arXiv subjects

Xiao Li

Publications and source records attributed to Xiao Li.

At least 127 records · Page 7Linked to original sources

arXiv2026

Data-Driven Deep MIMO Detection:Network Architectures and Generalization Analysis

In practical Multiuser Multiple-Input Multiple-Output (MU-MIMO) systems, symbol detection remains challenging due to severe inter-user interference and sensitivity to Channel State Information (CSI) uncertainty. In contrast to the mostly studied belief propagation-type model-driven methods, which incur high computational complexity, Soft Interference Cancellation (SIC) strikes a good balance between performance and complexity. To further address CSI mismatch and nonlinear effects, the recently proposed data-driven deep neural receivers, such as DeepSIC, leverage the advantages of deep neural networks for interference cancellation and symbol detection, demonstrating strong empirical performance. However, there is still a lack of theoretical underpinning for why and to what extent DeepSIC could generalize with the number of training samples. This paper proposes inspecting the fully data-driven DeepSIC detection within a Network-of-MLPs architecture, which is composed of multiple interconnected MLPs via outer and inner Directed Acyclic Graphs (DAGs). Within such an architecture, DeepSIC can be upgraded as a graph-based message-passing process using Graph Neural Networks (GNNs), termed GNNSIC, with shared model parameters across users and iterations. Notably, GNNSIC achieves excellent expressivity comparable to DeepSIC with substantially fewer trainable parameters, resulting in improved sample efficiency and enhanced user generalization. By conducting a norm-based generalization analysis using Rademacher complexity, we reveal that an exponential dependence on the number of iterations for DeepSIC can be eliminated in GNNSIC due to parameter sharing. Simulation results demonstrate that GNNSIC attains comparable or improved Symbol Error Rate (SER) performance to DeepSIC with significantly fewer parameters and training samples.

eess.SP↗

arXiv2026

Navigating heterogeneous protein landscapes through geometry-aware smoothing

The evolutionary fitness landscape of biological molecules is extremely sparse and heterogeneous, with functional sequences forming isolated dense ``islands'' within a vast combinatorial space of largely non-functional variants. Protein sequences, in particular, exemplify this structure, yet most generative artificial intelligence models implicitly assume a homogeneous data distribution. We show that this assumption fundamentally breaks down in heterogeneous biological sequence spaces: fixed global noise levels impose a destructive trade-off, either oversmoothing dense functional clusters or fragmenting sparse regions and producing non-functional hallucinations. To address this limitation, we introduce \emph{Density-Dependent Smoothing} (DDS), a geometry-aware generative framework that adapts stochastic smoothing to the local density of the underlying sequence landscape. By inversely coupling diffusion noise to estimated sequence density, DDS enables gentle refinement in high-density functional regions while promoting controlled exploration across sparse regions. Implemented as a plug-in mechanism for discrete molecular sampling, DDS consistently outperforms state-of-the-art diffusion and autoregressive models across antibody repertoires, therapeutic antibody design, antimicrobial peptide generation and coronavirus antibody design. Together, these results show that fixed global smoothing assumptions fundamentally limit generative modeling in sparse biological sequence spaces, and that geometry-aware smoothing removes this constraint, enabling reliable exploration and design previously unattainable with fixed-noise generative models.

cs.CE↗

arXiv2026

A Generalized Version of Chung's Lemma and its Applications

Chung's Lemma is a classical tool for establishing asymptotic convergence rates of (stochastic) optimization methods under strong convexity-type assumptions and appropriate polynomial diminishing step sizes. In this work, we develop a generalized version of Chung's Lemma, which provides a simple non-asymptotic convergence framework for a more general family of step size rules. We demonstrate broad applicability of the proposed generalized lemma by deriving tight non-asymptotic convergence rates for a large variety of stochastic methods. In particular, we obtain partially new non-asymptotic complexity results for stochastic optimization methods, such as Stochastic Gradient Descent (SGD) and Random Reshuffling (RR), under a general $(θ,μ)$-Polyak-Lojasiewicz (PL) condition and for various step sizes strategies, including polynomial, constant, exponential, and cosine step sizes rules. Notably, as a by-product of our analysis, we observe that exponential step sizes exhibit superior adaptivity to both landscape geometry and gradient noise; specifically, they achieve optimal convergence rates without requiring exact knowledge of the underlying landscape or separate parameter selection strategies for noisy and noise-free regimes. Our results demonstrate that the developed variant of Chung's Lemma offers a versatile, systematic, and streamlined approach to establish non-asymptotic convergence rates under general step size rules.

