arXiv subjects

Yin Huang

Publications and source records attributed to Yin Huang.

At least 37 records · Page 2Linked to original sources

arXiv2025

Multi-Scale Target-Aware Representation Learning for Fundus Image Enhancement

High-quality fundus images provide essential anatomical information for clinical screening and ophthalmic disease diagnosis. Yet, due to hardware limitations, operational variability, and patient compliance, fundus images often suffer from low resolution and signal-to-noise ratio. Recent years have witnessed promising progress in fundus image enhancement. However, existing works usually focus on restoring structural details or global characteristics of fundus images, lacking a unified image enhancement framework to recover comprehensive multi-scale information. Moreover, few methods pinpoint the target of image enhancement, e.g., lesions, which is crucial for medical image-based diagnosis. To address these challenges, we propose a multi-scale target-aware representation learning framework (MTRL-FIE) for efficient fundus image enhancement. Specifically, we propose a multi-scale feature encoder (MFE) that employs wavelet decomposition to embed both low-frequency structural information and high-frequency details. Next, we design a structure-preserving hierarchical decoder (SHD) to fuse multi-scale feature embeddings for real fundus image restoration. SHD integrates hierarchical fusion and group attention mechanisms to achieve adaptive feature fusion while retaining local structural smoothness. Meanwhile, a target-aware feature aggregation (TFA) module is used to enhance pathological regions and reduce artifacts. Experimental results on multiple fundus image datasets demonstrate the effectiveness and generalizability of MTRL-FIE for fundus image enhancement. Compared to state-of-the-art methods, MTRL-FIE achieves superior enhancement performance with a more lightweight architecture. Furthermore, our approach generalizes to other ophthalmic image processing tasks without supervised fine-tuning, highlighting its potential for clinical applications.

eess.IV↗

arXiv2025

Search for the $D^{}\bar{D}^{}$ Molecular State $X_{2}(4013)$ in $K^{-}p$ and $pp$ Collisions

Motivated by the interpretation of $X(3872)$ as a $D\bar{D}^{*}$ molecular state, heavy-quark spin symmetry predicts a spin-2 partner, $X_{2}(4013)$, which can be regarded as a $D^{*}\bar{D}^{*}$ molecule with quantum numbers $J^{PC} = 2^{++}$. Its experimental confirmation, however, remains elusive. In this work, we investigate the production mechanisms of $X_{2}(4013)$ in the reactions $K^{-}p \to Λ_{c}^{+} D_{s}^{-} X_{2}(4013)$ and $pp \to Λ_{c}^{+}Λ_{c}^{+} X_{2}(4013)$ within an effective Lagrangian framework. The production processes are modeled via $t$-channel $D/\bar{D}^{*}$ meson exchanges, while initial-state interactions (ISI) mediated by Pomeron and Reggeon exchanges are also taken into account. Our calculations indicate that the total cross sections can reach the pb level, suggesting that $X_{2}(4013)$ may be accessible at current and future experiments such as AMBER@CERN and LHCb. Inclusion of ISI enhances the cross sections by nearly one order of magnitude. The differential distributions show distinct angular behaviors for the two reactions: the $K^{-}p$ reaction exhibits a forward-peaked distribution, whereas the $pp$ reaction shows a dip near central angles. This study provides a quantitative theoretical benchmark for future experimental searches of $X_2(4013)$ and highlights the importance of initial-state interactions (ISI) in high-energy particle investigations.

hep-ph↗

arXiv2025

Knowledge Extraction on Semi-Structured Content: Does It Remain Relevant for Question Answering in the Era of LLMs?

