arXiv subjects

Hao Du

Publications and source records attributed to Hao Du.

At least 19 recordsLinked to original sources

arXiv2026

Increase of Magnetic Trap Loading Efficiency for $^{39}\mathrm{K}$ Bose--Einstein Condensation Experiments

During the experimental sequence, a large atom number is important for the creation of high-quality Bose--Einstein condensates. We report an experimental increase of about 15% in magnetic-trap (MT) loading efficiency by optimizing the MT loading process. The loading time is much shorter than the lifetime of the atoms in the MT, which means atom loss during loading due to the finite trap lifetime can be neglected. We obtain an almost pure $^{39}\mathrm{K}$ condensate with $1.2\times10^{5}$ atoms after evaporative cooling in our optical trap.

cond-mat.quant-gas↗

arXiv2026

Complete Reduction for Derivatives in a Transcendental Liouvillian Extension

Transcendental Liouvillian extensions are differential fields, in which one can model poly-logarithmic, hyperexponential, and trigonometric functions, logarithmic integrals, and their (nested) rational expressions. For such an extension, we construct, over the subfield of constants, a complement of the subspace of derivatives, and develop an algorithm that decomposes any element of the field into the sum of a derivative and a component lying in the complement. Consequently, an element is a derivative if and only if its complementary component vanishes. Moreover, the algorithm enables us to determine elementary integrability over the extension by computing parametric logarithmic parts, and leads to a reduction-based approach to constructing telescopers for elements in the extension, provided that an a priori order bound is given.

cs.SC↗

arXiv2026

Nonreciprocal impurity scattering as a probe for pairing symmetries in kagome superconductors

The superconducting (SC) pairing symmetry and its link to time-reversal symmetry breaking (TRSB) in the vanadium-based kagome superconductors remain unresolved, with ambiguities stemming from sublattice interference and charge-density-wave (CDW) entanglement with superconductivity. Using two representative SC pairings, i.e., the conventional on-site $s$-wave and the TRSB $d_{x^2-y^2}+id_{xy}$-wave, as a model study, we theoretically show that while single magnetic impurity yield qualitatively identical spectral behavior of local density of states (LDOS) for these two symmetries, two magnetic impurities give rise to distinct LDOS patterns. For the conventional on-site $s$-wave pairing, time-reversal symmetry (TRS) enforces equivalent forward and backward scattering between two impurities across all impurity configurations, leading to near disappearance of a Yu-Shiba-Rusinov (YSR) state pair along the line connecting the two impurities. However, for the TRSB $d_{x^2-y^2}+id_{xy}$-wave pairing, this scattering equivalence holds only for inversion-symmetric impurity configurations, with a pair of YSR disappearance restricted to this case. These distinct spectral features are resolvable in scanning tunneling microscopy (STM) experiments, providing a direct avenue to discriminate TRSB and non-TRSB SC pairing symmetries in kagome superconductors and an alternative method to probe SC nonreciprocity that circumvents the ambiguities of conventional critical current-based techniques.

cond-mat.supr-con↗

arXiv2025

Accelerating ground-state auxiliary-field quantum Monte Carlo simulations by delayed update and block force-bias update

Ground-state auxiliary-field quantum Monte Carlo (AFQMC) methods have become key numerical tools for studying quantum phases and phase transitions in interacting many-fermion systems. Despite the broad applicability, the efficiency of these algorithms is often limited by the bottleneck associated with the {\it local update} of the field configuration. In this work, we propose two novel update schemes, the {\it delayed update} and {\it block force-bias update}, both of which can generally and efficiently accelerate ground-state AFQMC simulations. The {\it delayed update}, with a predetermined delay rank, is an elegantly improved version of the {\it local update}, accelerating the process by replacing multiple vector-vector outer products in the latter with a single matrix-matrix multiplication. The {\it block force-bias update} is a block variant of the conventional force-bias update, which is a highly efficient scheme for dilute systems but suffers from the low acceptance ratio in lattice models. Our modified scheme maintains the high efficiency while offering flexible tuning of the acceptance ratio, controlled by the block size, for any desired fermion filling. We apply these two update schemes to both the standard and spin-orbit coupled two-dimensional Hubbard models, demonstrating their speedup over the {\it local update} with respect to the delay rank and block size. We also explore their efficiencies across varying system sizes and model parameters. Our results identify a speedup of $\sim$$8$ for systems with $\sim$$1600$ lattice sites. Furthermore, we have investigated the broader applications as well as an application diagram of these update schemes to general correlated fermion systems.