math.OC↗

arXiv2026

Understanding Representation Dynamics of Diffusion Models via Low-Dimensional Modeling

Diffusion models, though originally designed for generative tasks, have demonstrated impressive self-supervised representation learning capabilities. A particularly intriguing phenomenon in these models is the emergence of unimodal representation dynamics, where the quality of learned features peaks at an intermediate noise level. In this work, we conduct a comprehensive theoretical and empirical investigation of this phenomenon. Leveraging the inherent low-dimensionality structure of image data, we theoretically demonstrate that the unimodal dynamic emerges when the diffusion model successfully captures the underlying data distribution. The unimodality arises from an interplay between denoising strength and class confidence across noise scales. Empirically, we further show that, in classification tasks, the presence of unimodal dynamics reliably reflects the generalization of the diffusion model: it emerges when the model generates novel images and gradually transitions to a monotonically decreasing curve as the model begins to memorize the training data.

cs.LG↗

arXiv2026

A Unified View of Attention and Residual Sinks: Outlier-Driven Rescaling is Essential for Transformer Training

We investigate the functional role of emergent outliers in large language models, specifically attention sinks (a few tokens that consistently receive large attention logits) and residual sinks (a few fixed dimensions with persistently large activations across most tokens). We hypothesize that these outliers, in conjunction with the corresponding normalizations (\textit{e.g.}, softmax attention and RMSNorm), effectively rescale other non-outlier components. We term this phenomenon \textit{outlier-driven rescaling} and validate this hypothesis across different model architectures and training token counts. This view unifies the origin and mitigation of both sink types. Our main conclusions and observations include: (1) Outliers function jointly with normalization: removing normalization eliminates the corresponding outliers but degrades training stability and performance; directly clipping outliers while retaining normalization leads to degradation, indicating that outlier-driven rescaling contributes to training stability. (2) Outliers serve more as rescale factors rather than contributors, as the final contributions of attention and residual sinks are significantly smaller than those of non-outliers. (3) Outliers can be absorbed into learnable parameters or mitigated via explicit gated rescaling, leading to improved training performance (average gain of 2 points) and enhanced quantization robustness (1.2 points degradation under W4A4 quantization).

cs.CL↗

arXiv2026

AMGFormer: Adaptive Multi-Granular Transformer for Brain Tumor Segmentation with Missing Modalities

Multimodal MRI is essential for brain tumor segmentation, yet missing modalities in clinical practice cause existing methods to exhibit >40% performance variance across modality combinations, rendering them clinically unreliable. We propose AMGFormer, achieving significantly improved stability through three synergistic modules: (1) QuadIntegrator Bridge (QIB) enabling spatially adaptive fusion maintaining consistent predictions regardless of available modalities, (2) Multi-Granular Attention Orchestrator (MGAO) focusing on pathological regions to reduce background sensitivity, and (3) Modality Quality-Aware Enhancement (MQAE) preventing error propagation from corrupted sequences. On BraTS 2018, our method achieves 89.33% WT, 82.70% TC, 67.23% ET Dice scores with <0.5% variance across 15 modality combinations, solving the stability crisis. Single-modality ET segmentation shows 40-81% relative improvements over state-of-the-art methods. The method generalizes to BraTS 2020/2021, achieving up to 92.44% WT, 89.91% TC, 84.57% ET. The model demonstrates potential for clinical deployment with 1.2s inference. Code: https://github.com/guochengxiangives/AMGFormer.

eess.IV↗

arXiv2026

AoI-Driven Queue Management and Power Control in V2V Networks: A GNN-Enhanced MARL Approach

Queue management and resource allocation play a critical role in enabling cooperative status awareness in vehicular networks. This paper investigates the problem of age of information (AoI)-aware status updates in vehicle-to-vehicle (V2V) communication, where each vehicle's status is represented by multiple interdependent packets. To enable fine-grained queue management at the packet level under resource constraints, we formulate a joint optimization problem that simultaneously learns active packet dropping and transmit power control strategies. A hybrid action space is designed to support both discrete dropping decisions and continuous power control. To exploit the graph-structured interference inherent in V2V topology, a graph neural network (GNN) is introduced to aggregate slowly varying large-scale fading, allowing agents to capture topological dependencies implicitly without frequent message exchange. The overall framework is built upon multi-agent proximal policy optimization (MAPPO), with centralized training and decentralized execution (CTDE). Simulations demonstrate that the proposed method significantly reduces average AoI across a wide range of network densities, channel conditions, and traffic loads, consistently outperforming several baselines.

