arXiv subjects

Xiaofeng Chen

Publications and source records attributed to Xiaofeng Chen.

At least 19 recordsLinked to original sources

arXiv2026

Multi-Source Multi-View Graph Domain Adaptation with Hyperbolic Residual Encoding for Cross-Site MDD Identification from rs-fMRI

Cross-site identification of major depressive disorder (MDD) from resting-state functional magnetic resonance imaging (rs-fMRI) is hindered by inter-site distribution shifts and heterogeneous functional connectivity (FC) views. These views capture complementary neural relationships but exhibit distinct site biases and graph topologies, complicating alignment without sacrificing disease-relevant information or cross-view consistency. Existing studies largely treat multi-view connectome learning and cross-site adaptation separately. To the best of our knowledge, few studies have jointly modeled multiple FC views under multi-source unsupervised domain adaptation for cross-site rs-fMRI-based MDD classification. We construct Pearson correlation, sparse representation, and Granger causality graphs, each encoded by a view-specific graph attention network. Dual-stream adaptive fusion explicitly integrates pairwise cross-view interactions, followed by lightweight hyperbolic residual encoding for curvature-aware representation refinement. Class-wise Cauchy--Schwarz alignment reduces inter-source and source-target discrepancies, complemented by adversarial learning, information maximization, and confidence-aware pseudo-labeling. Across seven unlabeled target domains, our framework achieves 73.60% mean accuracy and 71.90% AUC, demonstrating effective generalization under heterogeneous acquisition conditions. These results highlight the effectiveness of unified heterogeneous-view modeling, curvature-aware refinement, and multi-source domain adaptation for cross-site MDD identification.The source code is at https://github.com/OPUS-Lightphenexx/MM-HyperGDA

cs.CV↗

arXiv2026

EchoFlow: A Workload-Aware Parameter Tuning Method for Blockchain Systems

Blockchain systems expose a large number of tunable parameters that significantly influence system performance. However, in practice, a single parameter configuration is often applied across different workloads, leaving substantial unexploited performance potential. To address this, we propose EchoFlow, a blockchain parameter tuning framework that adaptively adjusts parameter configurations based on workload characteristics, enabling continuous performance optimization. EchoFlow employs a distributed reinforcement learning approach in which multiple actors perform parallel sampling to mitigate the substantial time required for sample generation in blockchain environments. To further accelerate convergence, we introduce a genetic algorithm during the initial phase of training to generate high-quality samples. Extensive experimental evaluations demonstrate that EchoFlow consistently outperforms existing methods across diverse workload scenarios while also reducing training time, highlighting its effectiveness and practical value.

cs.DC↗

arXiv2026

CATA: Continual Machine Unlearning via Conflict-Averse Task Arithmetic

Vision-language models (VLMs) have shown remarkable ability in aligning visual and textual representations, enabling a wide range of multimodal applications. However, their large-scale training data inevitably raises concerns about privacy, copyright, and undesirable content, creating a strong need for machine unlearning. While existing studies mainly focus on single-shot unlearning, practical VLM deployment often involves sequential removal requests over time, giving rise to continual machine unlearning. In this work, we make the first attempt to study continual unlearning for VLMs and identify three key challenges in this setting: effectiveness in removing target knowledge, fidelity in preserving retained model utility, and persistence in preventing knowledge re-emergence under sequential updates. To address these challenges, we propose CATA, a conflict-averse task arithmetic method that represents each forget request as an unlearning task vector. By maintaining historical task vectors and performing sign-aware conflict-averse aggregation, CATA suppresses conflicting update components that may weaken previous forgetting effects. Extensive experiments under both single-shot and continual settings show that CATA outperforms baselines in terms of forgetting effectiveness, model fidelity, and forgetting persistence.

