arXiv subjects

Haoran Li

Publications and source records attributed to Haoran Li.

At least 145 records · Page 8Linked to original sources

arXiv2026

Learning to Fuse and Reconstruct Multi-View Graphs for Diabetic Retinopathy Grading

Diabetic retinopathy (DR) is one of the leading causes of vision loss worldwide, making early and accurate DR grading critical for timely intervention. Recent clinical practices leverage multi-view fundus images for DR detection with a wide coverage of the field of view (FOV), motivating deep learning methods to explore the potential of multi-view learning for DR grading. However, existing methods often overlook the inter-view correlations when fusing multi-view fundus images, failing to fully exploit the inherent consistency across views originating from the same patient. In this work, we present MVGFDR, an end-to-end Multi-View Graph Fusion framework for DR grading. Different from existing methods that directly fuse visual features from multiple views, MVGFDR is equipped with a novel Multi-View Graph Fusion (MVGF) module to explicitly disentangle the shared and view-specific visual features. Specifically, MVGF comprises three key components: (1) Multi-view Graph Initialization, which constructs visual graphs via residual-guided connections and employs Discrete Cosine Transform (DCT) coefficients as frequency-domain anchors; (2) Multi-view Graph Fusion, which integrates selective nodes across multi-view graphs based on frequency-domain relevance to capture complementary view-specific information; and (3) Masked Cross-view Reconstruction, which leverages masked reconstruction of shared information across views to facilitate view-invariant representation learning. Extensive experimental results on MFIDDR, by far the largest multi-view fundus image dataset, demonstrate the superiority of our proposed approach over existing state-of-the-art approaches in diabetic retinopathy grading.

cs.CV↗

arXiv2026

Real-time Calibration-free Imaging Through Dynamic and Distinct Multimode Fibers via Spatial Harmonic Invariant Nonlinear Encoding (SHINE)

Multimode fibers (MMFs) provide a compact, high-throughput platform for minimally invasive imaging and information transmission. However, their utility is fundamentally constrained by mode mixing, which renders image transmission spatially disrupted and sensitive to external perturbations. Current imaging methods typically rely on transmission matrix measurement or deep learning models that are fragile to fiber movement, necessitating frequent, time-consuming calibrations and re-calibrations that are easily disrupted and fail to generalize across different fiber configurations, let alone across entirely distinct fibers. Here, we propose a calibration and feedback-free MMF coherent imaging paradigm, that we termed Spatial Harmonic Invariant Nonlinear Encoding (SHINE). By leveraging the angle-dependent phase-matching conditions of second-harmonic generation, we encode spatial features into broadband spectral signatures that possess intrinsic insensitivity not only to modal scrambling but also to fiber bending, movement, as well as structural variations. This spectral representation enables a deep learning model to robustly reconstruct images in real time despite dynamic perturbations and even generalizes well to distinct MMFs without recalibration or feedback. We achieve experimentally an average Pearson correlation coefficient (PCC) of 0.82 for image reconstruction tasks on Fashion-MNIST and a classification accuracy of 92.3% on HERLEV biomedical dataset. Uniquely, our method exhibits remarkable cross-fiber generalization: a model trained on a single MMF successfully reconstructs images transmitted through entirely distinct, previously unseen MMFs with a PCC of 0.74. These results establish a robust, calibration-free framework for imaging through MMFs in real time, paving the way for practical, resilient optical diagnostics that operate without distal-end feedback.

physics.optics↗

arXiv2026

TKN: Transformer-based Keypoint Prediction Network For Real-time Video Prediction

Video prediction is a complex time-series forecasting task with great potential in many use cases. However, traditional methods prioritize accuracy and overlook slow prediction speeds due to complex model structures, redundant information, and excessive GPU memory consumption. These methods often predict frames sequentially, making acceleration difficult and limiting their applicability in real-time scenarios like danger prediction and warning.Therefore, we propose a transformer-based keypoint prediction neural network (TKN). TKN extracts dynamic content from video frames in an unsupervised manner, reducing redundant feature computation. And, TKN uses an acceleration matrix to reduce the computational cost of attention and employs a parallel computing structure for prediction acceleration. To the best of our knowledge, TKN is the first real-time video prediction solution that achieves a prediction rate of 1,176 fps, significantly reducing computation costs while maintaining other performance. Qualitative and quantitative experiments on multiple datasets have demonstrated the superiority of our method, suggesting that TKN has great application potential.

