arXiv subjects

Lili He

Publications and source records attributed to Lili He.

14 recordsLinked to original sources

arXiv2026

PCFlow: Physics-Conditioned Flow Matching for GPR B-Scan Image Synthesis

Ground-penetrating radar (GPR) B-scan image synthesis is important for data augmentation, algorithm validation, and simulation acceleration, yet generating radargrams with both visual realism and physical consistency remains challenging. Existing learning-based generative models often emphasize visual appearance but provide limited control over response geometry. In this paper, we propose PCFlow, a physics-conditioned flow matching framework for fast GPR B-scan image synthesis. The core of PCFlow is a Maxwell-informed dense physical condition field constructed from the parameterized physical model used for electromagnetic simulation, including material properties, target geometry, propagation cues, and response-domain priors. This condition field provides an interpretable interface between physical scene parameters and radar response geometry, and guides conditional flow matching in the VAE latent space toward physically feasible generation paths. We evaluate PCFlow on a gprMax-based buried-pipeline dataset with both in-distribution and out-of-distribution test cases. Experimental results show that PCFlow generates images with more accurate response geometry and high visual fidelity, demonstrating its effectiveness for controllable and physically faithful radar image synthesis.

cs.LG↗

arXiv2026

Generalized Whiting transform and Flux Estimates for Wave Equations in Subextremal Kerr

All recent advances \cite{DRS}, \cite{S-Rita1}, \cite{S-Rita2}, \cite{Millet}, \cite{MaS}, \cite{MaS2} on the boundedness and decay for wave equations on subextremal Kerr spacetimes $\KK(a,m)$, with non-small angular rotation, rely in an essential way on Whiting's groundbreaking discovery of a powerful integro-differential transformation with the help of which he ruled out exponentially growing modes in \cite{W} (see also \cite{Yacov}, \cite{AMPW} and \cite{Rita} for further developments). We give a purely \textit{physical space}, generalized version, of the transformation which takes general solutions of {spin-$\sk$} wave equations in $\KK=\KK(a,m)$ to solutions of a secondary wave equation on a new metric manifold $\KKt=\KKt(a,m)$ with two stationary, asymptotically flat, ends. Moreover, for a certain subdomain $\DDt\subset\DDt_-$, which can be identified as the exterior of a black hole region, the new metric $\gt$ of $\KKt(a, m)$ is Lorentzian and the time translation $\T $ is timelike. Using the geometric properties of the two ends one can then deduce, by classical asymptotic Fourier analysis techniques, the desired bound for the flux of $\psi$ at the future event horizon. This estimate played a crucial role in our companion paper \cite{He-K1}.

math.AP↗

arXiv2026

A Physical space derivation of Morawetz-Energy estimates in Kerr spacetimes with large angular momentum

We revisit the derivation of Morawetz--energy estimates for scalar wave equations in the domain of outer communication of a Kerr spacetime $\KK(a,m)$. Our goal is to develop robust physical-space methods which are well suited for extension to realistic perturbations of Kerr. The proof rests on several ingredients. First, we derive conditional Morawetz estimates which extend the physical-space techniques initiated by Andersson and Blue \cite{AB}, and later adapted in \cite{GKS} to perturbations of slowly rotating Kerr, by exploiting a physical-space characterization of the full $r$-range of trapped null geodesics. Second, we use an idea introduced by Stogin \cite{St} in the axially symmetric case to handle the low-frequency difficulties in the Morawetz estimates. In the general case, the control of the lower-order terms also requires making full use of the principal trapping term in the Morawetz bulk norm, together with a new use of Hardy-type inequalities. A further new ingredient is the control of the boundary terms generated by the Morawetz estimates. This is based on two additional ideas: physical-space versions of Whiting's transform \cite{W}, developed in a forthcoming paper \cite{H-K2}, which yield a flux-independent energy estimate; and an adaptation of the Andersson--Blue invariant-operator method, which turns that estimate into a bound for the horizon flux. Finally, a continuity argument yields an unconditional global-in-time Morawetz estimate, while a new energy estimate is obtained from the construction of a causal vectorfield which is Killing on the trapping set. The results proved here are restricted to scalar wave equations, corresponding to spin $0$, in the range $|a|/m\leq 0.75$. We expect this restriction to be technical, and the methods developed in this paper to extend to the Teukolsky equation.

gr-qc↗

arXiv2026

TurboGS: Accelerating 3D Gaussian Splatting via Error-Guided Sparse Pixel Sampling and Optimization