cs.CV↗

arXiv2026

Depth Charge: Jailbreak Large Language Models from Deep Safety Attention Heads

Currently, open-sourced large language models (OSLLMs) have demonstrated remarkable generative performance. However, as their structure and weights are made public, they are exposed to jailbreak attacks even after alignment. Existing attacks operate primarily at shallow levels, such as the prompt or embedding level, and often fail to expose vulnerabilities rooted in deeper model components, which creates a false sense of security for successful defense. In this paper, we propose \textbf{\underline{S}}afety \textbf{\underline{A}}ttention \textbf{\underline{H}}ead \textbf{\underline{A}}ttack (\textbf{SAHA}), an attention-head-level jailbreak framework that explores the vulnerability in deeper but insufficiently aligned attention heads. SAHA contains two novel designs. Firstly, we reveal that deeper attention layers introduce more vulnerability against jailbreak attacks. Based on this finding, \textbf{SAHA} introduces \textit{Ablation-Impact Ranking} head selection strategy to effectively locate the most vital layer for unsafe output. Secondly, we introduce a boundary-aware perturbation method, \textit{i.e. Layer-Wise Perturbation}, to probe the generation of unsafe content with minimal perturbation to the attention. This constrained perturbation guarantees higher semantic relevance with the target intent while ensuring evasion. Extensive experiments show the superiority of our method: SAHA improves ASR by 14\% over SOTA baselines, revealing the vulnerability of the attack surface on the attention head. Our code is available at https://anonymous.4open.science/r/SAHA.

cs.CR↗

arXiv2026

Knowing without Acting: The Disentangled Geometry of Safety Mechanisms in Large Language Models

Safety alignment is often conceptualized as a monolithic process wherein harmfulness detection automatically triggers refusal. However, the persistence of jailbreak attacks suggests a fundamental mechanistic decoupling. We propose the \textbf{\underline{D}}isentangled \textbf{\underline{S}}afety \textbf{\underline{H}}ypothesis \textbf{(DSH)}, positing that safety computation operates on two distinct subspaces: a \textit{Recognition Axis} ($\mathbf{v}_H$, ``Knowing'') and an \textit{Execution Axis} ($\mathbf{v}_R$, ``Acting''). Our geometric analysis reveals a universal ``Reflex-to-Dissociation'' evolution, where these signals transition from antagonistic entanglement in early layers to structural independence in deep layers. To validate this, we introduce \textit{Double-Difference Extraction} and \textit{Adaptive Causal Steering}. Using our curated \textsc{AmbiguityBench}, we demonstrate a causal double dissociation, effectively creating a state of ``Knowing without Acting.'' Crucially, we leverage this disentanglement to propose the \textbf{Refusal Erasure Attack (REA)}, which achieves State-of-the-Art attack success rates by surgically lobotomizing the refusal mechanism. Furthermore, we uncover a critical architectural divergence, contrasting the \textit{Explicit Semantic Control} of Llama3.1 with the \textit{Latent Distributed Control} of Qwen2.5. The code and dataset are available at https://anonymous.4open.science/r/DSH.

cs.CR↗

arXiv2026

Partial domain adaptation enables cross domain cell type annotation between scRNA-seq and snRNA-seq

Accurate cell type annotation across datasets is a key challenge in single-cell analysis. snRNA-seq enables profiling of frozen or difficult-to-dissociate tissues, complementing scRNA-seq by capturing fragile or rare cell types. However, cross-annotation between these two datasets remains largely unexplored, as existing methods treat them independently. We introduce ScNucAdapt, a method designed for cross-annotation between paired and unpaired scRNA-seq and snRNA-seq datasets. To address distributional and cell composition differences, ScNucAdapt employs partial domain adaptation. Experiments across both unpaired and paired scRNA-seq and snRNA-seq show that ScNucAdapt achieves robust and accurate cell type annotation, outperforming existing approaches. Therefore, ScNucAdapt provides a practical framework for the cross-domain cell type annotation between scRNA-seq and snRNA seq data.

q-bio.GN↗

arXiv2026

Single-exposure holographic 3D printing via inverse-designed phase masks

Additive manufacturing using light is commonly constrained by serial voxel-by-voxel or layer-by-layer processing, which fundamentally limits fabrication speed and scalability. Here, we introduce a single-exposure holographic three-dimensional (3D) printing approach that synthesizes an entire volumetric dose distribution optically in one step. The method combines inverse-designed microstructured phase masks with photopolymer resins engineered for controlled optical absorption. By precisely tailoring the phase-mask topography, we generate arbitrary 3D light-intensity distributions within the resin, including intentionally encoded dark regions that define hollow internal features. Simultaneously, the resin formulation is designed to balance optical penetration with sufficient local energy deposition to achieve high-fidelity polymerization throughout the volume. Using this approach, millimeter-scale architectures comprising more than $10^{6}$ addressable voxels are fabricated in a single 7.5~s exposure, corresponding to a volumetric throughput of $\sim$1~mm$^{3}$/s ($>10^{5}$~voxels/s). The demonstrated performance is presently limited by resin kinetics and illumination geometry rather than by the phase-mask framework itself. Because the volumetric information capacity scales with the space--bandwidth product of the phase mask, this approach provides a clear pathway toward substantially higher throughput, enabling scalable fabrication of micro-optical components, biomedical scaffolds, and other precision-engineered mesoscale systems.

