arXiv subjects

Han Yue

Publications and source records attributed to Han Yue.

11 recordsLinked to original sources

arXiv2026

Breaking optoelectronic SNR limitations via physics-consistent computational diffractive imaging

Ptychography is a powerful lensless imaging technique capable of approaching the diffraction limit, yet its performance is increasingly constrained by non-ideal detection hardware. In photon-limited measurements, weak high-frequency diffraction signals often overlap with spatially heterogeneous detector noise, whereas most reconstruction algorithms still treat the detector as an ideal measurement plane. Here, we introduce detector-informed measurement consistency into ptychographic reconstruction. By calibrating the pixelwise sensor response, the method construct a spatially resolved confidence map and embed it into the iterative amplitude constraint, allowing unreliable detector residuals to be down-weighted while preserving physically meaningful diffraction information. Experiments across transmission, reflection, and weak biological phase imaging show improved diffraction-data quality, an approximately twofold signal-to-noise ratio (SNR) enhancement, and reconstruction approaching the Rayleigh limit with a measured (k)-factor of about 0.65. Compared with previous advanced denoising methods, the proposed framework achieves a better balance between suppressing detector-induced background and preserving structural diffraction information. These results show that detector reliability can be used as an in-loop physical constraint to extend the performance of ptychographic imaging with imperfect sensors.

cs.GR↗

arXiv2026

Pulsar anti-glitches: starquakes driven by magnetism?

In the conventional starquake model of pulsar glitches, it is usually assumed that such events arise from fault slip induced by the self-gravity of compact objects. This inevitably decreases the moment of inertia, producing a glitch with an amplitude of only $\Delta\nu/\nu > 0$. However, an increasing number of anti-glitches ($\Delta\nu/\nu < 0$) have been observed in extremely magnetized pulsars, the magnetars, and this cannot be explained by that framework. In the present study, we hypothesis that magnetic stresses within a compact object can make for elastic deformations that trigger fault slipping, resulting in a ``magnetism-driven starquake'' when the local breaking threshold is exceeded. This process can then either decrease or increase the moment of inertia, naturally generating a glitch or an anti-glitch, respectively. With an order-of-magnitude calculation in this brief report, we present a simple relationship between the magnetic field $B$ and the amplitude $\Delta\nu/\nu$, which is consistent with the observational distribution of existing glitch and anti-glitch data. Further discoveries of glitch/anti-glitch events, alongside more quantitative models of elastic-magnetic stress coupling, would be welcome and could eventually provide clear tests for the hypothesis.

astro-ph.HE↗

arXiv2026

Adaptive Retrieval Strategies for Biomedical Question Answering

Biomedical question answering (QA) encompasses diverse question types, including yes/no, factoid, list, and summary questions, each requiring distinct forms of evidence and reasoning. However, most retrieval-augmented QA systems rely on a unified retrieval pipeline, regardless of the information needs of different question categories. This one-size-fits-all approach may limit the effectiveness of evidence acquisition and downstream answer generation. In this work, we propose an adaptive retrieval framework that selects retrieval and evidence aggregation strategies according to question type. The system combines query understanding, biomedical document retrieval, reranking, knowledge graph augmentation, document clustering, and large language model-based answer generation. For yes/no questions, it focuses on precise evidence retrieval; for factoid and list questions, it emphasizes entity-oriented retrieval and clustering; and for summary questions, it performs broader evidence collection and synthesis. We evaluate the proposed framework on the BioASQ benchmark and demonstrate that adaptive retrieval strategies improve evidence relevance and answer quality across multiple question types. Our results suggest that aligning retrieval mechanisms with question-specific information needs provides an effective direction for enhancing retrieval-augmented biomedical QA systems.

cs.IR↗

arXiv2024

A Physics-Inspired Deep Learning Framework with Polar Coordinate Attention for Ptychographic Imaging

