We formulate a unified taxonomy of scientific data and a hierarchical model of scientific knowledge, emphasizing the multimodal, cross-scale, and domain-specific challenges that differentiate scientific corpora from general natural language processing datasets. We systematically review recent Sci-LLMs, from general-purpose foundations to specialized models across diverse scientific disciplines, alongside an extensive analysis of over 270 pre-/post-training datasets, showing why Sci-LLMs pose distinct demands -- heterogeneous, multi-scale, uncertainty-laden corpora that require representations preserving domain invariance and enabling cross-modal reasoning. On evaluation, we examine over 190 benchmark datasets and trace a shift from static exams toward process- and discovery-oriented assessments with advanced evaluation protocols. These data-centric analyses highlight persistent issues in scientific data development and discuss emerging solutions involving semi-automated annotation pipelines and expert validation. Finally, we outline a paradigm shift toward closed-loop systems where autonomous agents based on Sci-LLMs actively experiment, validate, and contribute to a living, evolving knowledge base. Collectively, this work provides a roadmap for building trustworthy, continually evolving artificial intelligence (AI) systems that function as a true partner in accelerating scientific discovery.

cs.CL↗

arXiv2025

SPOT: Sensing-augmented Trajectory Planning via Obstacle Threat Modeling

UAVs equipped with a single depth camera encounter significant challenges in dynamic obstacle avoidance due to limited field of view and inevitable blind spots. While active vision strategies that steer onboard cameras have been proposed to expand sensing coverage, most existing methods separate motion planning from sensing considerations, resulting in less effective and delayed obstacle response. To address this limitation, we introduce SPOT (Sensing-augmented Planning via Obstacle Threat modeling), a unified planning framework for observation-aware trajectory planning that explicitly incorporates sensing objectives into motion optimization. At the core of our method is a Gaussian Process-based obstacle belief map, which establishes a unified probabilistic representation of both recognized (previously observed) and potential obstacles. This belief is further processed through a collision-aware inference mechanism that transforms spatial uncertainty and trajectory proximity into a time-varying observation urgency map. By integrating urgency values within the current field of view, we define differentiable objectives that enable real-time, observation-aware trajectory planning with computation times under 10 ms. Simulation and real-world experiments in dynamic, cluttered, and occluded environments show that our method detects potential dynamic obstacles 2.8 seconds earlier than baseline approaches, increasing dynamic obstacle visibility by over 500\%, and enabling safe navigation through cluttered, occluded environments.

cs.RO↗

arXiv2025

FinHEAR: Human Expertise and Adaptive Risk-Aware Temporal Reasoning for Financial Decision-Making

Financial decision-making presents unique challenges for language models, demanding temporal reasoning, adaptive risk assessment, and responsiveness to dynamic events. While large language models (LLMs) show strong general reasoning capabilities, they often fail to capture behavioral patterns central to human financial decisions-such as expert reliance under information asymmetry, loss-averse sensitivity, and feedback-driven temporal adjustment. We propose FinHEAR, a multi-agent framework for Human Expertise and Adaptive Risk-aware reasoning. FinHEAR orchestrates specialized LLM-based agents to analyze historical trends, interpret current events, and retrieve expert-informed precedents within an event-centric pipeline. Grounded in behavioral economics, it incorporates expert-guided retrieval, confidence-adjusted position sizing, and outcome-based refinement to enhance interpretability and robustness. Empirical results on curated financial datasets show that FinHEAR consistently outperforms strong baselines across trend prediction and trading tasks, achieving higher accuracy and better risk-adjusted returns.

cs.LG↗

arXiv2025

MotionAgent: Fine-grained Controllable Video Generation via Motion Field Agent

We propose MotionAgent, enabling fine-grained motion control for text-guided image-to-video generation. The key technique is the motion field agent that converts motion information in text prompts into explicit motion fields, providing flexible and precise motion guidance. Specifically, the agent extracts the object movement and camera motion described in the text and converts them into object trajectories and camera extrinsics, respectively. An analytical optical flow composition module integrates these motion representations in 3D space and projects them into a unified optical flow. An optical flow adapter takes the flow to control the base image-to-video diffusion model for generating fine-grained controlled videos. The significant improvement in the Video-Text Camera Motion metrics on VBench indicates that our method achieves precise control over camera motion. We construct a subset of VBench to evaluate the alignment of motion information in the text and the generated video, outperforming other advanced models on motion generation accuracy.