physics.optics↗

arXiv2025

ShiftKD: Benchmarking Knowledge Distillation under Distribution Shift

Knowledge Distillation (KD) transfers knowledge from large models to small models and has recently achieved remarkable success. However, the reliability of existing KD methods in real-world applications, especially under distribution shift, remains underexplored. Distribution shift refers to the data distribution drifts between the training and testing phases, and this can adversely affect the efficacy of KD. In this paper, we propose a unified and systematic framework \textsc{ShiftKD} to benchmark KD against two general distributional shifts: diversity and correlation shift. The evaluation benchmark covers more than 30 methods from algorithmic, data-driven, and optimization perspectives for five benchmark datasets. Our development of \textsc{ShiftKD} conducts extensive experiments and reveals strengths and limitations of current SOTA KD methods. More importantly, we thoroughly analyze key factors in student model training process, including data augmentation, pruning methods, optimizers, and evaluation metrics. We believe \textsc{ShiftKD} could serve as an effective benchmark for assessing KD in real-world scenarios, thus driving the development of more robust KD methods in response to evolving demands. The code will be made available upon publication.

cs.LG↗

arXiv2025

Timestamp calibration for time-series single cell RNA-seq expression data

Timestamp automatic annotation (TAA) is a crucial procedure for analyzing time-series ScRNA-seq data, as they unveil dynamic biological developments and cell regeneration process. However, current TAA methods heavily rely on manual timestamps, often overlooking their reliability. This oversight can significantly degrade the performance of timestamp automatic annotation due to noisy timestamps. Nevertheless, the current approach for addressing this issue tends to select less critical cleaned samples for timestamp calibration. To tackle this challenge, we have developed a novel timestamp calibration model called ScPace for handling noisy labeled time-series ScRNA-seq data. This approach incorporates a latent variable indicator within a base classifier instead of probability sampling to detect noisy samples effectively. To validate our proposed method, we conducted experiments on both simulated and real time-series ScRNA-seq datasets. Cross-validation experiments with different artificial mislabeling rates demonstrate that ScPace outperforms previous approaches. Furthermore, after calibrating the timestamps of the original time-series ScRNA-seq data using our method, we performed supervised pseudotime analysis, revealing that ScPace enhances its performance significantly. These findings suggest that ScPace is an effective tool for timestamp calibration by enabling reclassification and deletion of detected noisy labeled samples while maintaining robustness across diverse ranges of time-series ScRNA-seq datasets. The source code is available at https://github.com/OPUS-Lightphenexx/ScPace.

q-bio.GN↗

arXiv2024

GATES: Graph Attention Network with Global Expression Fusion for Deciphering Spatial Transcriptome Architectures

Single-cell spatial transcriptomics (ST) offers a unique approach to measuring gene expression profiles and spatial cell locations simultaneously. However, most existing ST methods assume that cells in closer spatial proximity exhibit more similar gene expression patterns. Such assumption typically results in graph structures that prioritize local spatial information while overlooking global patterns, limiting the ability to fully capture the broader structural features of biological tissues. To overcome this limitation, we propose GATES (Graph Attention neTwork with global Expression fuSion), a novel model designed to capture structural details in spatial transcriptomics data. GATES first constructs an expression graph that integrates local and global information by leveraging both spatial proximity and gene expression similarity. The model then employs an autoencoder with adaptive attention to assign proper weights for neighboring nodes, enhancing its capability of feature extraction. By fusing features of both the spatial and expression graphs, GATES effectively balances spatial context with gene expression data. Experimental results across multiple datasets demonstrate that GATES significantly outperforms existing methods in identifying spatial domains, highlighting its potential for analyzing complex biological tissues. Our code can be accessed on GitHub at https://github.com/xiaoxiongtao/GATES.