cs.CV↗

arXiv2026

Dual-Granularity Contrastive Reward via Generated Episodic Guidance for Efficient Embodied RL

Designing suitable rewards poses a significant challenge in reinforcement learning (RL), especially for embodied manipulation. Trajectory success rewards are suitable for human judges or model fitting, but the sparsity severely limits RL sample efficiency. While recent methods have effectively improved RL via dense rewards, they rely heavily on high-quality human-annotated data or abundant expert supervision. To tackle these issues, this paper proposes Dual-granularity contrastive reward via generated Episodic Guidance (DEG), a novel framework to seek sample-efficient dense rewards without requiring human annotations or extensive supervision. Leveraging the prior knowledge of large video generation models, DEG only needs a small number of expert videos for domain adaptation to generate dedicated task guidance for each RL episode. Then, the proposed dual-granularity reward that balances coarse-grained exploration and fine-grained matching, will guide the agent to efficiently approximate the generated guidance video sequentially in the contrastive self-supervised latent space, and finally complete the target task. Extensive experiments on 18 diverse tasks across both simulation and real-world settings show that DEG can not only serve as an efficient exploration stimulus to help the agent quickly discover sparse success rewards, but also guide effective RL and stable policy convergence independently.

cs.LG↗

arXiv2026

CoPE: Clipped RoPE as A Scalable Free Lunch for Long Context LLMs

Rotary Positional Embedding (RoPE) is a key component of context scaling in Large Language Models (LLMs). While various methods have been proposed to adapt RoPE to longer contexts, their guiding principles generally fall into two categories: (1) out-of-distribution (OOD) mitigation, which scales RoPE frequencies to accommodate unseen positions, and (2) Semantic Modeling, which posits that the attention scores computed with RoPE should always prioritize semantically similar tokens. In this work, we unify these seemingly distinct objectives through a minimalist intervention, namely CoPE: soft clipping lowfrequency components of RoPE. CoPE not only eliminates OOD outliers and refines semantic signals, but also prevents spectral leakage caused by hard clipping. Extensive experiments demonstrate that simply applying our soft clipping strategy to RoPE yields significant performance gains that scale up to 256k context length, validating our theoretical analysis and establishing CoPE as a new state-of-the-art for length generalization. Our code, data, and models are available at https://github.com/hrlics/CoPE.

cs.CL↗

arXiv2026

LASS-ODE: Scaling ODE Computations to Connect Foundation Models with Dynamical Physical Systems

Foundation models have transformed language, vision, and time series data analysis, yet progress on dynamic predictions for physical systems remains limited. Given the complexity of physical constraints, two challenges stand out. $(i)$ Physics-computation scalability: physics-informed learning can enforce physical regularization, but its computation (e.g., ODE integration) does not scale to extensive systems. $(ii)$ Knowledge-sharing efficiency: the attention mechanism is primarily computed within each system, which limits the extraction of shared ODE structures across systems. We show that enforcing ODE consistency does not require expensive nonlinear integration: a token-wise locally linear ODE representation preserves physical fidelity while scaling to foundation-model regimes. Thus, we propose novel token representations that respect locally linear ODE evolution. Such linearity substantially accelerates integration while accurately approximating the local data manifold. Second, we introduce a simple yet effective inter-system attention that augments attention with a common structure hub (CSH) that stores shared tokens and aggregates knowledge across systems. The resulting model, termed LASS-ODE (\underline{LA}rge-\underline{S}cale \underline{S}mall \underline{ODE}), is pretrained on our $40$GB ODE trajectory collections to enable strong in-domain performance, zero-shot generalization across diverse ODE systems, and additional improvements through fine-tuning.