The advent of Large Language Models (LLMs) has significantly advanced web-based Question Answering (QA) systems over semi-structured content, raising questions about the continued utility of knowledge extraction for question answering. This paper investigates the value of triple extraction in this new paradigm by extending an existing benchmark with knowledge extraction annotations and evaluating commercial and open-source LLMs of varying sizes. Our results show that web-scale knowledge extraction remains a challenging task for LLMs. Despite achieving high QA accuracy, LLMs can still benefit from knowledge extraction, through augmentation with extracted triples and multi-task learning. These findings provide insights into the evolving role of knowledge triple extraction in web-based QA and highlight strategies for maximizing LLM effectiveness across different model sizes and resource settings.

cs.CL↗

arXiv2025

Embodied Representation Alignment with Mirror Neurons

Mirror neurons are a class of neurons that activate both when an individual observes an action and when they perform the same action. This mechanism reveals a fundamental interplay between action understanding and embodied execution, suggesting that these two abilities are inherently connected. Nonetheless, existing machine learning methods largely overlook this interplay, treating these abilities as separate tasks. In this study, we provide a unified perspective in modeling them through the lens of representation learning. We first observe that their intermediate representations spontaneously align. Inspired by mirror neurons, we further introduce an approach that explicitly aligns the representations of observed and executed actions. Specifically, we employ two linear layers to map the representations to a shared latent space, where contrastive learning enforces the alignment of corresponding representations, effectively maximizing their mutual information. Experiments demonstrate that this simple approach fosters mutual synergy between the two tasks, effectively improving representation quality and generalization.

cs.AI↗

arXiv2025

Reconfigurable structural color generation in multicolor pixels using phase-change materials

We introduce multilayer structures based on phase-change materials for reconfigurable structural color generation. These structures can produce multiple distinct colors within a single pixel. Specifically, we design structures that generate either two or four maximally distinct structural colors. We employ a memetic optimization algorithm coupled with the impedance method to identify the optimal combination of materials and layer thicknesses that maximizes the distinctiveness of the generated colors. We demonstrate that our design approach achieves strong color contrast between the generated colors. The proposed multilayer design eliminates the need for subwavelength lithography, making it highly suitable for large-scale applications. Our results could lead to a new class of single-cell multicolor pixels which retain each color without power consumption, making them particularly appealing for low refresh rate displays.

physics.optics↗

arXiv2025

Parking Availability Prediction via Fusing Multi-Source Data with A Self-Supervised Learning Enhanced Spatio-Temporal Inverted Transformer

The rapid growth of private car ownership has worsened the urban parking predicament, underscoring the need for accurate and effective parking availability prediction to support urban planning and management. To address key limitations in modeling spatio-temporal dependencies and exploiting multi-source data for parking availability prediction, this study proposes a novel approach with SST-iTransformer. The methodology leverages K-means clustering to establish parking cluster zones (PCZs), extracting and integrating traffic demand characteristics from various transportation modes (i.e., metro, bus, online ride-hailing, and taxi) associated with the targeted parking lots. Upgraded on vanilla iTransformer, SST-iTransformer integrates masking-reconstruction-based pretext tasks for self-supervised spatio-temporal representation learning, and features an innovative dual-branch attention mechanism: Series Attention captures long-term temporal dependencies via patching operations, while Channel Attention models cross-variate interactions through inverted dimensions. Extensive experiments using real-world data from Chengdu, China, demonstrate that SST-iTransformer outperforms baseline deep learning models (including Informer, Autoformer, Crossformer, and iTransformer), achieving state-of-the-art performance with the lowest mean squared error (MSE) and competitive mean absolute error (MAE). Comprehensive ablation studies quantitatively reveal the relative importance of different data sources: incorporating ride-hailing data provides the largest performance gains, followed by taxi, whereas fixed-route transit features (bus/metro) contribute marginally. Spatial correlation analysis further confirms that excluding historical data from correlated parking lots within PCZs leads to substantial performance degradation, underscoring the importance of modeling spatial dependencies.

cs.LG↗

arXiv2025

PrismRAG: Boosting RAG Factuality with Distractor Resilience and Strategized Reasoning

