arXiv subjects

Haiyang Li

Publications and source records attributed to Haiyang Li.

At least 19 recordsLinked to original sources

arXiv2026

KANResDiff: Learning Local Residual Diffusion via Kolmogorov-Arnold Network for Ambiguous Medical Image Segmentation

Ambiguous medical image segmentation aims to provide a series of diverse but plausible segmentation hypotheses. However, existing methods introduce stochasticity in a fixed and pre-defined manner, failing to form a progressive semantic modeling process. To address these challenges, we propose KANResDiff to learn local residual diffusion with Kolmogorov-Arnold Network, thereby assigning distinct roles across stages for ambiguity modeling. Specifically, we propose Independent Time Encoding that offers spline-based time embeddings instead of linear ones from MLPs, which enhances the independence across inference stages and assigns progressive semantic roles to different stages. We propose Residual Schrodinger Bridge that injects deterministic residual prior with learnable weights by constructing local Schrodinger Bridge instead of following manually settings, achieving a flexible deterministic-stochastic interaction and stage-aware ambiguity modeling thanks to local optimal diffusion path. Extensive experimental results on two public datasets demonstrate that KANResDiff achieves SOTA performance on GED and HM-IoU, with maximum improvements of 16.8% and 7.7%, respectively, while maintaining competitive performance on the MDM metric. Source code is available at https://github.com/PerceptionComputingLab/KANResDiff.

cs.CV↗

arXiv2026

AGVBench: A Reliability-Oriented Benchmark of Data Augmentation for Vein Recognition

Vein recognition is a secure biometric technology often constrained by limited annotated data and imaging variations. While data augmentation mitigates this, strategies designed for natural images may disrupt the fine-grained topology and textures essential for identity discrimination. We present AGVBench, which evaluates 30 representative augmentation strategies on five public palm- and finger-vein datasets with seven backbone architectures, covering classic CNNs, vision transformers, and vein-specific recognition models. Our results show that multi-image mixing methods (e.g., MixUp, PuzzleMix, StarMixup) generally provide the strongest recognition performance. However, they are often poorly calibrated and vulnerable to adversarial perturbations, revealing a clear inconsistency between clean accuracy and adversarial security. We also find that severe geometric transformations frequently degrade recognition, which is potentially due to feature misalignment or spatial cropping, and that augmentation effectiveness varies across palm and finger vein datasets. These findings prove that accuracy-centric evaluation is insufficient for biometric augmentation. AGVBench provides standardized protocols to support reproducible research and guide the design of reliable, secure, and robust vein recognition systems. Our codebase is available at https://github.com/Advance-VeinTech-Innovators/AGVBench.

cs.CV↗

arXiv2025

Adaptive Fusion Graph Network for 3D Strain Field Prediction in Solid Rocket Motor Grains

Local high strain in solid rocket motor grains is a primary cause of structural failure. However, traditional numerical simulations are computationally expensive, and existing surrogate models cannot explicitly establish geometric models and accurately capture high-strain regions. Therefore, this paper proposes an adaptive graph network, GrainGNet, which employs an adaptive pooling dynamic node selection mechanism to effectively preserve the key mechanical features of structurally critical regions, while concurrently utilising feature fusion to transmit deep features and enhance the model's representational capacity. In the joint prediction task involving four sequential conditions--curing and cooling, storage, overloading, and ignition--GrainGNet reduces the mean squared error by 62.8% compared to the baseline graph U-Net model, with only a 5.2% increase in parameter count and an approximately sevenfold improvement in training efficiency. Furthermore, in the high-strain regions of debonding seams, the prediction error is further reduced by 33% compared to the second-best method, offering a computationally efficient and high-fidelity approach to evaluate motor structural safety.

physics.app-ph↗

arXiv2025

Visualized Geometric Phase of Caustic Geometric Beams

Detecting Pancharatnam-Berry geometric phases of light typically requires interferometry or diffraction through a specially truncated aperture. Here, we introduce a simpler method that allows direct and fully visual detection of geometric phases in structured light without using interferometers or beam truncation. Our approach takes advantage of the geometric phase that naturally arises in SU(2) structured beams, where spatial wave packets follow caustic trajectories during propagation. By observing the evolution of these caustic-linked wave packets, we directly visualize both the geometric phase and the Gouy phase. This visual detection method provides new insight into geometric phases in complex optical fields and expands the possibilities for designing optical systems that exploit phase geometry.