cs.LG↗

arXiv2026

PWAVEP: Purifying Imperceptible Adversarial Perturbations in 3D Point Clouds via Spectral Graph Wavelets

Recent progress in adversarial attacks on 3D point clouds, particularly in achieving spatial imperceptibility and high attack performance, presents significant challenges for defenders. Current defensive approaches remain cumbersome, often requiring invasive model modifications, expensive training procedures or auxiliary data access. To address these threats, in this paper, we propose a plug-and-play and non-invasive defense mechanism in the spectral domain, grounded in a theoretical and empirical analysis of the relationship between imperceptible perturbations and high-frequency spectral components. Building upon these insights, we introduce a novel purification framework, termed PWAVEP, which begins by computing a spectral graph wavelet domain saliency score and local sparsity score for each point. Guided by these values, PWAVEP adopts a hierarchical strategy, it eliminates the most salient points, which are identified as hardly recoverable adversarial outliers. Simultaneously, it applies a spectral filtering process to a broader set of moderately salient points. This process leverages a graph wavelet transform to attenuate high-frequency coefficients associated with the targeted points, thereby effectively suppressing adversarial noise. Extensive evaluations demonstrate that the proposed PWAVEP achieves superior accuracy and robustness compared to existing approaches, advancing the state-of-the-art in 3D point cloud purification. Code and datasets are available at https://github.com/a772316182/pwavep

cs.CV↗

arXiv2026

Do Reasoning Models Enhance Embedding Models?

State-of-the-art embedding models are increasingly derived from decoder-only Large Language Model (LLM) backbones adapted via contrastive learning. Given the emergence of reasoning models trained via Reinforcement Learning with Verifiable Rewards (RLVR), a natural question arises: do enhanced reasoning translate to superior semantic representations when these models serve as embedding initializations? Contrary to expectation, our evaluation on MTEB and BRIGHT reveals a **null effect**: embedding models initialized from RLVR-tuned backbones yield no consistent performance advantage over their base counterparts when subjected to identical training recipes. To unpack this paradox, we introduce **H**ierarchical **R**epresentation **S**imilarity **A**nalysis (HRSA), a framework that decomposes similarity across representation, geometry, and function levels. HRSA reveals that while RLVR induces irreversible latent manifold's local geometry reorganization and reversible coordinate basis drift, it preserves the global manifold geometry and linear readout. Consequently, subsequent contrastive learning drives strong alignment between base- and reasoning-initialized models, a phenomenon we term **Manifold Realignment**. Empirically, our findings suggest that unlike Supervised Fine-Tuning (SFT), RLVR optimizes trajectories within an existing semantic landscape rather than fundamentally restructuring the landscape itself.

cs.AI↗

arXiv2026

Agent.xpu: Efficient Scheduling of Agentic LLM Workloads on Heterogeneous SoC

Personal LLM agents increasingly combine foreground reactive interactions with background proactive monitoring, forming long-lived, stateful LLM flows that interleave prefill and token-by-token decode. While modern heterogeneous SoCs integrate CPUs, iGPUs, and NPUs to support on-device intelligence, existing LLM engines assume static, single-shot inference and lack mechanisms for flow-level concurrency, prioritization, and efficient accelerator coordination. As a result, commodity SoCs remain poorly matched to the dynamic, mixed-criticality execution patterns of personal agents. This paper presents Agent$.$xpu, the first LLM engine that orchestrates concurrent reactive and proactive LLM flows on commodity SoCs. Extensive profiling uncovers unique SoC characteristics of operator-accelerator affinity, asymmetric DDR contention, and stage-divergent batching behaviors distinct from cloud-serving assumptions. Agent$.$xpu introduces three key techniques: a heterogeneous execution graph (HEG) capturing NPU/iGPU affinity and elastic operator binding; flow-aware NPU-iGPU coordination with stage elasticity, decoupling prefill and decode to reduce bandwidth contention and enforce priorities; and fine-grained preemption with slack-aware piggybacking to guarantee reactive responsiveness without starving proactive work. Across realistic personal-agent workloads, Agent$.$xpu delivers 1.2-4.9$\times$ proactive throughput and reduces reactive latency by at least 91%, compared with both industrial iGPU-only serving engine and NPU-iGPU static inference with optimal tensor-partitioning schemes. Agent$.$xpu also minimizes energy consumption and graphics interference via controlled iGPU usage.