cs.CV↗

arXiv2025

BrainOmni: A Brain Foundation Model for Unified EEG and MEG Signals

Electroencephalography (EEG) and magnetoencephalography (MEG) measure neural activity non-invasively by capturing electromagnetic fields generated by dendritic currents. Although rooted in the same biophysics, EEG and MEG exhibit distinct signal patterns, further complicated by variations in sensor configurations across modalities and recording devices. Existing approaches typically rely on separate, modality- and dataset-specific models, which limits the performance and cross-domain scalability. This paper proposes BrainOmni, the first brain foundation model that generalises across heterogeneous EEG and MEG recordings. To unify diverse data sources, we introduce BrainTokenizer,the first tokenizer that quantises spatiotemporal brain activity into discrete representations. Central to BrainTokenizer is a novel Sensor Encoder that encodes sensor properties such as spatial layout, orientation, and type, enabling compatibility across devices and modalities. Building upon the discrete representations, BrainOmni learns unified semantic embeddings of brain signals by self-supervised pretraining. To the best of our knowledge, it is the first foundation model to support both EEG and MEG signals, as well as the first to incorporate large-scale MEG pretraining. A total of 1,997 hours of EEG and 656 hours of MEG data are curated and standardised from publicly available sources for pretraining. Experiments show that BrainOmni outperforms both existing foundation models and state-of-the-art task-specific models on a range of downstream tasks. It also demonstrates strong generalisation to unseen EEG and MEG devices. Further analysis reveals that joint EEG-MEG (EMEG) training yields consistent improvements across both modalities. Code and checkpoints are publicly available at https://github.com/OpenTSLab/BrainOmni.

eess.SP↗

arXiv2025

Towards Multimodal Query-Based Spatial Audio Source Extraction

Query-based audio source extraction seeks to recover a target source from a mixture conditioned on a query. Existing approaches are largely confined to single-channel audio, leaving the spatial information in multi-channel recordings underexploited. We introduce a query-based spatial audio source extraction framework for recovering dry target signals from first-order ambisonics (FOA) mixtures. Our method accepts either an audio prompt or a text prompt as condition input, enabling flexible end-to-end extraction. The core of our proposed model lies in a tri-axial Transformer that jointly models temporal, frequency, and spatial channel dependencies. The model uses contrastive language-audio pretraining (CLAP) embeddings to enable unified audio-text conditioning via feature-wise linear modulation (FiLM). To eliminate costly annotations and improve generalization, we propose a label-free data pipeline that dynamically generates spatial mixtures and corresponding targets for training. The result of our experiment with high separation quality demonstrates the efficacy of multimodal conditioning and tri-axial modeling. This work establishes a new paradigm for high-fidelity spatial audio separation in immersive applications.

eess.AS↗

arXiv2025

Macro-from-Micro Planning for High-Quality and Parallelized Autoregressive Long Video Generation

Current autoregressive diffusion models excel at video generation but are generally limited to short temporal durations. Our theoretical analysis indicates that the autoregressive modeling typically suffers from temporal drift caused by error accumulation and hinders parallelization in long video synthesis. To address these limitations, we propose a novel planning-then-populating framework centered on Macro-from-Micro Planning (MMPL) for long video generation. MMPL sketches a global storyline for the entire video through two hierarchical stages: Micro Planning and Macro Planning. Specifically, Micro Planning predicts a sparse set of future keyframes within each short video segment, offering motion and appearance priors to guide high-quality video segment generation. Macro Planning extends the in-segment keyframes planning across the entire video through an autoregressive chain of micro plans, ensuring long-term consistency across video segments. Subsequently, MMPL-based Content Populating generates all intermediate frames in parallel across segments, enabling efficient parallelization of autoregressive generation. The parallelization is further optimized by Adaptive Workload Scheduling for balanced GPU execution and accelerated autoregressive video generation. Extensive experiments confirm that our method outperforms existing long video generation models in quality and stability. Generated videos and comparison results are in our project page.