q-bio.GN↗

arXiv2024

Mixture Data for Training Cannot Ensure Out-of-distribution Generalization

Deep neural networks often face generalization problems to handle out-of-distribution (OOD) data, and there remains a notable theoretical gap between the contributing factors and their respective impacts. Literature evidence from in-distribution data has suggested that generalization error can shrink if the size of mixture data for training increases. However, when it comes to OOD samples, this conventional understanding does not hold anymore -- Increasing the size of training data does not always lead to a reduction in the test generalization error. In fact, diverse trends of the errors have been found across various shifting scenarios including those decreasing trends under a power-law pattern, initial declines followed by increases, or continuous stable patterns. Previous work has approached OOD data qualitatively, treating them merely as samples unseen during training, which are hard to explain the complicated non-monotonic trends. In this work, we quantitatively redefine OOD data as those situated outside the convex hull of mixed training data and establish novel generalization error bounds to comprehend the counterintuitive observations better. Our proof of the new risk bound agrees that the efficacy of well-trained models can be guaranteed for unseen data within the convex hull; More interestingly, but for OOD data beyond this coverage, the generalization cannot be ensured, which aligns with our observations. Furthermore, we attempted various OOD techniques to underscore that our results not only explain insightful observations in recent OOD generalization work, such as the significance of diverse data and the sensitivity to unseen shifts of existing algorithms, but it also inspires a novel and effective data selection strategy.

cs.LG↗

arXiv2024

SAME: Sample Reconstruction against Model Extraction Attacks

While deep learning models have shown significant performance across various domains, their deployment needs extensive resources and advanced computing infrastructure. As a solution, Machine Learning as a Service (MLaaS) has emerged, lowering the barriers for users to release or productize their deep learning models. However, previous studies have highlighted potential privacy and security concerns associated with MLaaS, and one primary threat is model extraction attacks. To address this, there are many defense solutions but they suffer from unrealistic assumptions and generalization issues, making them less practical for reliable protection. Driven by these limitations, we introduce a novel defense mechanism, SAME, based on the concept of sample reconstruction. This strategy imposes minimal prerequisites on the defender's capabilities, eliminating the need for auxiliary Out-of-Distribution (OOD) datasets, user query history, white-box model access, and additional intervention during model training. It is compatible with existing active defense methods. Our extensive experiments corroborate the superior efficacy of SAME over state-of-the-art solutions. Our code is available at https://github.com/xythink/SAME.

cs.CR↗

arXiv2022

Deep Hierarchy Quantization Compression algorithm based on Dynamic Sampling

Unlike traditional distributed machine learning, federated learning stores data locally for training and then aggregates the models on the server, which solves the data security problem that may arise in traditional distributed machine learning. However, during the training process, the transmission of model parameters can impose a significant load on the network bandwidth. It has been pointed out that the vast majority of model parameters are redundant during model parameter transmission. In this paper, we explore the data distribution law of selected partial model parameters on this basis, and propose a deep hierarchical quantization compression algorithm, which further compresses the model and reduces the network load brought by data transmission through the hierarchical quantization of model parameters. And we adopt a dynamic sampling strategy for the selection of clients to accelerate the convergence of the model. Experimental results on different public datasets demonstrate the effectiveness of our algorithm.

cs.LG↗

arXiv2022

CDEdit: A Highly Applicable Redactable Blockchain with Controllable Editing Privilege and Diversified Editing Types

Redactable blockchains allow modifiers or voting committees with modification privileges to edit the data on the chain. Trapdoor holders in chameleon-based hash redactable blockchains can quickly compute hash collisions for arbitrary data, and without breaking the link of the hash-chain. However, chameleon-based hash redactable blockchain schemes have difficulty solving the problem of multi-level editing requests and competing for modification privileges. In this paper, we propose CDEdit, a highly applicable redactable blockchain with controllable editing privilege and diversified editing types. The proposed scheme increases the cost of invalid or malicious requests by paying the deposit on each edit request. At the same time, the editing privilege is subdivided into request, modification, and verification privileges, and the modification privilege token is distributed efficiently to prevent the abuse of the modification privilege and collusion attacks. We use chameleon hashes with ephemeral trapdoor (CHET) and ciphertext policy attribute-based encryption (CP-ABE) to implement two editing types of transaction-level and block-level, and present a practical instantiation and security analysis. Finally, the implementation and evaluation show that our scheme only costs low-performance overhead and is suitable for multi-level editing requests and modification privilege competition scenarios.