cond-mat.str-el↗

arXiv2025

PRING: Rethinking Protein-Protein Interaction Prediction from Pairs to Graphs

Deep learning-based computational methods have achieved promising results in predicting protein-protein interactions (PPIs). However, existing benchmarks predominantly focus on isolated pairwise evaluations, overlooking a model's capability to reconstruct biologically meaningful PPI networks, which is crucial for biology research. To address this gap, we introduce PRING, the first comprehensive benchmark that evaluates protein-protein interaction prediction from a graph-level perspective. PRING curates a high-quality, multi-species PPI network dataset comprising 21,484 proteins and 186,818 interactions, with well-designed strategies to address both data redundancy and leakage. Building on this golden-standard dataset, we establish two complementary evaluation paradigms: (1) topology-oriented tasks, which assess intra and cross-species PPI network construction, and (2) function-oriented tasks, including protein complex pathway prediction, GO module analysis, and essential protein justification. These evaluations not only reflect the model's capability to understand the network topology but also facilitate protein function annotation, biological module detection, and even disease mechanism analysis. Extensive experiments on four representative model categories, consisting of sequence similarity-based, naive sequence-based, protein language model-based, and structure-based approaches, demonstrate that current PPI models have potential limitations in recovering both structural and functional properties of PPI networks, highlighting the gap in supporting real-world biological applications. We believe PRING provides a reliable platform to guide the development of more effective PPI prediction models for the community. The dataset and source code of PRING are available at https://github.com/SophieSarceau/PRING.

cs.LG↗

arXiv2025

Complete Reduction for Derivatives in a Primitive Tower

A complete reduction $ϕ$ for derivatives in a differential field is a linear operator on the field over its constant subfield. The reduction enables us to decompose an element $f$ as the sum of a derivative and the remainder $ϕ(f)$. A direct application of $ϕ$ is that $f$ is in-field integrable if and only if $ϕ(f) = 0.$ In this paper, we present a complete reduction for derivatives in a primitive tower algorithmically. Typical examples for primitive towers are differential fields generated by (poly-)logarithmic functions and logarithmic integrals. Using remainders and residues, we provide a necessary and sufficient condition for an element from a primitive tower to have an elementary integral, and discuss how to construct telescopers for non-D-finite functions in some special primitive towers.

cs.SC↗

arXiv2025

A Unified Reduction for Hypergeometric and q-Hypergeometric Creative Telescoping

We adapt the theory of normal and special polynomials from symbolic integration to the summation setting, and then built up a general framework embracing both the usual shift case and the $q$-shift case. In the context of this general framework, we develop a unified reduction algorithm, and subsequently a creative telescoping algorithm, applicable to both hypergeometric terms and their $q$-analogues. Our algorithms allow to split up the usual shift case and the $q$-shift case only when it is really necessary, and thus instantly reveal the intrinsic differences between these two cases. Computational experiments are also provided.

cs.SC↗

arXiv2025

GRAM-MAMBA: Holistic Feature Alignment for Wireless Perception with Adaptive Low-Rank Compensation