Retrieval-augmented generation (RAG) often falls short when retrieved context includes confusing semi-relevant passages, or when answering questions require deep contextual understanding and reasoning. We propose an efficient fine-tuning framework, called PrismRAG, that (i) trains the model with distractor-aware QA pairs mixing gold evidence with subtle distractor passages, and (ii) instills reasoning-centric habits that make the LLM plan, rationalize, and synthesize without relying on extensive human engineered instructions. Evaluated across 12 open-book RAG QA benchmarks spanning diverse application domains and scenarios, PrismRAG improves average factuality by 5.4%, outperforming state-of-the-art solutions.

cs.CL↗

arXiv2025

Multimodal Federated Learning: A Survey through the Lens of Different FL Paradigms

Multimodal Federated Learning (MFL) lies at the intersection of two pivotal research areas: leveraging complementary information from multiple modalities to improve downstream inference performance and enabling distributed training to enhance efficiency and preserve privacy. Despite the growing interest in MFL, there is currently no comprehensive taxonomy that organizes MFL through the lens of different Federated Learning (FL) paradigms. This perspective is important because multimodal data introduces distinct challenges across various FL settings. These challenges, including modality heterogeneity, privacy heterogeneity, and communication inefficiency, are fundamentally different from those encountered in traditional unimodal or non-FL scenarios. In this paper, we systematically examine MFL within the context of three major FL paradigms: horizontal FL (HFL), vertical FL (VFL), and hybrid FL. For each paradigm, we present the problem formulation, review representative training algorithms, and highlight the most prominent challenge introduced by multimodal data in distributed settings. We also discuss open challenges and provide insights for future research. By establishing this taxonomy, we aim to uncover the novel challenges posed by multimodal data from the perspective of different FL paradigms and to offer a new lens through which to understand and advance the development of MFL.

cs.LG↗

arXiv2025

Interplay between defects and the non-Hermitian skin effect

The non-Hermitian skin effect (NHSE) is an intriguing phenomenon in which an extensive number of bulk eigenstates localize at the boundaries of a non-Hermitian system with non-reciprocal hoppings. Here we study the interplay of this effect and a defect in non-reciprocal one-dimensional lattices. We show that the interplay of the NHSE and defects is size-dependent. We demonstrate a novel class of hybrid skin-defect states in finite-size systems resulting from the coupling between the skin and defect states. Next, we consider a single defect in a topologically nontrivial lattice with time-reversal symmetry based on the non-reciprocal Su-Schrieffer-Heeger configuration. We unveil how topologically nontrivial defect states and non-Hermiticity interplay by competing with each other, exhibiting a transition from topologically nontrivial defect states to skin states. In addition, we show that decreasing the defect strength can result in a transition from trivial defect states to skin states. Our work promotes the understanding of the interplay between defects and the NHSE, and especially the importance of the energy spectra of the system with the defect under periodic boundary conditions.

quant-ph↗

arXiv2025

A Self-Supervised Transformer for Unusable Shared Bike Detection

The rapid expansion of bike-sharing systems (BSS) has greatly improved urban "last-mile" connectivity, yet large-scale deployments face escalating operational challenges, particularly in detecting faulty bikes. Existing detection approaches either rely on static model-based thresholds that overlook dynamic spatiotemporal (ST) usage patterns or employ supervised learning methods that struggle with label scarcity and class imbalance. To address these limitations, this paper proposes a novel Self-Supervised Transformer (SSTransformer) framework for automatically detecting unusable shared bikes, leveraging ST features extracted from GPS trajectories and trip records. The model incorporates a self-supervised pre-training strategy to enhance its feature extraction capabilities, followed by fine-tuning for efficient status recognition. In the pre-training phase, the Transformer encoder learns generalized representations of bike movement via a self-supervised objective; in the fine-tuning phase, the encoder is adapted to a downstream binary classification task. Comprehensive experiments on a real-world dataset of 10,730 bikes (1,870 unusable, 8,860 normal) from Chengdu, China, demonstrate that SSTransformer significantly outperforms traditional machine learning, ensemble learning, and deep learning baselines, achieving the best accuracy (97.81%), precision (0.8889), and F1-score (0.9358). This work highlights the effectiveness of self-supervised Transformer on ST data for capturing complex anomalies in BSS, paving the way toward more reliable and scalable maintenance solutions for shared mobility.