Consumer-level applications require fast optimization of 3D Gaussian Splatting (3DGS) with high-fidelity novel view rendering. However, existing 3DGS acceleration approaches still incur substantial computation on redundant pixels while sacrificing fine details. In this paper, we present TurboGS, an error-guided training framework that accelerates 3DGS by concentrating optimization on perceptually informative pixels. TurboGS is built upon four core components: (1) a tile-wise sparse pixel sampling, which, driven by multi-view reconstruction errors during training, prioritizes challenging regions and skips well-reconstructed ones to avoid redundant gradient computation; (2) a tile-wise structure-aware loss with sparse Normalized Cross-Correlation, which provides sparse yet effective supervision to preserve fine details and stabilize training; (3) an error-driven Gaussian density control strategy, which dynamically allocates model capacity and removes redundant primitives; and (4) a tailored hybrid optimizer that couples Hessian-informed updates with Adam moment damping to stabilize and improve convergence under sparse supervision. Experiments on standard benchmarks demonstrate that TurboGS can deliver on par or superior rendering quality within 100 seconds on a single RTX 5090 GPU card (up to 10x training speedup over vanilla 3DGS).

cs.CV↗

arXiv2026

SCOPE: Tree-based Self-Correcting Online Log Parsing via Syntactic-Semantic Collaboration

Log parsing is a critical step for automated log analysis in complex systems. Traditional heuristic-based methods offer high efficiency but are limited in accuracy due to overlooking semantic context. In contrast, recent LLM-based parsers improve accuracy via se mantic understanding but incur high latency from frequent model calls. To address this, we propose SCOPE, the first self-correcting online log parsing method that integrates the strengths of both heuristic and LLM-based paradigms. SCOPE introduces a novel bi-directional tree structure that enables efficient template match ing from both forward and reverse directions, resulting in a higher overall matching rate. Additionally, it adopts a two-stage syntactic semantic collaboration framework: a lightweight NLP model first utilizes part-of-speech (POS) information for syntax-based match ing, while the LLM is selectively invoked as a fallback to handle semantically complex cases when uncertainty remains. This design significantly reduces LLM API usage while maintaining high ac curacy, achieving a balance between efficiency and effectiveness. Extensive evaluations on diverse benchmark datasets show that SCOPE outperforms state-of-the-art methods in both accuracy and efficiency. The implementation and datasets are publicly released to facilitate further research.

cs.CL↗

arXiv2024

Self-Adaptive Active Damping Method for Stability Enhancement of Systems With Black-Box Inverters Considering Operating Points

Due to the black-box nature of inverters and the wide variation range of operating points, it is challenging to on-line predict and adaptively enhance the stability of inverter-based systems. To solve this problem, this paper provides a feasible self-adaptive active damping method to eliminate potential small-signal instability of systems with black-box inverters under multiple operating points. First, the framework that includes grid impedance estimation, inverters' admittance identification, and self-adaptive strategy is presented. Second, a widely-applicable and engineering-friendly method for inductive-resistive grid impedance estimation is studied, in which a frequency-integral-based dq-axis aligning method is presented to avoid the inaccuracy resulting from the disturbance theta. Then, to make the system have a sufficient stable margin under different operating points, a self-adaptive active damper (SAD) as well as its control strategy with lag compensator modification is proposed, in which the SAD's damping compensation mechanism for the system's stability enhancement is investigated and revealed. Finally, the mapping between system's parameter variations and SAD's parameters is established based on the artificial neural network (ANN) technique, serving as a computationally light model surrogate that is favorable for on-line parameter-tuning for SAD to compensate the system's damping according to operating points. The effectiveness of the proposed method is verified by simulations in PSACD/EMTDC and experiments in RT-Lab platforms.

eess.SY↗

arXiv2024

RadCLIP: Enhancing Radiologic Image Analysis through Contrastive Language-Image Pre-training