cs.CV↗

arXiv2025

Laminar: A Scalable Asynchronous RL Post-Training Framework

Reinforcement learning (RL) post-training for Large Language Models (LLMs) is now scaling to large clusters and running for extended durations to enhance model reasoning performance. However, the scalability of existing RL frameworks is limited, as extreme long-tail skewness in RL trajectory generation causes severe GPU underutilization. Current asynchronous RL systems attempt to mitigate this, but they rely on global weight synchronization between the actor and all rollouts, which creates a rigid model update schedule. This global synchronization is ill-suited for the highly skewed and evolving distribution of trajectory generation latency in RL training, crippling training efficiency. Our key insight is that efficient scaling requires breaking this lockstep through trajectory-level asynchrony, which generates and consumes each trajectory independently. We propose Laminar, a scalable and robust RL post-training system built on a fully decoupled architecture. First, we replace global updates with a tier of relay workers acting as a distributed parameter service. This enables asynchronous and fine-grained weight synchronization, allowing rollouts to pull the latest weight anytime without stalling the actor's training loop. Second, a dynamic repack mechanism consolidates long-tail trajectories onto a few dedicated rollouts, maximizing generation throughput. The fully decoupled design also isolates failures, ensuring robustness for long-running jobs. Our evaluation on a 1024-GPU cluster shows that Laminar achieves up to 5.48$\times$ training throughput speedup over state-of-the-art systems, while reducing model convergence time.

cs.LG↗

arXiv2025

Aixel: A Unified, Adaptive and Extensible System for AI-powered Data Analysis

A growing trend in modern data analysis is the integration of data management with learning, guided by accuracy, latency, and cost requirements. In practice, applications draw data of different formats from many sources. In the meanwhile, the objectives and budgets change over time. Existing systems handle these applications across databases, analysis libraries, and tuning services. Such fragmentation leads to complex user interaction, limited adaptability, suboptimal performance, and poor extensibility across components. To address these challenges, we present Aixel, a unified, adaptive, and extensible system for AI-powered data analysis. The system organizes work across four layers: application, task, model, and data. The task layer provides a declarative interface to capture user intent, which is parsed into an executable operator plan. An optimizer compiles and schedules this plan to meet specified goals in accuracy, latency, and cost. The task layer coordinates the execution of data and model operators, with built-in support for reuse and caching to improve efficiency. The model layer offers versioned storage for index, metadata, tensors, and model artifacts. It supports adaptive construction, task-aligned drift detection, and safe updates that reuse shared components. The data layer provides unified data management capabilities, including indexing, constraint-aware discovery, task-aligned selection, and comprehensive feature management. With the above designed layers, Aixel delivers a user friendly, adaptive, efficient, and extensible system.

cs.DB↗

arXiv2025

Learning Shared and Source-specific Subspaces across Multiple Data Sources for Functional Data

In the era of big data, integrating multi-source functional data to extract a subspace that captures the shared subspace across sources has attracted considerable attention. In practice, data collection procedures often follow source-specific protocols. Directly averaging sample covariance operators across sources implicitly assumes homogeneity, which may bias the recovery of both shared and source-specific variation patterns. To address this issue, we propose a projection-based data integration method that explicitly separates the shared and source-specific subspaces. The method first estimates source-specific projection operators via smoothing to accommodate the nonparametric nature of functional data. The shared subspace is then isolated by examining the eigenvalues of the averaged projection operator across all sources. If a source-specific subspace is of interest, we re-project the associated source-specific covariance estimator onto the subspace orthogonal to the estimated shared subspace, and estimate the source-specific subspace from the resulting projection. We further establish the asymptotic properties of both the shared and source-specific subspace estimators. Extensive simulation studies demonstrate the effectiveness of the proposed method across a wide range of settings. Finally, we illustrate its practical utility with an example of air pollutant data.

