arXiv subjects

Mingxuan Sun

Publications and source records attributed to Mingxuan Sun.

At least 19 recordsLinked to original sources

arXiv2026

Groundhog Bit-Flip Attack: Seeding Infinite Generation Loops in Mixture-of-Experts LLMs through Bit Flips

Mixture-of-Experts (MoE) architectures enable scalable and efficient large language models (LLMs) by selectively activating expert sub-networks through a routing mechanism. However, this adaptive design introduces a new attack surface: specific experts become disproportionately correlated with certain tokens (e.g., end-of-sequence), allowing adversaries to manipulate model behavior via lightweight perturbations. In this work, we present \textbf{Groundhog Bit-Flip Attack (GBFA)}, the first bit-flip-based \textit{ Denial-of-Wallet availability attack} against MoE-based LLMs. By identifying and flipping routing-layer bits associated with related expert activations, we demonstrate that GBFA substantially extends the decoding token usage across three different LLM modes: conversational, reasoning, and agentic tasks, while largely preserving semantic fidelity. Across four main real-world MoE-based LLMs, manually deactivating on average fewer than \textbf{4 experts} drives average output inflation to $\mathbf{5912\%}$, with the majority of test samples reaching max tokens. These results reveal a robustness vulnerability of MoE architectures to bit flip, and highlight the potential of GBFA as an availability attack against LLMs.

cs.CL↗

arXiv2026

Optimal Condition Numbers in Low-Rank Positive Semidefinite Matrix Sensing

In this paper we focus on the stability of positive semidefinite matrix sensing maps $\Phi_{\mathcal{A}}(X)=(\langle A_i,X\rangle)_{i=1}^m$ where $A_i\succeq 0$, $X\succeq0$ and $\operatorname{rank}(X)\le r$. We introduce the bi-Lipschitz constants of $\Phi_{\mathcal{A}}(X)$ and define the global condition numbers as the ratio of upper and lower Lipschitz constants. We give deterministic universal lower bounds for these condition numbers, that depend only on the rank $r$ and on the underlying field. We then investigate the random rank-one Gaussian measurements and show that our lower bounds on condition numbers are asymptotically sharp, and therefore the random rank-one Gaussian measurements are asymptotically optimal. As an application, we derive the stability guarantees for an $\ell_1$-residual PhaseLift-type estimator at the optimal sampling scale.

cs.IT↗

arXiv2026

CONSISTRE: A Unified Consistency-Aware Framework for Document-Level Relation Extraction with Large Language Models

Document-level relation extraction (DocRE) aims to extract relations among multiple entities across extended contexts while maintaining consistency across predicted triples. Although large language models (LLMs) show remarkable reasoning capabilities in information extraction, their predictions are typically generated independently for each candidate triple and may violate fundamental relational constraints such as transitivity, symmetry, and functional uniqueness, leading to contradictory and unreliable outputs. We propose CONSISTRE, a unified consistency-aware framework for DocRE that addresses this limitation through two complementary tracks. The first operates at inference time for black-box LLMs, combining constraint-aware prompting, constraint-based verification, and iterative self-reflection to refine predictions without task-specific fine-tuning. The second injects consistency knowledge into smaller open-source models via a knowledge distillation and reinforcement learning pipeline: reasoning traces from a powerful teacher are distilled into a student via supervised fine-tuning, followed by GRPO alignment using a composite reward that jointly optimizes extraction performance and relational consistency. Together, the two tracks cover both API-accessible and locally deployable scenarios under a unified consistency formulation. Experiments on DocRED show that both tracks outperform their baselines, with the inference-time track achieving competitive F1 using off-the-shelf black-box LLMs and the training-time track substantially narrowing the gap between 7--8B open-source models and state-of-the-art proprietary LLMs at a fraction of their inference cost. Ablation studies confirm that explicit consistency modeling mitigates relational contradictions and enhances the reliability of LLM-based DocRE across both deployment paradigms.