physics.optics↗

arXiv2025

M3PD Dataset: Dual-view Photoplethysmography (PPG) Using Front-and-rear Cameras of Smartphones in Lab and Clinical Settings

Portable physiological monitoring is essential for early detection and management of cardiovascular disease, but current methods often require specialized equipment that limits accessibility or impose impractical postures that patients cannot maintain. Video-based photoplethysmography on smartphones offers a convenient noninvasive alternative, yet it still faces reliability challenges caused by motion artifacts, lighting variations, and single-view constraints. Few studies have demonstrated reliable application to cardiovascular patients, and no widely used open datasets exist for cross-device accuracy. To address these limitations, we introduce the M3PD dataset, the first publicly available dual-view mobile photoplethysmography dataset, comprising synchronized facial and fingertip videos captured simultaneously via front and rear smartphone cameras from 60 participants (including 47 cardiovascular patients). Building on this dual-view setting, we further propose F3Mamba, which fuses the facial and fingertip views through Mamba-based temporal modeling. The model reduces heart-rate error by 21.9 to 30.2 percent over existing single-view baselines while improving robustness in challenging real-world scenarios. Data and code: https://github.com/Health-HCI-Group/F3Mamba.

cs.CV↗

arXiv2025

Seemingly Redundant Modules Enhance Robust Odor Learning in Fruit Flies

Biological circuits have evolved to incorporate multiple modules that perform similar functions. In the fly olfactory circuit, both lateral inhibition (LI) and neuronal spike frequency adaptation (SFA) are thought to enhance pattern separation for odor learning. However, it remains unclear whether these mechanisms play redundant or distinct roles in this process. In this study, we present a computational model of the fly olfactory circuit to investigate odor discrimination under varying noise conditions that simulate complex environments. Our results show that LI primarily enhances odor discrimination in low- and medium-noise scenarios, but this benefit diminishes and may reverse under higher-noise conditions. In contrast, SFA consistently improves discrimination across all noise levels. LI is preferentially engaged in low- and medium-noise environments, whereas SFA dominates in high-noise settings. When combined, these two sparsification mechanisms enable optimal discrimination performance. This work demonstrates that seemingly redundant modules in biological circuits can, in fact, be essential for achieving optimal learning in complex contexts.

cs.NE↗

arXiv2025

Towards Unified Multimodal Misinformation Detection in Social Media: A Benchmark Dataset and Baseline

Detecting deceptive multimodal content on social media has become an increasingly important problem. Two major types of deception dominate: human-crafted misinformation (e.g., rumors and misleading posts) and AI-generated content produced by image synthesis models or vision-language models (VLMs). However, these two types are usually addressed as separate tasks. Consequently, existing models are often specialized for only one type of fake content. In real deployments, however, the fake-content type of an incoming multimodal post is typically unknown, which limits the practicality of such specialized systems. To study this setting, we build OmniFake, a benchmark with 98K samples that combines human-curated misinformation from existing resources with newly created AI-generated examples. To address this new task, we propose Unified Multimodal Fake Content Detection (UMFDet), a framework designed to handle both types of deception. UMFDet builds on a VLM backbone augmented with a Category-aware Mixture-of-Experts (CMoE) adapter to capture category-specific cues. We further introduce an Expert-wise Discriminative Regularization to enforce intra-expert compactness. In addition, cross-modal consistency alignment is proposed to improve the perceptual capability of experts for handling different deception types. Experiments show that UMFDet consistently outperforms competitive specialized baselines across both deception types.

cs.AI↗

arXiv2025

NEFT: A Unified Transformer Framework for Efficient Near-Field CSI Feedback in XL-MIMO Systems