cs.LG↗

arXiv2025

A Survey on Parameter-Efficient Fine-Tuning for Foundation Models in Federated Learning

Foundation models have revolutionized artificial intelligence by providing robust, versatile architectures pre-trained on large-scale datasets. However, adapting these massive models to specific downstream tasks requires fine-tuning, which can be prohibitively expensive in computational resources. Parameter-Efficient Fine-Tuning (PEFT) methods address this challenge by selectively updating only a small subset of parameters. Meanwhile, Federated Learning (FL) enables collaborative model training across distributed clients without sharing raw data, making it ideal for privacy-sensitive applications. This survey provides a comprehensive review of the integration of PEFT techniques within federated learning environments. We systematically categorize existing approaches into three main groups: Additive PEFT (which introduces new trainable parameters), Selective PEFT (which fine-tunes only subsets of existing parameters), and Reparameterized PEFT (which transforms model architectures to enable efficient updates). For each category, we analyze how these methods address the unique challenges of federated settings, including data heterogeneity, communication efficiency, computational constraints, and privacy concerns. We further organize the literature based on application domains, covering both natural language processing and computer vision tasks. Finally, we discuss promising research directions, including scaling to larger foundation models, theoretical analysis of federated PEFT methods, and sustainable approaches for resource-constrained environments.

cs.LG↗

arXiv2025

Direct Evidence for the $\bar{D}D^/D\bar{D}^$ Molecular Nature of $G(3900)$ Through Triangular Singularity Mechanisms

The exotic hadron $G(3900)$, initially observed in the process $e^+ e^- \to D\bar{D}$, has been further supported by analyses from the BESIII Collaboration, which classify it as a $P$-wave molecular state of $D^{+}D^{*-}/D^{-}D^{*+}$. However, theoretical discussions raise concerns about its status as a true particle, emphasizing the need for additional studies. In this Letter, we employ the triangular singularity mechanism to investigate $G(3900)$ across various reaction channels, allowing us to produce significant peaks without relying on the existence of a real particle. We identify $X(4020)$, $Y(4320)$, and the tentative $X(4014)$ as potential sources of these peaks via decay processes to $γG(3900)$ or $πG(3900)$. We stress the importance of experimental explorations of $e^+ e^- \to X/Y \to γ(π) G(3900)$, which are essential for confirming the molecular composition of $D^{+}D^{*-}/D^{-}D^{*+}$ and refining the mass measurement of $G(3900)$.

hep-ph↗

arXiv2024

Predicting New Above-Threshold Molecular States Via Triangular Singularities

Considering that the experimentally observed molecular states are significantly fewer than those predicted theoretically, and that these states are traditionally classified as lying below thresholds while several candidates are found above them, we propose to broaden the definition of molecular states to include those that exist just above the thresholds. Identifying resonance peaks in invariant mass distributions and scattering cross-sections is crucial for probing these states, yet the mechanisms responsible for such enhancements remain unclear, complicating our understanding of new particle production. While the peaks linked to triangular singularities do not correspond to true hadronic states, the associated production mechanisms may provide valuable insight into the search for genuine hadrons. In this work, we propose employing the triangular singularity mechanism to theoretically investigate yet-to-be-observed molecular states, particularly those that could test heavy quark symmetry. We argue that these states may have true masses surpassing the thresholds of their constituent components, rather than being predicted to be below them by theoretical models. Our findings suggest the possible existence of 18 predicted heavy quark molecular states, including $X(4014)$, $Z_{cs}(4123)$, $X_{c0}(4500)$, $X_{c1}(4685)$, $Y(4320)$, $Z(4430)$, and $Υ(11020)$, which are posited to contain $D^{*}\bar{D}^{*}$, $D^{*}\bar{D}^{*}_{s}$, $ D_{s1}^{+}D_{s}^{-} $,$ D_{s1}^{+}D_{s}^{*-} $, $D_1\bar{D}$, $D_1\bar{D}^{*}$, and $B_{1}(5721)\bar{B}$ constituents, respectively. The recognition of these states would substantiate heavy quark symmetry and enhance our understanding of hadronic dynamics and molecular states formation.