cs.CL↗

arXiv2026

KAP-CPD: Kernel Aggregation for Change-Point Detection in Dynamic Networks

Change-point detection in dynamic networks has received much attention due to its broad applications in social networks and biological systems. Kernel-based methods have shown strong potential for this problem. However, their performance can depend sensitively on the choice of kernel, and selecting an appropriate kernel is challenging when the underlying change pattern is unknown. Motivated by this challenge, we propose KAP-CPD, a new kernel-based testing framework for change-point detection in dynamic networks. KAP-CPD aggregates information from multiple kernels, allowing it to adapt to diverse change patterns. The proposed method does not assume specific underlying network distribution, and achieves strong empirical power across a wide range of network change scenarios. To improve scalability, we further develop a fast analytic testing procedure, KAPf-CPD, that substantially reduces computation time for long network sequences compared with permutation-based alternatives and current state-of-the-art methods. We evaluate our proposed framework through extensive simulations and real-world data on email communication networks and brain functional connectivity networks.

stat.ME↗

arXiv2026

Spatiotemporal-Aware Bit-Flip Injection on DNN-based Advanced Driver Assistance Systems (extended version)

Modern advanced driver assistance systems (ADAS) rely on deep neural networks (DNNs) for perception and planning. Since DNNs' parameters reside in DRAM during inference, bit flips caused by cosmic radiation or low-voltage operation may corrupt DNN computations, distort driving decisions, and lead to real-world incidents. This paper presents a SpatioTemporal-Aware Fault Injection (STAFI) framework to locate critical fault sites in DNNs for ADAS efficiently. Spatially, we propose a Progressive Metric-guided Bit Search (PMBS) that efficiently identifies critical network weight bits whose corruption causes the largest deviations in driving behavior (e.g., unintended acceleration or steering). Furthermore, we develop a Critical Fault Time Identification (CFTI) mechanism that determines when to trigger these faults, taking into account the context of real-time systems and environmental states, to maximize the safety impact. Experiments on DNNs for a production ADAS demonstrate that STAFI uncovers 29.56x more hazard-inducing critical faults than the strongest baseline.

cs.CR↗

arXiv2025

FLEX-MoE: Federated Mixture-of-Experts with Load-balanced Expert Assignment for Edge Computing

Mixture-of-Experts (MoE) models enable scalable neural networks through conditional computation, offering enhanced effectiveness and efficiency for next-generation wireless communications. However, deploying MoE with federated learning (FL) over wireless and IoT edge networks faces two critical challenges: 1) resource-constrained clients cannot store large AI models with full expert sets, and 2) non-IID data distributions cause severe expert load imbalance that degrades model performance. To this end, we propose FLEX-MoE, a federated MoE framework that jointly optimizes expert assignment and load balancing under limited client capacity. Specifically, our approach introduces client-expert fitness scores that quantify expert suitability for local datasets through training feedback, and employs an optimization-based algorithm to maximize client-expert specialization while enforcing balanced expert utilization system-wide. Unlike greedy methods that focus solely on personalization while ignoring load imbalance, FLEX-MoE addresses expert utilization skew, which is particularly severe in heterogeneous edge FL. Our experimental results demonstrate superior accuracy and consistently balanced expert utilization across diverse resource-constrained scenarios for edge computing.

cs.LG↗

arXiv2025

Learning to Rank with Top-$K$ Fairness

Fairness in ranking models is crucial, as disparities in exposure can disproportionately affect protected groups. Most fairness-aware ranking systems focus on ensuring comparable average exposure for groups across the entire ranked list, which may not fully address real-world concerns. For example, when a ranking model is used for allocating resources among candidates or disaster hotspots, decision-makers often prioritize only the top-$K$ ranked items, while the ranking beyond top-$K$ becomes less relevant. In this paper, we propose a list-wise learning-to-rank framework that addresses the issues of inequalities in top-$K$ rankings at training time. Specifically, we propose a top-$K$ exposure disparity measure that extends the classic exposure disparity metric in a ranked list. We then learn a ranker to balance relevance and fairness in top-$K$ rankings. Since direct top-$K$ selection is computationally expensive for a large number of items, we transform the non-differentiable selection process into a differentiable objective function and develop efficient stochastic optimization algorithms to achieve both high accuracy and sufficient fairness. Extensive experiments demonstrate that our method outperforms existing methods.