stat.ME↗

arXiv2025

Entropy Engineering-Regulated Electron-Phonon Coupling for Highly Efficient Photoluminescence in Se-doped WS2

The limited quantum yield of strained monolayer transition metal dichalcogenides grown by vapor-phase methods and during transfer-based stacking poses a fundamental challenge for their optoelectronic applications. Here, we introduce the concept of "entropy engineering" as a transformative strategy to selectively enhance light-matter interactions through controlled electron-phonon coupling. We unveil how tailored entropy introduced via precise selenium doping or interfacial van der Waals proximity can significantly amplify radiative recombination from momentum-dark excitons in WS2 monolayers. Notably, we discover that slight selenium doping drastically enhances the photoluminescence (PL) of WS2 under strain. While both undoped and heavily doped WS2 suffer from strong PL quenching owing to the direct-to-indirect bandgap transition, lightly Se-doped samples exhibit an order-of-magnitude increase in emission intensity. This counterintuitive boost is traced to doping-induced structural disorder, which intensifies electron-phonon interactions and unlocks efficient phonon-assisted emission from otherwise non-radiative indirect excitons. Moreover, we demonstrate that van der Waals coupling to adjacent Se-doped layers can impart interfacial entropy and further augment PL via proximity effects. Our work highlights entropy engineering via controlled doping as a powerful strategy for activating high-efficiency light emission in atomically thin semiconductors.

cond-mat.mtrl-sci↗

arXiv2025

Local-Antisymmetric Flat Band and Coexisting Correlated stripe charge orders in WSe2-Modulated Twisted Bilayer Graphene

Insulating, atomically flat transition metal dichalcogenides (TMDs) like WSe2 are ideal substrates for probing intrinsic graphene properties. Conventionally, their influence on graphene's band structure is assumed negligible, particularly when small moire patterns form. Combining scanning tunneling microscopy/spectroscopy and theoretical analysis, we reveal that the atomic registry in graphene/WSe2 heterostructures profoundly modulates the electronic structure of magic-angle twisted bilayer graphene (MATBG). At special graphene/WSe2 twist angles, an incommensurate moire superlattice hosts three distinct atomic stacking configurations (A, B, X types). These induce position-dependent potentials that asymmetrically shift MATBG's flat bands, transforming them from hole-side to electron-side asymmetric within a single AA-stacked region. This symmetry breaking enables the unprecedented coexistence of orthogonal stripe charge orders in the correlated regime-a phenomenon previously considered mutually exclusive due to Coulomb repulsion. This band modulation arises from the synergistic effects of the graphene/WSe2 interfacial atomic registry and heterostrain within the MATBG, exhibiting multi-field tunability. Our work establishes interfacial atomic registry as a critical, previously overlooked tuning parameter for flat-band physics, opening avenues to engineer correlated quantum states in van der Waals heterostructures.

cond-mat.mes-hall↗

arXiv2025

Latent Class Logit Kernel Framework for Surrogate Safety: Identifying Behavioural Thresholds through Conflict Indicator Profiles

Crash data objectively characterize road safety but are rare and often unsuitable for proactive safety management. Traffic conflict indicators such as time-to-collision (TTC) provide continuous measures of collision proximity but require thresholds to distinguish routine from safety-critical interactions. Extreme Value Theory (EVT) offers statistically defined thresholds, yet these do not necessarily represent how drivers perceive and respond to conflict. This study introduces a behavioural modelling framework that identifies candidate behavioural thresholds (CBTs) by explicitly modelling how drivers adjust their movements under conflict conditions. The framework is based on a Latent Class Logit Kernel (LC-LK) model that captures inter-class heterogeneity (routine vs. defensive driving) and intra-class correlation between overlapping spatial alternatives. This yields probability curves showing how the likelihood of defensive manoeuvres varies with conflict indicators, from which CBTs such as inflection points and crossovers can be extracted. The framework tests four hypotheses: (1) drivers exhibit varying degrees of membership in both low- and high-risk classes; (2) membership shifts systematically with conflict values, revealing behavioural thresholds; (3) this relationship follows a logistic shape, with stable behaviour at safe levels and rapid transitions near critical points; and (4) even in free flow, drivers maintain a baseline caution level. Application to naturalistic roundabout trajectories revealed stable TTC thresholds (0.8-1.1 s) but unstable MTTC2 estimates (e.g., 34 s), suggesting cognitive limits in processing complex indicators. Overall, the framework complements EVT by offering a structured, behaviourally grounded method for identifying and validating thresholds in surrogate safety analysis.