hep-ph↗

arXiv2024

Leverage Multi-source Traffic Demand Data Fusion with Transformer Model for Urban Parking Prediction

The escalation in urban private car ownership has worsened the urban parking predicament, necessitating effective parking availability prediction for urban planning and management. However, the existing prediction methods suffer from low prediction accuracy with the lack of spatial-temporal correlation features related to parking volume, and neglect of flow patterns and correlations between similar parking lots within certain areas. To address these challenges, this study proposes a parking availability prediction framework integrating spatial-temporal deep learning with multi-source data fusion, encompassing traffic demand data from multiple sources (e.g., metro, bus, taxi services), and parking lot data. The framework is based on the Transformer as the spatial-temporal deep learning model and leverages K-means clustering to establish parking cluster zones, extracting and integrating traffic demand characteristics from various transportation modes (i.e., metro, bus, online ride-hailing, and taxi) connected to parking lots. Real-world empirical data was used to verify the effectiveness of the proposed method compared with different machine learning, deep learning, and traditional statistical models for predicting parking availability. Experimental results reveal that, with the proposed pipeline, the developed Transformer model outperforms other models in terms of various metrics, e.g., Mean Squared Error (MSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). By fusing multi-source demanding data with spatial-temporal deep learning techniques, this approach offers the potential to develop parking availability prediction systems that furnish more accurate and timely information to both drivers and urban planners, thereby fostering more efficient and sustainable urban mobility.

cs.LG↗

arXiv2024

Quantum Entanglement Path Selection and Qubit Allocation via Adversarial Group Neural Bandits

Quantum Data Networks (QDNs) have emerged as a promising framework in the field of information processing and transmission, harnessing the principles of quantum mechanics. QDNs utilize a quantum teleportation technique through long-distance entanglement connections, encoding data information in quantum bits (qubits). Despite being a cornerstone in various quantum applications, quantum entanglement encounters challenges in establishing connections over extended distances due to probabilistic processes influenced by factors like optical fiber losses. The creation of long-distance entanglement connections between quantum computers involves multiple entanglement links and entanglement swapping techniques through successive quantum nodes, including quantum computers and quantum repeaters, necessitating optimal path selection and qubit allocation. Current research predominantly assumes known success rates of entanglement links between neighboring quantum nodes and overlooks potential network attackers. This paper addresses the online challenge of optimal path selection and qubit allocation, aiming to learn the best strategy for achieving the highest success rate of entanglement connections between two chosen quantum computers without prior knowledge of the success rate and in the presence of a QDN attacker. The proposed approach is based on multi-armed bandits, specifically adversarial group neural bandits, which treat each path as a group and view qubit allocation as arm selection. Our contributions encompass formulating an online adversarial optimization problem, introducing the EXPNeuralUCB bandits algorithm with theoretical performance guarantees, and conducting comprehensive simulations to showcase its superiority over established advanced algorithms.