cs.CR↗

arXiv2021

Privacy Inference Attacks and Defenses in Cloud-based Deep Neural Network: A Survey

Deep Neural Network (DNN), one of the most powerful machine learning algorithms, is increasingly leveraged to overcome the bottleneck of effectively exploring and analyzing massive data to boost advanced scientific development. It is not a surprise that cloud computing providers offer the cloud-based DNN as an out-of-the-box service. Though there are some benefits from the cloud-based DNN, the interaction mechanism among two or multiple entities in the cloud inevitably induces new privacy risks. This survey presents the most recent findings of privacy attacks and defenses appeared in cloud-based neural network services. We systematically and thoroughly review privacy attacks and defenses in the pipeline of cloud-based DNN service, i.e., data manipulation, training, and prediction. In particular, a new theory, called cloud-based ML privacy game, is extracted from the recently published literature to provide a deep understanding of state-of-the-art research. Finally, the challenges and future work are presented to help researchers to continue to push forward the competitions between privacy attackers and defenders.

cs.CR↗

arXiv2021

An Efficient Matrix Multiplication with Enhanced Privacy Protection in Cloud Computing and Its Applications

As one of the most important basic operations, matrix multiplication computation (MMC) has varieties of applications in the scientific and engineering community such as linear regression, k-nearest neighbor classification and biometric identification. However handling these tasks with large-scale datasets will lead to huge computation beyond resource-constrained client s computation power. With the rapid development of cloud computing, outsourcing intensive tasks to cloud server has become a promising method. While the cloud server is generally out of the control of clients, there are still many challenges concerned with the privacy security of clients sensitive data. Motivated by this, Lei et al. presented an efficient encryption scheme based on random permutation to protect the privacy of client s data in outsourcing MMC task. Nevertheless, there exists inherent security flaws in their scheme, revealing the statistic information of zero elements in the original data thus not satisfying the computational indistinguishability (IND-ZEA). Aiming to enhance the security of the outsourcing MMC task, we propose a new encryption scheme based on subtly designed invertible matrix where the additive perturbation is introduced besides the multiplicative perturbation. Furthermore, we show that the proposed encryption scheme can be applied to not only MMC task but also other kinds of outsourced tasks such as linear regression and principal component analysis. Theoretical analyses and experiments indicate that our methods are more secure in terms of data privacy, with comparable performance to the state-of-the-art scheme based on matrix transformation.

cs.CR↗

arXiv2020

An Efficient Secure Dynamic Skyline Query Model

It is now cost-effective to outsource large dataset and perform query over the cloud. However, in this scenario, there exist serious security and privacy issues that sensitive information contained in the dataset can be leaked. The most effective way to address that is to encrypt the data before outsourcing. Nevertheless, it remains a grand challenge to process queries in ciphertext efficiently. In this work, we shall focus on solving one representative query task, namely dynamic skyline query, in a secure manner over the cloud. However, it is difficult to be performed on encrypted data as its dynamic domination criteria require both subtraction and comparison, which cannot be directly supported by a single encryption scheme efficiently. To this end, we present a novel framework called SCALE. It works by transforming traditional dynamic skyline domination into pure comparisons. The whole process can be completed in single-round interaction between user and the cloud. We theoretically prove that the outsourced database, query requests, and returned results are all kept secret under our model. Moreover, we also present an efficient strategy for dynamic insertion and deletion of stored records. Empirical study over a series of datasets demonstrates that our framework improves the efficiency of query processing by nearly three orders of magnitude compared to the state-of-the-art.

cs.DB↗

arXiv2019

The (b, c)-inverse in semigroups and rings with involution

In this paper, we first prove that if a is both left (b, c)-invertible and left (c, b)-invertible, then a is both (b, c)-invertible and (c, b)-invertible in a *-monoid, which generalized the recent result about the inverse along an element by Wang and Mosic, under the conditions (ab)* = ab and (ac)* = ac. In addition, we consider that ba is (c, b)- invertible, and at the same time ca is (b, c)-invertible under the same conditions, which extend the related results about Moore-Penrose inverses by Chen et al. to (b, c)-inverses. As applications, we obtain that under condition (a2)* = a2, a is an EP element if and only if a is one-sided core invertible if and only if a is group invertible.

math.RA↗