Extremely large-scale multiple-input multiple-output (XL-MIMO) systems, operating in the near-field region due to their massive antenna arrays, are key enablers of next-generation wireless communications but face significant challenges in channel state information (CSI) feedback. Deep learning has emerged as a powerful tool by learning compact channel features for feedback. However, existing methods struggle to capture the intricate structure of near-field CSI and incur prohibitive computational overhead on practical mobile devices. To overcome these limitations, we propose the near-field efficient feedback Transformer (NEFT) family for accurate near-field CSI feedback with reduced overhead under diverse hardware constraints. NEFT builds on a hierarchical vision Transformer backbone with progressive token reduction and multi-scale feature extraction, enabling compact and effective modeling of near-field channel characteristics. Furthermore, NEFT is extended with lightweight variants: NEFT-Compact applies multi-level knowledge distillation (KD) to reduce model complexity while preserving accuracy; NEFT-Hybrid adopts an attention-free CNN encoder to reduce encoder-side computation; and NEFT-Edge combines NEFT-Hybrid with KD to enable deployment on highly resource-constrained edge devices. Extensive simulations show that NEFT achieves a 15--21dB improvement in normalized mean-squared error over state-of-the-art methods, NEFT-Compact and NEFT-Edge reduce total FLOPs by 25-36% with negligible accuracy loss, while NEFT-Hybrid reduces encoder-side complexity by up to 64%, enabling deployment in highly asymmetric device scenarios. These results establish NEFT as a practical and scalable solution for near-field CSI feedback in XL-MIMO systems.

eess.SP↗

arXiv2025

GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models

We present GLM-4.5, an open-source Mixture-of-Experts (MoE) large language model with 355B total parameters and 32B activated parameters, featuring a hybrid reasoning method that supports both thinking and direct response modes. Through multi-stage training on 23T tokens and comprehensive post-training with expert model iteration and reinforcement learning, GLM-4.5 achieves strong performance across agentic, reasoning, and coding (ARC) tasks, scoring 70.1% on TAU-Bench, 91.0% on AIME 24, and 64.2% on SWE-bench Verified. With much fewer parameters than several competitors, GLM-4.5 ranks 3rd overall among all evaluated models and 2nd on agentic benchmarks. We release both GLM-4.5 (355B parameters) and a compact version, GLM-4.5-Air (106B parameters), to advance research in reasoning and agentic AI systems. Code, models, and more information are available at https://github.com/zai-org/GLM-4.5.

cs.CL↗

arXiv2025

MRG-Bench: Evaluating and Exploring the Requirements of Context for Repository-Level Code Generation

Large Language Models (LLMs) have demonstrated impressive capabilities in code generation. However, current evaluation datasets suffer from issues such as the lack of runnable test cases, deviation from the distribution of real-world code, and the ability to evaluate only the Python language. These limitations undermine the credibility of the evaluation results. To address these limitations, we introduce \textbf{MRG-Bench} (Multi-language Repository-level Code Generation Benchmark), a novel dataset that provides a more accurate evaluation of LLMs in practical repository-level code generation tasks. MRG-Bench has three main features: (1) practical data sourced from real-world code repositories that align to the practical distribution, (2) multiple programming languages support, including Python, Java, and Go, and (3) project-level runnable test cases to assess the quality of the generated code. Based on MRG-Bench, we conducted extensive experiments including large language models, long-context models, and RAG-related methods. These evaluation results demonstrate that \textbf{current repository-level code generation techniques suffer from significant performance deficiencies}. To further investigate why models fail, we designed novel experiments to annotate the underlying causes of generation errors. The results explicitly show that the majority of methods suffer from "\textbf{difficulty in understanding user requirements}," failing to comprehend their assigned tasks accurately. Moreover, the impact of different repository-level contexts on this issue exhibits significant disparities across different programming languages, suggesting that, in practice, specialized contextual information needs to be designed for different languages.

cs.SE↗

arXiv2025

I Know What You Said: Unveiling Hardware Cache Side-Channels in Local Large Language Model Inference

Large Language Models (LLMs) that can be deployed locally have recently gained popularity for privacy-sensitive tasks, with companies such as Meta, Google, and Intel playing significant roles in their development. However, the security of local LLMs through the lens of hardware cache side-channels remains unexplored. In this paper, we unveil novel side-channel vulnerabilities in local LLM inference: token value and token position leakage, which can expose both the victim's input and output text, thereby compromising user privacy. Specifically, we found that adversaries can infer the token values from the cache access patterns of the token embedding operation, and deduce the token positions from the timing of autoregressive decoding phases. To demonstrate the potential of these leaks, we design a novel eavesdropping attack framework targeting both open-source and proprietary LLM inference systems. The attack framework does not directly interact with the victim's LLM and can be executed without privilege. We evaluate the attack on a range of practical local LLM deployments (e.g., Llama, Falcon, and Gemma), and the results show that our attack achieves promising accuracy. The restored output and input text have an average edit distance of 5.2% and 17.3% to the ground truth, respectively. Furthermore, the reconstructed texts achieve average cosine similarity scores of 98.7% (input) and 98.0% (output).

