arXiv subjects

Yilin Wang

Publications and source records attributed to Yilin Wang.

At least 91 records · Page 5Linked to original sources

arXiv2025

OmniStyle: Filtering High Quality Style Transfer Data at Scale

In this paper, we introduce OmniStyle-1M, a large-scale paired style transfer dataset comprising over one million content-style-stylized image triplets across 1,000 diverse style categories, each enhanced with textual descriptions and instruction prompts. We show that OmniStyle-1M can not only enable efficient and scalable of style transfer models through supervised training but also facilitate precise control over target stylization. Especially, to ensure the quality of the dataset, we introduce OmniFilter, a comprehensive style transfer quality assessment framework, which filters high-quality triplets based on content preservation, style consistency, and aesthetic appeal. Building upon this foundation, we propose OmniStyle, a framework based on the Diffusion Transformer (DiT) architecture designed for high-quality and efficient style transfer. This framework supports both instruction-guided and image-guided style transfer, generating high resolution outputs with exceptional detail. Extensive qualitative and quantitative evaluations demonstrate OmniStyle's superior performance compared to existing approaches, highlighting its efficiency and versatility. OmniStyle-1M and its accompanying methodologies provide a significant contribution to advancing high-quality style transfer, offering a valuable resource for the research community.

cs.CV↗

arXiv2025

Classifying metal-poor stars with machine learning using nucleosynthesis calculations

We apply the capabilities of machine learning (ML) to discern patterns in order to classify metal-poor stars. To do so, we train an ML model on a bank of nucleosynthesis calculations derived from hydrodynamic simulations for events such as neutron star mergers where the rapid ($r$) neutron capture process can take place. Likewise we consider a bank of calculations from simulations of the slow ($s$) neutron capture process and also consider a few calculations for the intermediate ($i$) neutron capture process. We demonstrate that the ML does well overall in recognizing the $s$ process from the $r$ process, and after training on theoretical calculations ML stellar assignments match conventional labels 87% of the time. We highlight that this method then points to stars that could benefit from additional observational measurements. We also demonstrate that the ML assigns some of the presently considered $i$-process stars to instead be of $r$ or $s$ in origin, but likewise, finds stars currently labeled as $s$ to be potentially more aligned with $i$ enrichment. This first application of ML to classify metal-poor star enrichment using theoretical nucleosynthesis calculations thus reveals the promise, and some challenges, associated with this new data-driven path forward.

nucl-th↗

arXiv2025

Unconventional band splitting of CeSb in the devil's staircase transition

The interplay between magnetism and electronic band structure is a central theme in condensed matter physics. CeSb, with its complex devil's staircase antiferromagnetic transition, offers a unique opportunity to explore this interplay. Using angle-resolved photoemission spectroscopy (ARPES), we investigate the electronic structure evolution across the devil's staircase transition. Upon entering the antiferromagnetic phase, we observe an intriguing band splitting of the electron pocket around the X point. The energy separation between the split bands changes abruptly with temperature, consistent with the characteristics of the first-order phase transition. However, their respective spectral weights behave gradually with temperature. Combined with our density functional theory (DFT) calculations, we suggest that this atypical behavior deviates from conventional magnetically induced band splitting and potentially arises from the intricate modulation of paramagnetic and antiferromagnetic layers within the devil's staircase transition. Our results provide insights into the complex relationship between electronic structure and magnetism in correlated electron systems.

cond-mat.str-el↗

arXiv2025

CreoPep: A Universal Deep Learning Framework for Target-Specific Peptide Design and Optimization

Target-specific peptides, such as conotoxins, exhibit exceptional binding affinity and selectivity toward ion channels and receptors. However, their therapeutic potential remains underutilized due to the limited diversity of natural variants and the labor-intensive nature of traditional optimization strategies. Here, we present CreoPep, a deep learning-based conditional generative framework that integrates masked language modeling with a progressive masking scheme to design high-affinity peptide mutants while uncovering novel structural motifs. CreoPep employs an integrative augmentation pipeline, combining FoldX-based energy screening with temperature-controlled multinomial sampling, to generate structurally and functionally diverse peptides that retain key pharmacological properties. We validate this approach by designing conotoxin inhibitors targeting the $α$7 nicotinic acetylcholine receptor, achieving submicromolar potency in electrophysiological assays. Structural analysis reveals that CreoPep-generated variants engage in both conserved and novel binding modes, including disulfide-deficient forms, thus expanding beyond conventional design paradigms. Overall, CreoPep offers a robust and generalizable platform that bridges computational peptide design with experimental validation, accelerating the discovery of next-generation peptide therapeutics.