Ptychographic imaging confronts inherent challenges in applying deep learning for phase retrieval from diffraction patterns. Conventional neural architectures, both convolutional neural networks and Transformer-based methods, are optimized for natural images with Euclidean spatial neighborhood-based inductive biases that exhibit geometric mismatch with the concentric coherent patterns characteristic of diffraction data in reciprocal space. In this paper, we present PPN, a physics-inspired deep learning network with Polar Coordinate Attention (PoCA) for ptychographic imaging, that aligns neural inductive biases with diffraction physics through a dual-branch architecture separating local feature extraction from non-local coherence modeling. It consists of a PoCA mechanism that replaces Euclidean spatial priors with physically consistent radial-angular correlations. PPN outperforms existing end-to-end models, with spectral and spatial analysis confirming its greater preservation of high-frequency details. Notably, PPN maintains robust performance compared to iterative methods even at low overlap ratios, making it well suited for high-throughput imaging in real-world acquisition scenarios for samples with consistent structural characteristics.

physics.optics↗

arXiv2024

On the Inflation of KNN-Shapley Value

Shapley value-based data valuation methods, originating from cooperative game theory, quantify the usefulness of each individual sample by considering its contribution to all possible training subsets. Despite their extensive applications, these methods encounter the challenge of value inflation - while samples with negative Shapley values are detrimental, some with positive values can also be harmful. This challenge prompts two fundamental questions: the suitability of zero as a threshold for distinguishing detrimental from beneficial samples and the determination of an appropriate threshold. To address these questions, we focus on KNN-Shapley and propose Calibrated KNN-Shapley (CKNN-Shapley), which calibrates zero as the threshold to distinguish detrimental samples from beneficial ones by mitigating the negative effects of small-sized training subsets. Through extensive experiments, we demonstrate the effectiveness of CKNN-Shapley in alleviating data valuation inflation, detecting detrimental samples, and assessing data quality. We also extend our approach beyond conventional classification settings, applying it to diverse and practical scenarios such as learning with mislabeled data, online learning with stream data, and active learning for label annotation.

cs.LG↗

arXiv2024

Revisit, Extend, and Enhance Hessian-Free Influence Functions

Influence functions serve as crucial tools for assessing sample influence in model interpretation, subset training set selection, noisy label detection, and more. By employing the first-order Taylor extension, influence functions can estimate sample influence without the need for expensive model retraining. However, applying influence functions directly to deep models presents challenges, primarily due to the non-convex nature of the loss function and the large size of model parameters. This difficulty not only makes computing the inverse of the Hessian matrix costly but also renders it non-existent in some cases. Various approaches, including matrix decomposition, have been explored to expedite and approximate the inversion of the Hessian matrix, with the aim of making influence functions applicable to deep models. In this paper, we revisit a specific, albeit naive, yet effective approximation method known as TracIn. This method substitutes the inverse of the Hessian matrix with an identity matrix. We provide deeper insights into why this simple approximation method performs well. Furthermore, we extend its applications beyond measuring model utility to include considerations of fairness and robustness. Finally, we enhance TracIn through an ensemble strategy. To validate its effectiveness, we conduct experiments on synthetic data and extensive evaluations on noisy label detection, sample selection for large language model fine-tuning, and defense against adversarial attacks.

cs.LG↗

arXiv2023

The precursor of GRB211211A: a tide-induced giant quake?

The equilibrium configuration of a solid strange star in the final inspiral phase with another compact object is generally discussed, and the starquake-related issue is revisited, for a special purpose to understand the precursor emission of binary compact star merger events (e.g., that of GRB211211A). As the binary system inspirals inward due to gravitational wave radiation, the ellipticity of the solid strangeon star increases due to the growing tidal field of its compact companion. Elastic energy is hence accumulated during the inspiral stage which might trigger a starquake before the merger when exceeds a critical value. The energy released during such starquakes is calculated and compared to the precursor observation of GRB211211A. The result shows that the energy might be insufficient for binary strangeon-star case unless the entire solid strangeon star shatters, and hence favors a black hole-strangeon star scenario for GRB211211A. The timescale of the precursor as well as the frequency of the observed quasi-periodic-oscillation have also been discussed in the starquake model.