Multi-modal fusion is crucial for Internet of Things (IoT) perception, widely deployed in smart homes, intelligent transport, industrial automation, and healthcare. However, existing systems often face challenges: high model complexity hinders deployment in resource-constrained environments, unidirectional modal alignment neglects inter-modal relationships, and robustness suffers when sensor data is missing. These issues impede efficient and robust multimodal perception in real-world IoT settings. To overcome these limitations, we propose GRAM-MAMBA. This framework utilizes the linear-complexity Mamba model for efficient sensor time-series processing, combined with an optimized GRAM matrix strategy for pairwise alignment among modalities, addressing the shortcomings of traditional single-modality alignment. Inspired by Low-Rank Adaptation (LoRA), we introduce an adaptive low-rank layer compensation strategy to handle missing modalities post-training. This strategy freezes the pre-trained model core and irrelevant adaptive layers, fine-tuning only those related to available modalities and the fusion process. Extensive experiments validate GRAM-MAMBA's effectiveness. On the SPAWC2021 indoor positioning dataset, the pre-trained model shows lower error than baselines; adapting to missing modalities yields a 24.5% performance boost by training less than 0.2% of parameters. On the USC-HAD human activity recognition dataset, it achieves 93.55% F1 and 93.81% Overall Accuracy (OA), outperforming prior work; the update strategy increases F1 by 23% while training less than 0.3% of parameters. These results highlight GRAM-MAMBA's potential for achieving efficient and robust multimodal perception in resource-constrained environments.

cs.CV↗

arXiv2025

Pushing the Limits of Safety: A Technical Report on the ATLAS Challenge 2025

Multimodal Large Language Models (MLLMs) have enabled transformative advancements across diverse applications but remain susceptible to safety threats, especially jailbreak attacks that induce harmful outputs. To systematically evaluate and improve their safety, we organized the Adversarial Testing & Large-model Alignment Safety Grand Challenge (ATLAS) 2025}. This technical report presents findings from the competition, which involved 86 teams testing MLLM vulnerabilities via adversarial image-text attacks in two phases: white-box and black-box evaluations. The competition results highlight ongoing challenges in securing MLLMs and provide valuable guidance for developing stronger defense mechanisms. The challenge establishes new benchmarks for MLLM safety evaluation and lays groundwork for advancing safer multimodal AI systems. The code and data for this challenge are openly available at https://github.com/NY1024/ATLAS_Challenge_2025.

cs.CR↗

arXiv2025

Can Differentially Private Fine-tuning LLMs Protect Against Privacy Attacks?

Fine-tuning large language models (LLMs) has become an essential strategy for adapting them to specialized tasks; however, this process introduces significant privacy challenges, as sensitive training data may be inadvertently memorized and exposed. Although differential privacy (DP) offers strong theoretical guarantees against such leakage, its empirical privacy effectiveness on LLMs remains unclear, especially under different fine-tuning methods. In this paper, we systematically investigate the impact of DP across fine-tuning methods and privacy budgets, using both data extraction and membership inference attacks to assess empirical privacy risks. Our main findings are as follows: (1) Differential privacy reduces model utility, but its impact varies significantly across different fine-tuning methods. (2) Without DP, the privacy risks of models fine-tuned with different approaches differ considerably. (3) When DP is applied, even a relatively high privacy budget can substantially lower privacy risk. (4) The privacy-utility trade-off under DP training differs greatly among fine-tuning methods, with some methods being unsuitable for DP due to severe utility degradation. Our results provide practical guidance for privacy-conscious deployment of LLMs and pave the way for future research on optimizing the privacy-utility trade-off in fine-tuning methodologies.

cs.CR↗

arXiv2025

SVLTA: Benchmarking Vision-Language Temporal Alignment via Synthetic Video Situation

Vision-language temporal alignment is a crucial capability for human dynamic recognition and cognition in real-world scenarios. While existing research focuses on capturing vision-language relevance, it faces limitations due to biased temporal distributions, imprecise annotations, and insufficient compositionally. To achieve fair evaluation and comprehensive exploration, our objective is to investigate and evaluate the ability of models to achieve alignment from a temporal perspective, specifically focusing on their capacity to synchronize visual scenarios with linguistic context in a temporally coherent manner. As a preliminary step, we present the statistical analysis of existing benchmarks and reveal the existing challenges from a decomposed perspective. To this end, we introduce SVLTA, the Synthetic Vision-Language Temporal Alignment derived via a well-designed and feasible control generation method within a simulation environment. The approach considers commonsense knowledge, manipulable action, and constrained filtering, which generates reasonable, diverse, and balanced data distributions for diagnostic evaluations. Our experiments reveal diagnostic insights through the evaluations in temporal question answering, distributional shift sensitiveness, and temporal alignment adaptation.