q-bio.BM↗

arXiv2025

PTCL: Pseudo-Label Temporal Curriculum Learning for Label-Limited Dynamic Graph

Dynamic node classification is critical for modeling evolving systems like financial transactions and academic collaborations. In such systems, dynamically capturing node information changes is critical for dynamic node classification, which usually requires all labels at every timestamp. However, it is difficult to collect all dynamic labels in real-world scenarios due to high annotation costs and label uncertainty (e.g., ambiguous or delayed labels in fraud detection). In contrast, final timestamp labels are easier to obtain as they rely on complete temporal patterns and are usually maintained as a unique label for each user in many open platforms, without tracking the history data. To bridge this gap, we propose PTCL(Pseudo-label Temporal Curriculum Learning), a pioneering method addressing label-limited dynamic node classification where only final labels are available. PTCL introduces: (1) a temporal decoupling architecture separating the backbone (learning time-aware representations) and decoder (strictly aligned with final labels), which generate pseudo-labels, and (2) a Temporal Curriculum Learning strategy that prioritizes pseudo-labels closer to the final timestamp by assigning them higher weights using an exponentially decaying function. We contribute a new academic dataset (CoOAG), capturing long-range research interest in dynamic graph. Experiments across real-world scenarios demonstrate PTCL's consistent superiority over other methods adapted to this task. Beyond methodology, we propose a unified framework FLiD (Framework for Label-Limited Dynamic Node Classification), consisting of a complete preparation workflow, training pipeline, and evaluation standards, and supporting various models and datasets. The code can be found at https://github.com/3205914485/FLiD.

cs.LG↗

arXiv2025

Leveraging Large Self-Supervised Time-Series Models for Transferable Diagnosis in Cross-Aircraft Type Bleed Air System

Bleed Air System (BAS) is critical for maintaining flight safety and operational efficiency, supporting functions such as cabin pressurization, air conditioning, and engine anti-icing. However, BAS malfunctions, including overpressure, low pressure, and overheating, pose significant risks such as cabin depressurization, equipment failure, or engine damage. Current diagnostic approaches face notable limitations when applied across different aircraft types, particularly for newer models that lack sufficient operational data. To address these challenges, this paper presents a self-supervised learning-based foundation model that enables the transfer of diagnostic knowledge from mature aircraft (e.g., A320, A330) to newer ones (e.g., C919). Leveraging self-supervised pretraining, the model learns universal feature representations from flight signals without requiring labeled data, making it effective in data-scarce scenarios. This model enhances both anomaly detection and baseline signal prediction, thereby improving system reliability. The paper introduces a cross-model dataset, a self-supervised learning framework for BAS diagnostics, and a novel Joint Baseline and Anomaly Detection Loss Function tailored to real-world flight data. These innovations facilitate efficient transfer of diagnostic knowledge across aircraft types, ensuring robust support for early operational stages of new models. Additionally, the paper explores the relationship between model capacity and transferability, providing a foundation for future research on large-scale flight signal models.

eess.SP↗

arXiv2025

MotionDreamer: One-to-Many Motion Synthesis with Localized Generative Masked Transformer

Generative masked transformers have demonstrated remarkable success across various content generation tasks, primarily due to their ability to effectively model large-scale dataset distributions with high consistency. However, in the animation domain, large datasets are not always available. Applying generative masked modeling to generate diverse instances from a single MoCap reference may lead to overfitting, a challenge that remains unexplored. In this work, we present MotionDreamer, a localized masked modeling paradigm designed to learn internal motion patterns from a given motion with arbitrary topology and duration. By embedding the given motion into quantized tokens with a novel distribution regularization method, MotionDreamer constructs a robust and informative codebook for local motion patterns. Moreover, a sliding window local attention is introduced in our masked transformer, enabling the generation of natural yet diverse animations that closely resemble the reference motion patterns. As demonstrated through comprehensive experiments, MotionDreamer outperforms the state-of-the-art methods that are typically GAN or Diffusion-based in both faithfulness and diversity. Thanks to the consistency and robustness of the quantization-based approach, MotionDreamer can also effectively perform downstream tasks such as temporal motion editing, \textcolor{update}{crowd animation}, and beat-aligned dance generation, all using a single reference motion. Visit our project page: https://motiondreamer.github.io/

cs.CV↗

arXiv2025

Driving behavior recognition via self-discovery learning

Autonomous driving systems require a deep understanding of human driving behaviors to achieve higher intelligence and safety.Despite advancements in deep learning, challenges such as long-tail distribution due to scarce samples and confusion from similar behaviors hinder effective driving behavior detection.Existing methods often fail to address sample confusion adequately, as datasets frequently contain ambiguous samples that obscure unique semantic information.