astro-ph.HE↗

arXiv2023

Quakes of Compact Stars

Glitches are commonly observed for pulsars, which are explained by various mechanisms. One hypothesis attributes the glitch effect to the instantaneous moment of inertia change of the whole star caused by a starquake, which is similar to earthquakes caused by fast dislocation occurring on planar faults for the static stress, though the quake-induced dynamics responsible for glitch (superfluid vortex vs. pure starquake) remains still unknown. However, a theoretical model to quantitatively explain the stress loading, types of starquakes, and co-seismic change of moment of inertia is rarely discussed. In this study, we incorporate elastic deformation theories of earthquakes into the starquake problems. We compute the field of stress loading associated with rotation deceleration and determine the optimal type of starquakes at various locations. Two types of pulsar structure models, i.e. neutron and strangeon star models, are included in the computation and their differences are notable. Our calculation shows that the observed glitch amplitude can be explained by the starquakes in the strangeon star model, though the required scaled starquake magnitude is much larger than that occurred on the Earth. We further discuss the possibility to compute the energy budget and other glitch phenomena using the starquake model in the elastic medium framework.

astro-ph.HE↗

arXiv2022

Multi-task Envisioning Transformer-based Autoencoder for Corporate Credit Rating Migration Early Prediction

Corporate credit ratings issued by third-party rating agencies are quantified assessments of a company's creditworthiness. Credit Ratings highly correlate to the likelihood of a company defaulting on its debt obligations. These ratings play critical roles in investment decision-making as one of the key risk factors. They are also central to the regulatory framework such as BASEL II in calculating necessary capital for financial institutions. Being able to predict rating changes will greatly benefit both investors and regulators alike. In this paper, we consider the corporate credit rating migration early prediction problem, which predicts the credit rating of an issuer will be upgraded, unchanged, or downgraded after 12 months based on its latest financial reporting information at the time. We investigate the effectiveness of different standard machine learning algorithms and conclude these models deliver inferior performance. As part of our contribution, we propose a new Multi-task Envisioning Transformer-based Autoencoder (META) model to tackle this challenging problem. META consists of Positional Encoding, Transformer-based Autoencoder, and Multi-task Prediction to learn effective representations for both migration prediction and rating prediction. This enables META to better explore the historical data in the training stage for one-year later prediction. Experimental results show that META outperforms all baseline models.

cs.LG↗

arXiv2022

Label-invariant Augmentation for Semi-Supervised Graph Classification

Recently, contrastiveness-based augmentation surges a new climax in the computer vision domain, where some operations, including rotation, crop, and flip, combined with dedicated algorithms, dramatically increase the model generalization and robustness. Following this trend, some pioneering attempts employ the similar idea to graph data. Nevertheless, unlike images, it is much more difficult to design reasonable augmentations without changing the nature of graphs. Although exciting, the current graph contrastive learning does not achieve as promising performance as visual contrastive learning. We conjecture the current performance of graph contrastive learning might be limited by the violation of the label-invariant augmentation assumption. In light of this, we propose a label-invariant augmentation for graph-structured data to address this challenge. Different from the node/edge modification and subgraph extraction, we conduct the augmentation in the representation space and generate the augmented samples in the most difficult direction while keeping the label of augmented data the same as the original samples. In the semi-supervised scenario, we demonstrate our proposed method outperforms the classical graph neural network based methods and recent graph contrastive learning on eight benchmark graph-structured data, followed by several in-depth experiments to further explore the label-invariant augmentation in several aspects.

cs.CV↗

arXiv2017

FRB 121102: a star quake-induced repeater?

Since its initial discovery, the Fast radio burst (FRB) FRB 121102 has been found to be repeating with millisecond-duration pulses. Very recently, 15 new bursts were detected by the Green Bank Telescope (GBT) during its continous monitoring observations. In this letter, we show that the burst energy distribution has a power law form which is very similar to the Gutenberg-Richter law of earthquakes. In addition, the waiting time of the burst has a Gaussian distribution, which is also a distinctive feature of earthquakes. These findings suggest that the repeating FRB pulses may originate from the starquakes of a pulsar. Noting that the soft gamma-ray repeaters (SGRs) also exhibit such distributions, the FRB could be powered by some mechanism associated with the SGRs, including crustal acitivity of a magnetar, solidification-induced stresss of a young strangeon star, and oscillation-drivern magnetic activities of a slowly rotating pulsar. These conjectures could be tested with more repeating samples.

astro-ph.HE↗