The integration of artificial intelligence (AI) with radiology marks a transformative era in medicine. Vision foundation models have been adopted to enhance radiologic imaging analysis. However, the distinct complexities of radiologic 2D and 3D radiologic data pose unique challenges that existing models, pre-trained on general non-medical images, fail to address adequately. To bridge this gap and capitalize on the diagnostic precision required in radiologic imaging, we introduce Radiologic Contrastive Language-Image Pre-training (RadCLIP): a cross-modal vision-language foundational model that harnesses Vision Language Pre-training (VLP) framework to improve radiologic image analysis. Building upon Contrastive Language-Image Pre-training (CLIP), RadCLIP incorporates a slice pooling mechanism tailored for volumetric image analysis and is pre-trained using a large and diverse dataset of radiologic image-text pairs. The RadCLIP was pre-trained to effectively align radiologic images with their corresponding text annotations, creating a robust vision backbone for radiologic images. Extensive experiments demonstrate RadCLIP's superior performance in both uni-modal radiologic image classification and cross-modal image-text matching, highlighting its significant promise for improving diagnostic accuracy and efficiency in clinical settings. Our Key contributions include curating a large dataset with diverse radiologic 2D/3D radiologic image-text pairs, a slice pooling adapter using an attention mechanism for integrating 2D images, and comprehensive evaluations of RadCLIP on various radiologic downstream tasks.

cs.CV↗

arXiv2024

Optical detection of bacterial cells on stainless-steel surface with a low-magnification light microscope

A Rapid and cost-effective method for detecting bacterial cells on surfaces is critical to protect public health from various aspects, including food safety, clinical hygiene, and pharmacy quality. Herein, we first established an optical detection method based on a gold chip coating with 3-mercaptophenylboronic acid (3-MPBA) to capture bacterial cells, which allows for the detection and quantification of bacterial cells with a standard light microscope under low-magnification (10 fold) objective lens. Then, integrating the developed optical detection method with swab sampling to achieve to detect bacterial cells loading on stainless-steel surfaces. Using Salmonella enterica (SE1045) and Escherichia coli as model bacterial cells, we achieved a capture efficiency of up to 76.0 % for SE1045 cells and 81.1 % for E. coli cells at Log 3 CFU/mL upon the optimized conditions. Our assay showed good linear relationship between the concentrations of bacterial cells with the cell counting in images with the limit of detection (LOD) of Log 3 CFU/mL for both SE1045 and E. coli cells. A further increase in sensitivity in detecting E. coli cells was achieved through a heat treatment, enabling the LOD to be pushed as low as Log 2 CFU/mL. Furthermore, successful application was observed in assessing bacterial contamination on stainless-steel surface following integrating with swab collection, achieving a recovery rate of approximately 70 % suggests future prospects for evaluating the cleanliness of surfaces. The entire process was completed within around 2 hours, with a cost of merely 2 dollars per sample. Given a standard light microscope cost around 250 dollars, our developed method has shown great potential in practical industrial applications for bacterial contamination control on surfaces in low-resource settings.

q-bio.QM↗

arXiv2023

Joint Self-Supervised and Supervised Contrastive Learning for Multimodal MRI Data: Towards Predicting Abnormal Neurodevelopment

The integration of different imaging modalities, such as structural, diffusion tensor, and functional magnetic resonance imaging, with deep learning models has yielded promising outcomes in discerning phenotypic characteristics and enhancing disease diagnosis. The development of such a technique hinges on the efficient fusion of heterogeneous multimodal features, which initially reside within distinct representation spaces. Naively fusing the multimodal features does not adequately capture the complementary information and could even produce redundancy. In this work, we present a novel joint self-supervised and supervised contrastive learning method to learn the robust latent feature representation from multimodal MRI data, allowing the projection of heterogeneous features into a shared common space, and thereby amalgamating both complementary and analogous information across various modalities and among similar subjects. We performed a comparative analysis between our proposed method and alternative deep multimodal learning approaches. Through extensive experiments on two independent datasets, the results demonstrated that our method is significantly superior to several other deep multimodal learning methods in predicting abnormal neurodevelopment. Our method has the capability to facilitate computer-aided diagnosis within clinical practice, harnessing the power of multimodal data.

eess.IV↗

arXiv2023

A Novel Collaborative Self-Supervised Learning Method for Radiomic Data

The computer-aided disease diagnosis from radiomic data is important in many medical applications. However, developing such a technique relies on annotating radiological images, which is a time-consuming, labor-intensive, and expensive process. In this work, we present the first novel collaborative self-supervised learning method to solve the challenge of insufficient labeled radiomic data, whose characteristics are different from text and image data. To achieve this, we present two collaborative pretext tasks that explore the latent pathological or biological relationships between regions of interest and the similarity and dissimilarity information between subjects. Our method collaboratively learns the robust latent feature representations from radiomic data in a self-supervised manner to reduce human annotation efforts, which benefits the disease diagnosis. We compared our proposed method with other state-of-the-art self-supervised learning methods on a simulation study and two independent datasets. Extensive experimental results demonstrated that our method outperforms other self-supervised learning methods on both classification and regression tasks. With further refinement, our method shows the potential advantage in automatic disease diagnosis with large-scale unlabeled data available.