cs.AI↗

arXiv2025

Revealing nanoscale structural phase separation in La$_{3}$Ni$_{2}$O$_{7-δ}$ single crystal via scanning near-field optical microscopy

The discovery of superconductivity in La3Ni2O7-$δ$ under high pressure,with an onset critical temperature around 80 K, has sparked significant interest in the superconducting phases of Ruddlesden-Popper nickelates, Lan+1NinO3n+1. While La4Ni3O10 exhibits nearly 100% superconductivity with Tc~30 K under high pressure, magnetic susceptibility studies on La3Ni2O7-$δ$, however, reveal a more complex picture, indicating either filamentary superconductivity or that approximately 50% of crystal phase becomes superconducting in polycrystalline samples. In this study, we employed scattering-type scanning near-field optical microscopy to visualize nanoscale structural phase separation in La3Ni2O7-$δ$, identifying enhanced optical conductivity with stripes approximately 183 nm wide. These stripes run diagonally with respect to the Ni-O-Ni bond directions in the a-b plane, ruling out the possibility that they arise from impurity phases, like the '1313', '214' or '4310' structures. The dark regions and bright stripes exhibit optical conductivities ~ 22% and 29% of gold's, respectively. Additionally, we find that the bright stripes constitute about 38% of the total field of view, while the remainder consists of dark regions and the transitional region between dark regions and bright stripes. Our results suggest that optical conductivity stripes originate from nanoscale structural phase separation. In contrast, La4Ni3O10 exhibits uniform and higher optical conductivity with no observable evidence of phase separation. Thus, our study represents a pioneering effort to directly image nanoscale phase separation in Lan+1NinO3n+1 nickelates. This observation could provide crucial insights into the factors that limit the superconducting volume fraction of La3Ni2O7-$δ$, highlighting SNOM as a powerful probe for exploring nanoscale low-energy physics in correlated quantum materials.

cond-mat.supr-con↗

arXiv2025

SigStyle: Signature Style Transfer via Personalized Text-to-Image Models

Style transfer enables the seamless integration of artistic styles from a style image into a content image, resulting in visually striking and aesthetically enriched outputs. Despite numerous advances in this field, existing methods did not explicitly focus on the signature style, which represents the distinct and recognizable visual traits of the image such as geometric and structural patterns, color palettes and brush strokes etc. In this paper, we introduce SigStyle, a framework that leverages the semantic priors that embedded in a personalized text-to-image diffusion model to capture the signature style representation. This style capture process is powered by a hypernetwork that efficiently fine-tunes the diffusion model for any given single style image. Style transfer then is conceptualized as the reconstruction process of content image through learned style tokens from the personalized diffusion model. Additionally, to ensure the content consistency throughout the style transfer process, we introduce a time-aware attention swapping technique that incorporates content information from the original image into the early denoising steps of target image generation. Beyond enabling high-quality signature style transfer across a wide range of styles, SigStyle supports multiple interesting applications, such as local style transfer, texture transfer, style fusion and style-guided text-to-image generation. Quantitative and qualitative evaluations demonstrate our approach outperforms existing style transfer methods for recognizing and transferring the signature styles.

cs.GR↗

arXiv2025

Vortex Interference Enables optimal 3D Interferometric Nanoscopy

Super-resolution imaging methods that combine interferometric (z) analysis with single-molecule localization microscopy (iSMLM) have achieved ultra-high 3D precision and contributed to the elucidation of important biological ultrastructures. However, their dependence on imaging multiple phase-shifted output channels necessitates complex instrumentation and operation. To solve this problem, we develop an interferometric super-resolution microscope capable of optimal direct axial nanoscopy, termed VILM (Vortex Interference Localization Microscopy). Using a pair of vortex phase plates with opposite orientation for each dual-opposed objective lenses, the detection point-spread functions (PSFs) adopt a bilobed profile whose rotation encodes the axial position. Thus, direct 3D single-molecule coordinate determination can be achieved with a single output image. By reducing the number of output channels to as few as one and utilizing a simple 50:50 beamsplitter, the imaging system is significantly streamlined, while the optimal iSMLM imaging performance is retained, with axial resolution ~2 times better than the lateral. The capability of VILM is demonstrated by resolving the architecture of microtubules and probing the organization of tyrosine-phosphorylated signalling proteins in integrin-based cell adhesions.