quant-ph↗

arXiv2024

Learning the Optimal Path and DNN Partition for Collaborative Edge Inference

Recent advancements in Deep Neural Networks (DNNs) have catalyzed the development of numerous intelligent mobile applications and services. However, they also introduce significant computational challenges for resource-constrained mobile devices. To address this, collaborative edge inference has been proposed. This method involves partitioning a DNN inference task into several subtasks and distributing these across multiple network nodes. Despite its potential, most current approaches presume known network parameters -- like node processing speeds and link transmission rates -- or rely on a fixed sequence of nodes for processing the DNN subtasks. In this paper, we tackle a more complex scenario where network parameters are unknown and must be learned, and multiple network paths are available for distributing inference tasks. Specifically, we explore the learning problem of selecting the optimal network path and assigning DNN layers to nodes along this path, considering potential security threats and the costs of switching paths. We begin by deriving structural insights from the DNN layer assignment with complete network information, which narrows down the decision space and provides crucial understanding of optimal assignments. We then cast the learning problem with incomplete network information as a novel adversarial group linear bandits problem with switching costs, featuring rewards generation through a combined stochastic and adversarial process. We introduce a new bandit algorithm, B-EXPUCB, which combines elements of the classical blocked EXP3 and LinUCB algorithms, and demonstrate its sublinear regret. Extensive simulations confirm B-EXPUCB's superior performance in learning for collaborative edge inference over existing algorithms.

cs.LG↗

arXiv2024

Studying the $Ξ(2030)$ as a predominantly $\bar{K}^{*}Σ$ molecular state

Since its discovery in 1977, the spin-parity of $Ξ(2030)$ has not been fully determined experimentally. The latest Particle Data Group (PDG) listing suggests it may be a baryon with $J=5/2$. Therefore, studying the mass spectrum and decay properties of $Ξ(2030)$ has become a current hot topic to definitively establish its spin-parity. As the three-quark model fails to explain $Ξ(2030)$, we previously proposed it may be a molecule primarily composed of $\bar{K}^{}Σ$ with $J^P=5/2^{+}$, based on its mass spectrum study. To verify its molecular state interpretation, this work proposes studying the strong decays of $Ξ(2030)$ assuming it is a $P$-wave $J^P=5/2^{+}$ meson-baryon molecule predominantly composed of $\bar{K}^{}Σ$. We calculated all experimentally measured two-body and three-body final state decay widths of $Ξ(2030)$, including $Ξ(2030) \to \bar{K}Λ, \bar{K}Σ, πΞ, πΞ^{*}$, and $Ξ(2030) \to ππΞ, π\bar{K}Σ, π\bar{K}Λ$. The results indicate that both the total decay width and partial decay widths agree well with experimental values within the error margins. This supports that $Ξ(2030)$ is a molecule with spin-parity $J^P = 5/2^{+}$, predominantly composed of $\bar{K}^{*}Σ$. Compared to the experimental central values, our results are slightly smaller, which suggests that $Ξ(2030)$ may contain additional components besides meson-baryon molecular components, such as three quark structures.

hep-ph↗

arXiv2024

Skin effect in Non-Hermitian systems with spin

The skin effect, where bulk modes collapse into boundary modes, is a key phenomenon in topological non-Hermitian systems, has been predominantly studied in spinless systems. Recent studies illustrate the magnetic suppression of the first-order skin effect while ignoring spin. However, the physical significance of a magnetic field in non-Hermitian skin effect with spin remains elusive. Here, we systematically explore non-Hermitian spinful systems based on generalized Hatano-Nelson models with SU(2) gauge potential fields. In an open one-dimensional lattice, the spin-up and spin-down states can be uniquely separated and localized at the two boundaries without magnetic field. When an external magnetic field is applied, the skin effect exhibits a smooth transition from bidirectional to unidirectional. Remarkably, we demonstrate that the first-order skin effect can be anomalously induced by a magnetic field in a topologically trivial non-Hermitian spinful system without any skin effect at zero field. The direction of such magnetically induced skin modes can be controlled by simply changing the amplitude and polarity of the magnetic field. In addition, we demonstrate a transition between non-Bloch PT and anti-PT symmetries in the system, and uncover the spindependent mechanism of non-Bloch PT symmetry. Our results pave the way for the investigation of non-Hermitian skin effect with spin degrees of freedom.

cond-mat.mes-hall↗