cs.LG↗

arXiv2025

Path to Intelligence: Measuring Similarity between Human Brain and Large Language Model Beyond Language Task

Large language models (LLMs) have demonstrated human-like abilities in language-based tasks. While language is a defining feature of human intelligence, it emerges from more fundamental neurophysical processes rather than constituting the basis of intelligence itself. In this work, we study the similarity between LLM internal states and human brain activity in a sensory-motor task rooted in anticipatory and visuospatial behavior. These abilities are essential for cognitive performance that constitute human intelligence. We translate the sensory-motor task into natural language in order to replicate the process for LLMs. We extract hidden states from pre-trained LLMs at key time steps and compare them to human intracranial EEG signals. Our results reveal that LLM-derived reactions can be linearly mapped onto human neural activity. These findings suggest that LLMs, with a simple natural language translation to make them understand temporal-relevant tasks, can approximate human neurophysical behavior in experiments involving sensory stimulants. In all, our contribution is two-fold: (1) We demonstrate similarity between LLM and human brain activity beyond language-based tasks. (2) We demonstrate that with such similarity, LLMs could help us understand human brains by enabling us to study topics in neuroscience that are otherwise challenging to tackle.

q-bio.NC↗

arXiv2025

Pruning and Malicious Injection: A Retraining-Free Backdoor Attack on Transformer Models

Transformer models have demonstrated exceptional performance and have become indispensable in computer vision (CV) and natural language processing (NLP) tasks. However, recent studies reveal that transformers are susceptible to backdoor attacks. Prior backdoor attack methods typically rely on retraining with clean data or altering the model architecture, both of which can be resource-intensive and intrusive. In this paper, we propose Head-wise Pruning and Malicious Injection (HPMI), a novel retraining-free backdoor attack on transformers that does not alter the model's architecture. Our approach requires only a small subset of the original data and basic knowledge of the model architecture, eliminating the need for retraining the target transformer. Technically, HPMI works by pruning the least important head and injecting a pre-trained malicious head to establish the backdoor. We provide a rigorous theoretical justification demonstrating that the implanted backdoor resists detection and removal by state-of-the-art defense techniques, under reasonable assumptions. Experimental evaluations across multiple datasets further validate the effectiveness of HPMI, showing that it 1) incurs negligible clean accuracy loss, 2) achieves at least 99.55% attack success rate, and 3) bypasses four advanced defense mechanisms. Additionally, relative to state-of-the-art retraining-dependent attacks, HPMI achieves greater concealment and robustness against diverse defense strategies, while maintaining minimal impact on clean accuracy.

cs.LG↗

arXiv2025

Efficient Training of Large-Scale AI Models Through Federated Mixture-of-Experts: A System-Level Approach

The integration of Federated Learning (FL) and Mixture-of-Experts (MoE) presents a compelling pathway for training more powerful, large-scale artificial intelligence models (LAMs) on decentralized data while preserving privacy. However, efficient federated training of these complex MoE-structured LAMs is hindered by significant system-level challenges, particularly in managing the interplay between heterogeneous client resources and the sophisticated coordination required for numerous specialized experts. This article highlights a critical, yet underexplored concept: the absence of robust quantitative strategies for dynamic client-expert alignment that holistically considers varying client capacities and the imperative for system-wise load balancing. Specifically, we propose a conceptual system design for intelligent client-expert alignment that incorporates dynamic fitness scoring, global expert load monitoring, and client capacity profiling. By tackling these systemic issues, we can unlock more scalable, efficient, and robust training mechanisms {with fewer communication rounds for convergence}, paving the way for the widespread deployment of large-scale federated MoE-structured LAMs in edge computing with ultra-high communication efficiency.