cs.CR↗

arXiv2025

Integrating earth observation data into the tri-environmental evaluation of the economic cost of natural disasters: a case study of 2025 LA wildfire

Wildfires in urbanized regions, particularly within the wildland-urban interface, have significantly intensified in frequency and severity, driven by rapid urban expansion and climate change. This study aims to provide a comprehensive, fine-grained evaluation of the recent 2025 Los Angeles wildfire's impacts, through a multi-source, tri-environmental framework in the social, built and natural environmental dimensions. This study employed a spatiotemporal wildfire impact assessment method based on daily satellite fire detections from the Visible Infrared Imaging Radiometer Suite (VIIRS), infrastructure data from OpenStreetMap, and high-resolution dasymetric population modeling to capture the dynamic progression of wildfire events in two distinct Los Angeles County regions, Eaton and Palisades, which occurred in January 2025. The modelling result estimated that the total direct economic losses reached approximately 4.86 billion USD with the highest single-day losses recorded on January 8 in both districts. Population exposure reached a daily maximum of 4,342 residents in Eaton and 3,926 residents in Palisades. Our modelling results highlight early, severe ecological and infrastructural damage in Palisades, as well as delayed, intense social and economic disruptions in Eaton. This tri-environmental framework underscores the necessity for tailored, equitable wildfire management strategies, enabling more effective emergency responses, targeted urban planning, and community resilience enhancement. Our study contributes a highly replicable tri-environmental framework for evaluating the natural, built and social environmental costs of natural disasters, which can be applied to future risk profiling, hazard mitigation, and environmental management in the era of climate change.

econ.GN↗

arXiv2025

Robust and digital hyper-polarization protocol of nuclear spins via magic sequential sequence

Hyper-polarization of nuclear spins is crucial for advancing nuclear magnetic resonance (NMR) and quantum information technologies, as nuclear spins typically exhibit extremely low polarization at room temperature due to their small gyro-magnetic ratios. A promising approach to achieving high nuclear spin polarization is transferring the polarization of electron to nuclear spin. The nitrogen-vacancy (NV) center in diamond has emerged as a highly effective medium for this purpose, and various hyper-polarization protocols have been developed. Among these, the pulsed polarization (PulsePol) method has been extensively studied due to its robustness against static energy shifts of the electron spin. In this study, we introduce a sequential polarization protocol and identify a series of magic and digital sequences for hyper-polarizing nuclear spins. Notably, we demonstrate that some of these magic sequences exhibit significantly greater robustness compared to the PulsePol protocol in the presence of finite half $\pi$ pulse duration of the protocol. This enhanced robustness positions our protocol as a more suitable candidate for hyper-polarizing nuclear spins species with large gyromagnetic ratios and also ensures better compatibility with high-efficiency readout techniques at high magnetic fields. Additionally, the generality of our protocol allows for its direct application to other solid-state quantum systems beyond the NV center.

quant-ph↗

arXiv2024

StreetviewLLM: Extracting Geographic Information Using a Chain-of-Thought Multimodal Large Language Model

Geospatial predictions are crucial for diverse fields such as disaster management, urban planning, and public health. Traditional machine learning methods often face limitations when handling unstructured or multi-modal data like street view imagery. To address these challenges, we propose StreetViewLLM, a novel framework that integrates a large language model with the chain-of-thought reasoning and multimodal data sources. By combining street view imagery with geographic coordinates and textual data, StreetViewLLM improves the precision and granularity of geospatial predictions. Using retrieval-augmented generation techniques, our approach enhances geographic information extraction, enabling a detailed analysis of urban environments. The model has been applied to seven global cities, including Hong Kong, Tokyo, Singapore, Los Angeles, New York, London, and Paris, demonstrating superior performance in predicting urban indicators, including population density, accessibility to healthcare, normalized difference vegetation index, building height, and impervious surface. The results show that StreetViewLLM consistently outperforms baseline models, offering improved predictive accuracy and deeper insights into the built environment. This research opens new opportunities for integrating the large language model into urban analytics, decision-making in urban planning, infrastructure management, and environmental monitoring.