eess.IV↗

arXiv2023

The linear stability of weakly charged and slowly rotating Kerr-Newman family of charged black holes

In this paper, we prove the linear stability of weakly charged and slowly rotating Kerr-Newman black holes under coupled gravitational and electromagnetic perturbations. We show that the solutions to the linearized Einstein-Maxwell equations decay at an inverse polynomial rate to a linearized Kerr-Newman solution plus a pure gauge term. This work builds on the framework developed by H\"{a}fner-Hintz-Vasy for the study of the Einstein vacuum equations. We work in the generalized wave map and Lorenz gauge. The proof involves the analysis of the resolvent of the Fourier transformed linearized Einstein-Maxwell operator on asymptotically flat spaces, which relies on recent advances in microlocal analysis and non-elliptic Fredholm theory developed by Vasy. The most delicate part of the proof is the description of the resolvent at low frequencies.

gr-qc↗

arXiv2022

A Novel Ontology-guided Attribute Partitioning Ensemble Learning Model for Early Prediction of Cognitive Deficits using Quantitative Structural MRI in Very Preterm Infants

Structural magnetic resonance imaging studies have shown that brain anatomical abnormalities are associated with cognitive deficits in preterm infants. Brain maturation and geometric features can be used with machine learning models for predicting later neurodevelopmental deficits. However, traditional machine learning models would suffer from a large feature-to-instance ratio (i.e., a large number of features but a small number of instances/samples). Ensemble learning is a paradigm that strategically generates and integrates a library of machine learning classifiers and has been successfully used on a wide variety of predictive modeling problems to boost model performance. Attribute (i.e., feature) bagging method is the most commonly used feature partitioning scheme, which randomly and repeatedly draws feature subsets from the entire feature set. Although attribute bagging method can effectively reduce feature dimensionality to handle the large feature-to-instance ratio, it lacks consideration of domain knowledge and latent relationship among features. In this study, we proposed a novel Ontology-guided Attribute Partitioning (OAP) method to better draw feature subsets by considering the domain-specific relationship among features. With the better partitioned feature subsets, we developed an ensemble learning framework, which is referred to as OAP-Ensemble Learning (OAP-EL). We applied the OAP-EL to predict cognitive deficits at 2 years of age using quantitative brain maturation and geometric features obtained at term equivalent age in very preterm infants. We demonstrated that the proposed OAP-EL approach significantly outperformed the peer ensemble learning and traditional machine learning approaches.

cs.LG↗

arXiv2021

Masses at null infinity for Einstein's equations in harmonic coordinates

In this work we give a complete picture of how to in a direct simple way define the mass at null infinity in harmonic coordinates in three different ways that we show satisfy the Bondi mass loss law. The first and second way involve only the limit of metric (Trautman mass) respectively the null second fundamental forms along asymptotically characteristic surfaces (asymptotic Hawking mass) that only depend on the ADM mass. The last in an original way involves construction of special characteristic coordinates at null infinity (Bondi mass). The results here rely on asymptotics of the metric derived in [24].

gr-qc↗

arXiv2021

Scattering from Infinity of the Maxwell Klein Gordon Equations in Lorenz Gauge

We prove global existence backwards from the scattering data posed at infinity for the Maxwell Klein Gordon equations in Lorenz gauge satisfying the weak null condition. The asymptotics of the solutions to the Maxwell Klein Gordon equations in Lorenz gauge were shown to be wave like at null infinity and homogeneous towards timelike infinity in arXiv:1803.11086 and expressed in terms of radiation fields, and thus our scattering data will be given in the form of radiation fields in the backward problem. We give a refinement of the asymptotics results in arXiv:1803.11086, and then making use of this refinement, we find a global solution which attains the prescribed scattering data at infinity. Our work starts from the approach in [22] and is more delicate since it involves nonlinearities with fewer derivatives. Our result corresponds to "existence of scattering states" in the scattering theory. The method of proof relies on a suitable construction of the approximate solution from the scattering data, a weighted conformal Morawetz energy estimate and a spacetime version of Hardy inequality.

math.AP↗