cs.LG↗

arXiv2025

VisAlgae 2023: A Dataset and Challenge for Algae Detection in Microscopy Images

Microalgae, vital for ecological balance and economic sectors, present challenges in detection due to their diverse sizes and conditions. This paper summarizes the second "Vision Meets Algae" (VisAlgae 2023) Challenge, aiming to enhance high-throughput microalgae cell detection. The challenge, which attracted 369 participating teams, includes a dataset of 1000 images across six classes, featuring microalgae of varying sizes and distinct features. Participants faced tasks such as detecting small targets, handling motion blur, and complex backgrounds. The top 10 methods, outlined here, offer insights into overcoming these challenges and maximizing detection accuracy. This intersection of algae research and computer vision offers promise for ecological understanding and technological advancement. The dataset can be accessed at: https://github.com/juntaoJianggavin/Visalgae2023/.

cs.CV↗

arXiv2025

Adaptive Neural Control with Desired Approximation: An Integral Lyapunov Function Approach

The inherent approximation ability of neural networks plays an essential role in adaptive neural control, where the prerequisite for existence of the compact set is crucial in the control designs. Instead of using practical system state, in this paper, the desired approximation approach is characterized to tackle such a problem, where the desired state signal is required only as the input to the network. An integral Lyapunov function-based adaptive controller is designed, in the sense of the error tracking, where the treatment of the state-dependent input gain is adopted. Theoretical results for the performance analysis of the integral and incremental adaptation algorithms are presented in details. In particular, the boundedness of the variables in the closed-loop is characterized, while the transient performance of the output error is analytically quantified. It is shown that the proposed control schemes assure that the tracking error converges to an adjustable set without any requirement on the knowledge of the region that the practical variables evolve, and remove the requirement for the setting of initial conditions including system states and weight estimates.

eess.SY↗

arXiv2025

Enhancing Time Series Forecasting via Multi-Level Text Alignment with LLMs

The adaptation of large language models (LLMs) to time series forecasting poses unique challenges, as time series data is continuous in nature, while LLMs operate on discrete tokens. Despite the success of LLMs in natural language processing (NLP) and other structured domains, aligning time series data with language-based representations while maintaining both predictive accuracy and interpretability remains a significant hurdle. Existing methods have attempted to reprogram time series data into text-based forms, but these often fall short in delivering meaningful, interpretable results. In this paper, we propose a multi-level text alignment framework for time series forecasting using LLMs that not only improves prediction accuracy but also enhances the interpretability of time series representations. Our method decomposes time series into trend, seasonal, and residual components, which are then reprogrammed into component-specific text representations. We introduce a multi-level alignment mechanism, where component-specific embeddings are aligned with pre-trained word tokens, enabling more interpretable forecasts. Experiments on multiple datasets demonstrate that our method outperforms state-of-the-art models in accuracy while providing good interpretability.

cs.CL↗

arXiv2024

DualGFL: Federated Learning with a Dual-Level Coalition-Auction Game

Despite some promising results in federated learning using game-theoretical methods, most existing studies mainly employ a one-level game in either a cooperative or competitive environment, failing to capture the complex dynamics among participants in practice. To address this issue, we propose DualGFL, a novel Federated Learning framework with a Dual-level Game in cooperative-competitive environments. DualGFL includes a lower-level hedonic game where clients form coalitions and an upper-level multi-attribute auction game where coalitions bid for training participation. At the lower-level DualGFL, we introduce a new auction-aware utility function and propose a Pareto-optimal partitioning algorithm to find a Pareto-optimal partition based on clients' preference profiles. At the upper-level DualGFL, we formulate a multi-attribute auction game with resource constraints and derive equilibrium bids to maximize coalitions' winning probabilities and profits. A greedy algorithm is proposed to maximize the utility of the central server. Extensive experiments on real-world datasets demonstrate DualGFL's effectiveness in improving both server utility and client utility.