cs.DC↗

arXiv2026

LORE: A Large Generative Model for Search Relevance

Achievement. We introduce LORE, a systematic framework for Large Generative Model-based relevance in e-commerce search. Deployed and iterated over three years, LORE achieves a cumulative +27\% improvement in online GoodRate metrics. This report shares the valuable experience gained throughout its development lifecycle, spanning data, features, training, evaluation, and deployment. Insight. While existing works apply Chain-of-Thought (CoT) to enhance relevance, they often hit a performance ceiling. We argue this stems from treating relevance as a monolithic task, lacking principled deconstruction. Our key insight is that relevance comprises distinct capabilities: knowledge and reasoning, multi-modal matching, and rule adherence. We contend that a qualitative-driven decomposition is essential for breaking through current performance bottlenecks. Contributions. LORE provides a complete blueprint for the LLM relevance lifecycle. Key contributions include: (1) A two-stage training paradigm combining progressive CoT synthesis via SFT with human preference alignment via RL. (2) A comprehensive benchmark, RAIR, designed to evaluate these core capabilities. (3) A query frequency-stratified deployment strategy that efficiently transfers offline LLM capabilities to the online system. LORE serves as both a practical solution and a methodological reference for other vertical domains.

cs.IR↗

arXiv2026

Evidence-Grounded Multi-Agent Planning Support for Urban Carbon Governance via RAG

Urban carbon governance requires planners to integrate heterogeneous evidence -- emission inventories, statistical yearbooks, policy texts, technical measures, and academic findings -- into actionable, cross-departmental plans. Large Language Models (LLMs) can assist planning workflows, yet their factual reliability and evidential traceability remain critical barriers in professional use. This paper presents an evidence-grounded multi-agent planning support system for urban carbon governance built upon standard text-based Retrieval-Augmented Generation (RAG) (without GraphRAG). We align the system with the typical planning workflow by decomposing tasks into four specialized agents: (i) evidence Q\&A for fact checking and compliance queries, (ii) emission status assessment for diagnostic analysis, (iii) planning recommendation for generating multi-sector governance pathways, and (iv) report integration for producing planning-style deliverables. We evaluate the system in two task families: factual retrieval and comprehensive planning generation. On factual retrieval tasks, introducing RAG increases the average score from below 6 to above 90, and dramatically improves key-field extraction (e.g., region and numeric values near 100\% detection). A real-city case study (Ningbo, China) demonstrates end-to-end report generation with strong relevance, coverage, and coherence in expert review, while also highlighting boundary inconsistencies across data sources as a practical limitation.

cs.CY↗

arXiv2025

The Primacy of Magnitude in Low-Rank Adaptation

Low-Rank Adaptation (LoRA) offers a parameter-efficient paradigm for tuning large models. While recent spectral initialization methods improve convergence and performance over the naive "Noise & Zeros" scheme, their extra computational and storage overhead undermines efficiency. In this paper, we establish update magnitude as the fundamental driver of LoRA performance and propose LoRAM, a magnitude-driven "Basis & Basis" initialization scheme that matches spectral methods without their inefficiencies. Our key contributions are threefold: (i) Magnitude of weight updates determines convergence. We prove low-rank structures intrinsically bound update magnitudes, unifying hyperparameter tuning in learning rate, scaling factor, and initialization as mechanisms to optimize magnitude regulation. (ii) Spectral initialization succeeds via magnitude amplification. We demystify that the presumed knowledge-driven benefit of the spectral component essentially arises from the boost in the weight update magnitude. (iii) A novel and compact initialization strategy, LoRAM, scales deterministic orthogonal bases using pretrained weight magnitudes to simulate spectral gains. Extensive experiments show that LoRAM serves as a strong baseline, retaining the full efficiency of LoRA while matching or outperforming spectral initialization across benchmarks.