cs.CL↗

arXiv2024

StarLKNet: Star Mixup with Large Kernel Networks for Palm Vein Identification

As a representative of a new generation of biometrics, vein identification technology offers a high level of security and convenience.Convolutional neural networks (CNNs), a prominent class of deep learning architectures, have been extensively utilized for vein identification. Since their performance and robustness are limited by small \emph{Effective Receptive Fields} (\emph{e.g.}, 3$\times$3 kernels) and insufficient training samples, however, they are unable to extract global feature representations from vein images effectively. To address these issues, we propose \textbf{StarLKNet}, a large kernel convolution-based palm-vein identification network, with the Mixup approach.Our StarMix learns effectively the distribution of vein features to expand samples. To enable CNNs to capture comprehensive feature representations from palm-vein images, we explored the effect of convolutional kernel size on the performance of palm-vein identification networks and designed LaKNet, a network leveraging large kernel convolution and gating mechanism. In light of the current state of knowledge, this represents an inaugural instance of the deployment of a CNN with large kernels in the domain of vein identification. Extensive experiments were conducted to validate the performance of StarLKNet on two public palm-vein datasets. The results demonstrated that \textbf{StarMix} provided superior augmentation, and \textbf{LakNet} exhibited more stable performance gains compared to mainstream approaches, resulting in the highest identification accuracy and lowest identification error.

cs.CV↗

arXiv2023

SR-HetGNN:Session-based Recommendation with Heterogeneous Graph Neural Network

The Session-Based Recommendation System aims to predict the user's next click based on their previous session sequence. The current studies generally learn user preferences according to the transitions of items in the user's session sequence. However, other effective information in the session sequence, such as user profiles, are largely ignored which may lead to the model unable to learn the user's specific preferences. In this paper, we propose SR-HetGNN, a novel session recommendation method that uses a heterogeneous graph neural network (HetGNN) to learn session embeddings and capture the specific preferences of anonymous users. Specifically, SR-HetGNN first constructs heterogeneous graphs containing various types of nodes according to the session sequence, which can capture the dependencies among items, users, and sessions. Second, HetGNN captures the complex transitions between items and learns the item embeddings containing user information. Finally, local and global session embeddings are combined with the attentional network to obtain the final session embedding, considering the influence of users' long and short-term preferences. SR-HetGNN is shown to be superior to the existing state-of-the-art session-based recommendation methods through extensive experiments over two real large datasets Diginetica and Tmall.

cs.IR↗

arXiv2020

Adaptive iterative singular value thresholding algorithm to low-rank matrix recovery

The problem of recovering a low-rank matrix from the linear constraints, known as affine matrix rank minimization problem, has been attracting extensive attention in recent years. In general, affine matrix rank minimization problem is a NP-hard. In our latest work, a non-convex fraction function is studied to approximate the rank function in affine matrix rank minimization problem and translate the NP-hard affine matrix rank minimization problem into a transformed affine matrix rank minimization problem. A scheme of iterative singular value thresholding algorithm is generated to solve the regularized transformed affine matrix rank minimization problem. However, one of the drawbacks for our iterative singular value thresholding algorithm is that the parameter $a$, which influences the behaviour of non-convex fraction function in the regularized transformed affine matrix rank minimization problem, needs to be determined manually in every simulation. In fact, how to determine the optimal parameter $a$ is not an easy problem. Here instead, in this paper, we will generate an adaptive iterative singular value thresholding algorithm to solve the regularized transformed affine matrix rank minimization problem. When doing so, our new algorithm will be intelligent both for the choice of the regularized parameter $λ$ and the parameter $a$.

math.OC↗

arXiv2019

Autonomous Time-Optimal Many-Revolution Orbit Raising for Electric Propulsion GEO Satellites via Neural Networks

Geostationary Earth orbit (GEO) satellites are of great significance in the space market. Low-thrust propulsion has been highly developed in the last decades because it is fuel-saving. Therefore, the design of GEO satellites is rapidly changing from classic high-thrust propulsion more and more toward low-thrust propulsion. However, the transfer time will be quite long using low-thrust propulsion and it will be very expensive if the ground supports the whole orbit raising. Therefore, autonomous orbit raising is necessary. Deep neural networks are trained to learn the optimal control. Results show that DNNs can be applied in this long-duration optimal control problem and have excellent performance.

math.OC↗