cs.CV↗

arXiv2025

Privacy in Fine-tuning Large Language Models: Attacks, Defenses, and Future Directions

Fine-tuning has emerged as a critical process in leveraging Large Language Models (LLMs) for specific downstream tasks, enabling these models to achieve state-of-the-art performance across various domains. However, the fine-tuning process often involves sensitive datasets, introducing privacy risks that exploit the unique characteristics of this stage. In this paper, we provide a comprehensive survey of privacy challenges associated with fine-tuning LLMs, highlighting vulnerabilities to various privacy attacks, including membership inference, data extraction, and backdoor attacks. We further review defense mechanisms designed to mitigate privacy risks in the fine-tuning phase, such as differential privacy, federated learning, and knowledge unlearning, discussing their effectiveness and limitations in addressing privacy risks and maintaining model utility. By identifying key gaps in existing research, we highlight challenges and propose directions to advance the development of privacy-preserving methods for fine-tuning LLMs, promoting their responsible use in diverse applications.

cs.AI↗

arXiv2025

Reduction systems and degree bounds for integration

In symbolic integration, the Risch--Norman algorithm aims to find closed forms of elementary integrals over differential fields by an ansatz for the integral, which usually is based on heuristic degree bounds. Norman presented an approach that avoids degree bounds and only relies on the completion of reduction systems. We give a formalization of his approach and we develop a refined completion process, which terminates in more instances. In some situations when the completion process does not terminate, one can detect patterns allowing to still describe infinite reduction systems that are complete. We present such infinite systems for the fields generated by Airy functions and complete elliptic integrals, respectively. Moreover, we show how complete reduction systems can be used to find rigorous degree bounds. In particular, we give a general formula for weighted degree bounds and we apply it to find tight bounds in the above examples.

cs.SC↗

arXiv2025

Tumor Detection, Segmentation and Classification Challenge on Automated 3D Breast Ultrasound: The TDSC-ABUS Challenge

Breast cancer is one of the most common causes of death among women worldwide. Early detection helps in reducing the number of deaths. Automated 3D Breast Ultrasound (ABUS) is a newer approach for breast screening, which has many advantages over handheld mammography such as safety, speed, and higher detection rate of breast cancer. Tumor detection, segmentation, and classification are key components in the analysis of medical images, especially challenging in the context of 3D ABUS due to the significant variability in tumor size and shape, unclear tumor boundaries, and a low signal-to-noise ratio. The lack of publicly accessible, well-labeled ABUS datasets further hinders the advancement of systems for breast tumor analysis. Addressing this gap, we have organized the inaugural Tumor Detection, Segmentation, and Classification Challenge on Automated 3D Breast Ultrasound 2023 (TDSC-ABUS2023). This initiative aims to spearhead research in this field and create a definitive benchmark for tasks associated with 3D ABUS image analysis. In this paper, we summarize the top-performing algorithms from the challenge and provide critical analysis for ABUS image examination. We offer the TDSC-ABUS challenge as an open-access platform at https://tdsc-abus2023.grand-challenge.org/ to benchmark and inspire future developments in algorithmic research.