cs.LG↗

arXiv2025

Videos are Sample-Efficient Supervisions: Behavior Cloning from Videos via Latent Representations

Humans can efficiently extract knowledge and learn skills from the videos within only a few trials and errors. However, it poses a big challenge to replicate this learning process for autonomous agents, due to the complexity of visual input, the absence of action or reward signals, and the limitations of interaction steps. In this paper, we propose a novel, unsupervised, and sample-efficient framework to achieve imitation learning from videos (ILV), named Behavior Cloning from Videos via Latent Representations (BCV-LR). BCV-LR extracts action-related latent features from high-dimensional video inputs through self-supervised tasks, and then leverages a dynamics-based unsupervised objective to predict latent actions between consecutive frames. The pre-trained latent actions are fine-tuned and efficiently aligned to the real action space online (with collected interactions) for policy behavior cloning. The cloned policy in turn enriches the agent experience for further latent action finetuning, resulting in an iterative policy improvement that is highly sample-efficient. We conduct extensive experiments on a set of challenging visual tasks, including both discrete control and continuous control. BCV-LR enables effective (even expert-level on some tasks) policy performance with only a few interactions, surpassing state-of-the-art ILV baselines and reinforcement learning methods (provided with environmental rewards) in terms of sample efficiency across 24/28 tasks. To the best of our knowledge, this work for the first time demonstrates that videos can support extremely sample-efficient visual policy learning, without the need to access any other expert supervision.

cs.LG↗

arXiv2025

DARTs: A Dual-Path Robust Framework for Anomaly Detection in High-Dimensional Multivariate Time Series

Multivariate time series anomaly detection (MTSAD) aims to accurately identify and localize complex abnormal patterns in the large-scale industrial control systems. While existing approaches excel in recognizing the distinct patterns under the low-dimensional scenarios, they often fail to robustly capture long-range spatiotemporal dependencies when learning representations from the high-dimensional noisy time series. To address these limitations, we propose DARTs, a robust long short-term dual-path framework with window-aware spatiotemporal soft fusion mechanism, which can be primarily decomposed into three complementary components. Specifically, in the short-term path, we introduce a Multi-View Sparse Graph Learner and a Diffusion Multi-Relation Graph Unit that collaborate to adaptively capture hierarchical discriminative short-term spatiotemporal patterns in the high-noise time series. While in the long-term path, we design a Multi-Scale Spatiotemporal Graph Constructor to model salient long-term dynamics within the high-dimensional representation space. Finally, a window-aware spatiotemporal soft-fusion mechanism is introduced to filter the residual noise while seamlessly integrating anomalous patterns. Extensive qualitative and quantitative experimental results across mainstream datasets demonstrate the superiority and robustness of our proposed DARTs. A series of ablation studies are also conducted to explore the crucial design factors of our proposed components. Our code and model will be made publicly open soon.

cs.LG↗

arXiv2025

Speckle-Driven Single-Shot Orbital Angular Momentum Recognition with Ultra-Low Sampling Density

Orbital angular momentum (OAM) recognition of vortex beams is critical for applications ranging from optical communications to quantum technologies. However, conventional approaches designed for free-space propagation struggle when light passes through scattering media, such as multimode fibers (MMF), and often rely on high-resolution sensors with tens of thousands of pixels to record detailed intensity profiles. Here, by harnessing scattering media as intrinsic encoders rather than detrimental factors, we introduce a speckle-driven OAM recognition technique termed patially multiplexed points detection (SMPD). This method extracts intensity information from a few spatially distributed points in a speckle plane, where object feature is naturally multiplexed, thereby transforming scattering from a detrimental effect into an efficient encoding mechanism while drastically reducing sampling requirements. Remarkably, it achieves over 99% retrieval accuracy for OAMs recognition using just 16 sampling points, corresponding to a sampling density of 0.024% compared with conventional imaging-based approaches. Furthermore, spatiotemporally interleaved vortex beams decoding, highcapacity OAM-multiplexed communication, MNIST, and Fashion-MNIST classification are implemented to verify the versatility of SMPD. This work establishes a scalable strategy for efficient optical information processing and fiberbased sensing in complex environments.