physics.optics↗

arXiv2025

Hardening of Ni-O bond-stretching phonons in LaNiO$_2$

We demonstrate that dynamical electron correlation and fluctuating local magnetic moments are crucial for the phonon spectra of the infinite-layer nickelate superconductor, LaNiO$_2$, using DFT plus dynamical mean-field theory (DFT+DMFT) calculations. We find significant hardening of optical Ni-O bond-stretching phonons when going from non-magnetic to paramagnetic state, and increasing Coulomb interaction will make them even harder. The electron correlation is found to be sensitive to the Ni-O bond-stretching distortions, indicating strong interplay between electron correlation and lattice. We find that the strong local electron correlation will not favor charge orders that couple to the Ni-O bond-stretching phonons, in support of the recent experiment that a $3a_0$ charge order is absent in the infinite-layer nickelates. Our results emphasize that the effects of local magnetic fluctuations should be fully taken into account when describing the lattice dynamics of the infinite-layer nickelates without long-range magnetic orders, and also provide evidence to ascribe the kink observed in the recent angle-resolved photoemission experiment to possible strong electron coupling to the Ni-O bond-stretching phonons.

cond-mat.str-el↗

arXiv2025

Universal Liouville action as a renormalized volume and its gradient flow

The universal Liouville action (also known as the Loewner energy for Jordan curves) is a Kähler potential on the Weil-Petersson universal Teichmüller space, which is identified with the family of Weil-Petersson quasicircles via conformal welding. Our main result shows that, under regularity assumptions, the universal Liouville action equals the renormalized volume of the hyperbolic $3$-manifold bounded by the two Epstein-Poincaré surfaces associated with the quasicircle. We also study the gradient descent flow of the universal Liouville action for the Weil-Petersson metric and show that the flow always converges to the origin (the circle). This provides a bound of the Weil-Petersson distance to the origin by the universal Liouville action.

math.DG↗

arXiv2025

Two optimization problems for the Loewner energy

A Jordan curve on the Riemann sphere can be encoded by its conformal welding, a circle homeomorphism. The Loewner energy measures how far a Jordan curve is away from being a circle, or equivalently, how far its welding homeomorphism is away from being a Möbius transformation. We consider two optimizing problems for the Loewner energy, one under the constraint for the curves to pass through $n$ given points on the Riemann sphere, which is the conformal boundary of hyperbolic $3$-space $\mathbb H^3$; the other under the constraint for $n$ given points on the circle to be welded to another $n$ given points of the circle. The latter problem can be viewed as optimizing positive curves on the boundary of AdS$^3$ space passing through $n$ prescribed points. We observe that the answers to the two problems exhibit interesting symmetries: optimizing the Jordan curve in $\partial_\infty \mathbb H^3$ gives rise to a welding homeomorphism that is the boundary of a pleated plane in AdS$^3$, whereas optimizing the positive curve in $\partial_\infty\!\operatorname{AdS}^3$ gives rise to a Jordan curve that is the boundary of a pleated plane in $\mathbb H^3$.

math.CV↗

arXiv2025

SuperNeRF-GAN: A Universal 3D-Consistent Super-Resolution Framework for Efficient and Enhanced 3D-Aware Image Synthesis