cs.GT↗

arXiv2024

Constructive RNNs: An Error-Recurrence Perspective on Time-Variant Zero Finding Problem Solving Under Uncertainty

When facing time-variant problems in analog computing, the desirable RNN design requires finite-time convergence and robustness with respect to various types of uncertainties, due to the time-variant nature and difficulties in implementation. It is very worthwhile to explore terminal zeroing neural networks, through examining and applying available attracting laws. In this paper, from a control-theoretic point of view, an error recurrence system approach is presented by equipping with uncertainty compensation in the pre-specified error dynamics, capable of enhancing robustness properly. Novel rectifying actions are designed to make finite-time settling so that the convergence speed and the computing accuracy of time-variant computing can be improved. Double-power and power-exponential rectifying actions are respectively formed to construct specific models, while the particular expressions of settling time function for the former are presented, and for the latter the proximate settling-time estimations are given, with which the fixed-time convergence of the corresponding models is in turn established. Moreover, the uncertainty compensation by the signum/smoothing-signum techniques are adopted for finite-duration stabilization. Theoretical results are presented to demonstrate effectiveness (involving fixed-time convergence and robustness) of the proposed computing schemes for the time-variant QP problem solving.

eess.SY↗

arXiv2023

Fractional Barrier Lyapunov Functions with Application to Learning Control

Barrier Lyapunov functions are suitable for learning control designs, due to their feature of finite duration tracking. This paper presents fractional barrier Lyapunov functions, provided and compared with the conventional ones in the error-constraint learning control designs. Two error models are adopted and the desired compensation control approach is applied for a non-parametric design, allowing two kinds of uncertainties involved in the error dynamics. Theoretical results about existence of the solution and convergence of the learning control schemes are presented. It is shown that fully-saturated learning algorithms play important role in assuring boundedness of the estimates, by which the error constraint objective can be achieved. Moreover, the robust technique is developed through modifying the discontinuous action involved in the learning control scheme that yields the expected tracking performance in the presence of residual.

eess.SY↗

arXiv2020

Time-Varying and Nonlinearly Scaled Consensus of Multiagent Systems: A Generic Attracting Law Approach

This paper presents the design and analysis of the finite/fixed-time scaled consensus for multiagent systems. A study on a generic attracting law, the certain classes of nonlinear systems that admit attractors with finite/fixed-time convergence, is at first given for the consensus purpose. The estimates for the lower and upper bounds on the settling time functions are provided through the two-phase analysis. The given estimates are initial state dependent, but the durations are finite, without regarding the values that the initial states take. According to the generic attracting law, distributed protocols are proposed for multiagent systems with undirected and detail-balanced directed graphs, respectively, where the scaled strategies, including time-varying and nonlinear scales, are adopted. It is shown that the finite/fixed-time consensus for the multiagent system undertaken can still be achieved, even though both time-varying and nonlinear scales are taken among agents. Numerical simulation of two illustrative examples are given to verify effectiveness of the proposed finite-duration consensus protocols.

eess.SY↗

arXiv2018

Intelligent Wireless Communications Enabled by Cognitive Radio and Machine Learning

The ability to intelligently utilize resources to meet the need of growing diversity in services and user behavior marks the future of wireless communication systems. Intelligent wireless communications aims at enabling the system to perceive and assess the available resources, to autonomously learn to adapt to the perceived wireless environment, and to reconfigure its operating mode to maximize the utility of the available resources. The perception capability and reconfigurability are the essential features of cognitive radio while modern machine learning techniques project great potential in system adaptation. In this paper, we discuss the development of the cognitive radio technology and machine learning techniques and emphasize their roles in improving spectrum and energy utility of wireless communication systems. We describe the state-of-the-art of relevant techniques, covering spectrum sensing and access approaches and powerful machine learning algorithms that enable spectrum- and energy-efficient communications in dynamic wireless environments. We also present practical applications of these techniques and identify further research challenges in cognitive radio and machine learning as applied to the existing and future wireless communication systems.

cs.IT↗