physics.optics↗

arXiv2025

Pockels effect induced strong Kerr nonlinearity in a lithium niobate waveguide

The utilization of Kerr nonlinearity in lithium niobate has been extensively investigated over the years. Nevertheless, the practical implementation of Kerr nonlinearity in waveguides has been constrained by the material's inherently low third-order nonlinear coefficients. Here, we present a significant advancement by demonstrating Pockels effect-induced strong Kerr nonlinearity in a periodically poled thin-film lithium niobate waveguide. Both effective four-wave mixing (FWM) and cascaded effective FWM processes are experimentally observed. The induced FWM process achieves a remarkable maximum output power of -8.5 dBm, spanning a wavelength spectrum of over 116.8 nm. Analysis reveals that the induced effective Kerr nonlinearity exhibits a substantial effective nonlinear refractive index as $2.9\times 10^{-15} m^{2}W^{-1}$, corresponding to an effective nonlinear refractive index enhancement factor of $1.6\times 10^{4}$ relative to the intrinsic value. Moreover, a wavelength-converting experiment demonstrates a flat optic-to-optic response over a broadband radiofrequency spectrum, confirming that signal integrity is well preserved after on-chip effective FWM conversion. Therefore, the demonstrated efficient and broadband Pockels effect induced effective Kerr nonlinearity paves the way for novel applications in diverse fields, including spectroscopy, parametric amplification, quantum correlation studies, and wavelength conversion technologies.

physics.optics↗

arXiv2025

Mathematical Proof as a Litmus Test: Revealing Failure Modes of Advanced Large Reasoning Models

Large reasoning models (e.g., R1, o3) have demonstrated remarkable mathematical problem-solving abilities. However, the high reported accuracy of these advanced models on popular datasets, reliance on purely numerical evaluation and potential benchmark leakage, often masks their true reasoning shortcomings. To address this, we propose leveraging the inherent rigor and methodological complexity of mathematical proofs as a diagnostic tool to expose these hidden failures. Specifically, we introduce the RFMDataset (Reveal Failure Modes), a collection of 200 diverse mathematical proof problems, and thoroughly evaluate advanced models' performance on it. Our in-depth analysis of their failures uncovers 10 fine-grained error types, which shows fundamental limitations in current large reasoning models: 1) large reasoning models grapple profoundly with mathematical proofs, with some generating entirely correct proofs for less than 20% of problems and failing even on basic ones; 2) models exhibit a diverse spectrum of reasoning failures, prominently demonstrating the lack of guarantees for the correctness and rigor of single-step reasoning; and 3) models show hallucination and incompleteness during the reasoning process. Our findings reveal that models' self-reflection is insufficient to resolve the current logical dilemmas, necessitating formalized and fine-grained logical training.

cs.AI↗

arXiv2025

Robust Extraction of Global 21 cm Spectrum from Experiments with a Chromatic Beam Based on Physics-Motivated Error Modeling

The extraction of the sky-averaged 21 cm signal from Cosmic Dawn and the Epoch of Reionization faces significant challenges. The bright and anisotropic Galactic foreground, which is 4 - 5 orders of magnitude brighter than the 21 cm signal, when convolved with the inevitably chromatic beam, introduces additional spectral structures that can easily mimic the real 21 cm signal. In this paper, we investigate the signal extraction for a lunar-orbit experiment, where the antenna moves fast in orbit and data from multiple orbits have to be used. We propose a physics-motivated and correlated modeling of both the foreground and the measurement errors. By dividing the sky into multiple regions according to the spectral index distribution and accounting for the full covariance of modeling errors, we jointly fit both the foreground and the 21 cm signal using simulated data for the Discovering the Sky at the Longest wavelength lunar orbit experiment. This method successfully extracts the 21 cm signals of various amplitudes from the simulated data even for a testing antenna with a relatively high level of chromaticity. This approach, which is robust against moderate beam chromaticity, significantly relaxes the stringent design and manufacturing requirements for the antenna, offering a practical solution for future 21 cm global signal experiments either on the ground or in space.

astro-ph.IM↗