eess.IV↗

arXiv2024

PGB: Benchmarking Differentially Private Synthetic Graph Generation Algorithms

Differentially private graph analysis is a powerful tool for deriving insights from diverse graph data while protecting individual information. Designing private analytic algorithms for different graph queries often requires starting from scratch. In contrast, differentially private synthetic graph generation offers a general paradigm that supports one-time generation for multiple queries. Although a rich set of differentially private graph generation algorithms has been proposed, comparing them effectively remains challenging due to various factors, including differing privacy definitions, diverse graph datasets, varied privacy requirements, and multiple utility metrics. To this end, we propose PGB (Private Graph Benchmark), a comprehensive benchmark designed to enable researchers to compare differentially private graph generation algorithms fairly. We begin by identifying four essential elements of existing works as a 4-tuple: mechanisms, graph datasets, privacy requirements, and utility metrics. We discuss principles regarding these elements to ensure the comprehensiveness of a benchmark. Next, we present a benchmark instantiation that adheres to all principles, establishing a new method to evaluate existing and newly proposed graph generation algorithms. Through extensive theoretical and empirical analysis, we gain valuable insights into the strengths and weaknesses of prior algorithms. Our results indicate that there is no universal solution for all possible cases. Finally, we provide guidelines to help researchers select appropriate mechanisms for various scenarios.

cs.DB↗

arXiv2024

Preserving Tumor Volumes for Unsupervised Medical Image Registration

Medical image registration is a critical task that estimates the spatial correspondence between pairs of images. However, current traditional and deep-learning-based methods rely on similarity measures to generate a deforming field, which often results in disproportionate volume changes in dissimilar regions, especially in tumor regions. These changes can significantly alter the tumor size and underlying anatomy, which limits the practical use of image registration in clinical diagnosis. To address this issue, we have formulated image registration with tumors as a constraint problem that preserves tumor volumes while maximizing image similarity in other normal regions. Our proposed strategy involves a two-stage process. In the first stage, we use similarity-based registration to identify potential tumor regions by their volume change, generating a soft tumor mask accordingly. In the second stage, we propose a volume-preserving registration with a novel adaptive volume-preserving loss that penalizes the change in size adaptively based on the masks calculated from the previous stage. Our approach balances image similarity and volume preservation in different regions, i.e., normal and tumor regions, by using soft tumor masks to adjust the imposition of volume-preserving loss on each one. This ensures that the tumor volume is preserved during the registration process. We have evaluated our strategy on various datasets and network architectures, demonstrating that our method successfully preserves the tumor volume while achieving comparable registration results with state-of-the-art methods. Our codes is available at: \url{https://dddraxxx.github.io/Volume-Preserving-Registration/}.

eess.IV↗

arXiv2023

Reducing Hyperexponential Functions over Monomial Extensions

We extend the shell and kernel reductions for hyperexponential functions over the field of rational functions to a monomial extension. Both of the reductions are incorporated into one algorithm. As an application, we present an additive decomposition in rationally hyperexponential towers. The decomposition yields an alternative algorithm for computing elementary integrals over such towers. The alternative can find some elementary integrals that are unevaluated by the integrators in the latest versions of Maple and Mathematica.

cs.SC↗

arXiv2023

Weakly-Supervised 3D Medical Image Segmentation using Geometric Prior and Contrastive Similarity

Medical image segmentation is almost the most important pre-processing procedure in computer-aided diagnosis but is also a very challenging task due to the complex shapes of segments and various artifacts caused by medical imaging, (i.e., low-contrast tissues, and non-homogenous textures). In this paper, we propose a simple yet effective segmentation framework that incorporates the geometric prior and contrastive similarity into the weakly-supervised segmentation framework in a loss-based fashion. The proposed geometric prior built on point cloud provides meticulous geometry to the weakly-supervised segmentation proposal, which serves as better supervision than the inherent property of the bounding-box annotation (i.e., height and width). Furthermore, we propose contrastive similarity to encourage organ pixels to gather around in the contrastive embedding space, which helps better distinguish low-contrast tissues. The proposed contrastive embedding space can make up for the poor representation of the conventionally-used gray space. Extensive experiments are conducted to verify the effectiveness and the robustness of the proposed weakly-supervised segmentation framework. The proposed framework is superior to state-of-the-art weakly-supervised methods on the following publicly accessible datasets: LiTS 2017 Challenge, KiTS 2021 Challenge, and LPBA40. We also dissect our method and evaluate the performance of each component.

eess.IV↗