physics.soc-ph↗

arXiv2025

Improving Hierarchical Representations of Vectorized HD Maps with Perspective Clues

The construction of vectorized High-Definition (HD) maps from onboard surround-view cameras has become a significant focus in autonomous driving. However, current map vector estimation pipelines face two key limitations: input-agnostic queries struggle to capture complex map structures, and the view transformation leads to information loss. These issues often result in inaccurate shape restoration or missing instances in map predictions. To address this concern, we propose a novel approach, namely \textbf{PerCMap}, which explicitly exploits clues from perspective-view features at both instance and point level. Specifically, at instance level, we propose Cross-view Instance Activation (CIA) to activate instance queries across surround-view images, thereby helping the model recover the instance attributes of map vectors. At point level, we design Dual-view Point Embedding (DPE), which fuses features from both views to generate input-aware positional embeddings and improve the accuracy of point coordinate estimation. Extensive experiments on \textit{nuScenes} and \textit{Argoverse 2} demonstrate that PerCMap achieves strong and consistent performance across benchmarks, reaching 67.1 and 70.5 mAP, respectively.

cs.CV↗

arXiv2025

Towards Reliable LLM-based Robot Planning via Combined Uncertainty Estimation

Large language models (LLMs) demonstrate advanced reasoning abilities, enabling robots to understand natural language instructions and generate high-level plans with appropriate grounding. However, LLM hallucinations present a significant challenge, often leading to overconfident yet potentially misaligned or unsafe plans. While researchers have explored uncertainty estimation to improve the reliability of LLM-based planning, existing studies have not sufficiently differentiated between epistemic and intrinsic uncertainty, limiting the effectiveness of uncertainty estimation. In this paper, we present Combined Uncertainty estimation for Reliable Embodied planning (CURE), which decomposes the uncertainty into epistemic and intrinsic uncertainty, each estimated separately. Furthermore, epistemic uncertainty is subdivided into task clarity and task familiarity for more accurate evaluation. The overall uncertainty assessments are obtained using random network distillation and multi-layer perceptron regression heads driven by LLM features. We validated our approach in two distinct experimental settings: kitchen manipulation and tabletop rearrangement experiments. The results show that, compared to existing methods, our approach yields uncertainty estimates that are more closely aligned with the actual execution outcomes.

cs.RO↗

arXiv2025

Order Determination for Functional Data

Dimension reduction is often necessary in functional data analysis, with functional principal component analysis being one of the most widely used techniques. A key challenge in applying these methods is determining the number of eigen-pairs to retain, a problem known as order determination. When a covariance function admits a finite representation, the challenge becomes estimating the rank of the associated covariance operator. While this problem is straightforward when the full trajectories of functional data are available, in practice, functional data are typically collected discretely and are subject to measurement error contamination. This contamination introduces a ridge to the empirical covariance function, which obscures the true rank of the covariance operator. We propose a novel procedure to identify the true rank of the covariance operator by leveraging the information of eigenvalues and eigenfunctions. By incorporating the nonparametric nature of functional data through smoothing techniques, the method is applicable to functional data collected at random, subject-specific points. Extensive simulation studies demonstrate the excellent performance of our approach across a wide range of settings, outperforming commonly used information-criterion-based methods and maintaining effectiveness even in high-noise scenarios. We further illustrate our method with two real-world data examples.

stat.ME↗