Neural volume rendering techniques, such as NeRF, have revolutionized 3D-aware image synthesis by enabling the generation of images of a single scene or object from various camera poses. However, the high computational cost of NeRF presents challenges for synthesizing high-resolution (HR) images. Most existing methods address this issue by leveraging 2D super-resolution, which compromise 3D-consistency. Other methods propose radiance manifolds or two-stage generation to achieve 3D-consistent HR synthesis, yet they are limited to specific synthesis tasks, reducing their universality. To tackle these challenges, we propose SuperNeRF-GAN, a universal framework for 3D-consistent super-resolution. A key highlight of SuperNeRF-GAN is its seamless integration with NeRF-based 3D-aware image synthesis methods and it can simultaneously enhance the resolution of generated images while preserving 3D-consistency and reducing computational cost. Specifically, given a pre-trained generator capable of producing a NeRF representation such as tri-plane, we first perform volume rendering to obtain a low-resolution image with corresponding depth and normal map. Then, we employ a NeRF Super-Resolution module which learns a network to obtain a high-resolution NeRF. Next, we propose a novel Depth-Guided Rendering process which contains three simple yet effective steps, including the construction of a boundary-correct multi-depth map through depth aggregation, a normal-guided depth super-resolution and a depth-guided NeRF rendering. Experimental results demonstrate the superior efficiency, 3D-consistency, and quality of our approach. Additionally, ablation studies confirm the effectiveness of our proposed components.

cs.CV↗

arXiv2025

Direct mass measurements of neutron-rich zinc and gallium isotopes: an investigation of the formation of the first r-process peak

The prediction of isotopic abundances resulting from the rapid neutron capture process (r-process) requires high-precision mass measurements. Using TITAN's on-line time-of-flight spectrometer, first time mass measurements are performed for $^{83}$Zn and $^{86}$Ga. These measurements reduced uncertainties, and are used to calculate isotopic abundances near the first r-process abundance peak using astrophysical conditions present during a binary neutron star (BNS) merger. Good agreement in abundance across a range of trajectories is found when comparing to several metal-poor stars while also strongly deviating from the solar r-process pattern. These findings point to a high degree of sensitivity to the electron fraction of a BNS merger on the final elemental abundance pattern for certain elements near the first r-process peak while others display universality. We find that small changes in electron fraction can produce distinct abundance patterns that match those of metal-poor stars with different classifications.

nucl-ex↗

arXiv2025

SLAM: Towards Efficient Multilingual Reasoning via Selective Language Alignment

Despite the significant improvements achieved by large language models (LLMs) in English reasoning tasks, these models continue to struggle with multilingual reasoning. Recent studies leverage a full-parameter and two-stage training paradigm to teach models to first understand non-English questions and then reason. However, this method suffers from both substantial computational resource computing and catastrophic forgetting. The fundamental cause is that, with the primary goal of enhancing multilingual comprehension, an excessive number of irrelevant layers and parameters are tuned during the first stage. Given our findings that the representation learning of languages is merely conducted in lower-level layers, we propose an efficient multilingual reasoning alignment approach that precisely identifies and fine-tunes the layers responsible for handling multilingualism. Experimental results show that our method, SLAM, only tunes 6 layers' feed-forward sub-layers including 6.5-8% of all parameters within 7B and 13B LLMs, achieving superior average performance than all strong baselines across 10 languages. Meanwhile, SLAM only involves one training stage, reducing training time by 4.1-11.9 compared to the two-stage method.

cs.CL↗

arXiv2025

Heavy-fermions in frustrated Hund's metal with portions of incipient flat-bands

Flat-bands induced by destructive interference of hoppings in frustrated lattices such as kagome metals, have been extensively studied in recent years. However, such flat-bands usually appear in small portions of Brillouin zone and are away from Fermi level in the non-interacting limit (dubbed as ``incipient"). Whether such incipient flat-band portions can induce very strong electron correlations is an open question. Recently, such incipient flat-bands are found in frustrated kagome CsCr$_3$Sb$_5$ and triangular CrB$_2$, which show moderately heavy-fermion behaviors and unconventional superconductivity. Here, by density functional theory plus dynamical mean-field theory calculations, we show that both compounds are typical Hund's metals. The incipient flat-band portions induce dips in hybridization functions, which further enhance the Kondo-like effect of Hund's metal, resulting in moderately heavy-fermion behaviors. This explains the origin of the heavy-fermion behaviors and pave the way for understanding the unconventional superconductivity in CsCr$_3$Sb$_5$ and CrB$_2$. Our work demonstrates a flexible route for generating $d$-electron heavy-fermion and for inducing unconventional superconductivity based on frustrated Hund's metals with portions of incipient flat-bands. A large family of materials with kagome, triangular, pyrochlore lattices consisting of Cr (d$^4$), Ru (d$^4$), Fe (d$^6$) ions are promising candidates.

cond-mat.str-el↗