eess.SP↗

arXiv2026

FormulaReasoning: A Dataset for Formula-Based Numerical Reasoning

The application of physics formulas is a fundamental human capability in numerical reasoning. While existing datasets often rely on implicit mathematical knowledge, they rarely explicitate the underlying formulas. To address this, we introduce FormulaReasoning, a new benchmark for formula-based numerical reasoning comprising 5,324 questions requiring calculations grounded in external physics principles. We provide high-quality, fine-grained annotations in English and Chinese--including formula structures, parameter names, symbols, values, and units--curated through manual effort and LLM-assisted validation. Additionally, we provide a consolidated formula database as an external knowledge source. To further challenge model performance, we develop an extended version of the dataset by coupling multiple questions. We evaluate various architectural and methodological frameworks, including retrieval-augmented methods, modular reasoning (formula generation, parameter extraction, and calculation), and preference-based optimization. Our analysis identifies critical challenges in formula-based reasoning, highlighting significant opportunities for future methodological advancement.

cs.CL↗

arXiv2026

PartImageNet++ Dataset: Enhancing Visual Models with High-Quality Part Annotations

To address the scarcity of high-quality part annotations in existing datasets, we introduce PartImageNet++ (PIN++), a dataset that provides detailed part annotations for all categories in ImageNet-1K. With 100 annotated images per category, totaling 100K images, PIN++ represents the most comprehensive dataset covering a diverse range of object categories. Leveraging PIN++, we propose a Multi-scale Part-supervised recognition Model (MPM) for robust classification on ImageNet-1K. We first trained a part segmentation network using PIN++ and used it to generate pseudo part labels for the remaining unannotated images. MPM then integrated a conventional recognition architecture with auxiliary bypass layers, jointly supervised by both pseudo part labels and the original part annotations. Furthermore, we conducted extensive experiments on PIN++, including part segmentation, object segmentation, and few-shot learning, exploring various ways to leverage part annotations in downstream tasks. Experimental results demonstrated that our approach not only enhanced part-based models for robust object recognition but also established strong baselines for multiple downstream tasks, highlighting the potential of part annotations in improving model performance. The dataset and the code are available at https://github.com/LixiaoTHU/PartImageNetPP.

cs.CV↗

arXiv2025

Classical vs quantum dynamics and the onset of chaos in a macrospin system

We study a periodically driven macrospin system with anisotropic long-range interactions and collective dissipation, described by a Lindblad master equation. In the thermodynamic limit ($N\to\infty$), a mean-field treatment yields classical equations of motion, whose dynamics are characterized via the maximal Lyapunov exponent (MLE). Focusing on the thermodynamic limit, we map out chaotic, quasiperiodic, and periodic phases via bifurcation diagrams, MLEs, and Fourier spectra of evolved observables, identifying classic period-doubling bifurcations and fractal boundaries in the regions of attractors. Finite-size quantum simulations in the Dicke basis reveal that while both quantum and classical systems exhibit diverse dynamical phases, finite-size effects suppress some behaviors present in the thermodynamic limit. The sign of $λ_{\mathrm{max}}$ serves as a key indicator of convergence between quantum and classical dynamics, which agree over timescales up to the Lyapunov time. Analysis of the density matrix shows that convergence occurs only when its nonzero elements are sharply localized. However, the nonconvergence does not imply a fundamental difference between quantum and classical dynamics: in chaotic regimes, although the evolution orbits of quantum and classical systems show significant differences, quantum evolution becomes mixed and diffusively explores the Hilbert space, signaling quantum chaos, which can be confirmed by the delocalized nature of the density matrix.

quant-ph↗

arXiv2025

Toward building next-generation Geocoding systems: a systematic review

Geocoding systems are widely used in both scientific research for spatial analysis and everyday life through location-based services. The quality of geocoded data significantly impacts subsequent processes and applications, underscoring the need for next-generation systems. In response to this demand, this review first characterizes the technical requirements for next-generation geocoding inputs and outputs. We then decompose the geocoding workflow into modular functional units and survey existing implementations. For each component, we identify methodological limitations, articulate domain-specific research questions and hypotheses, and outline evaluation strategies needed. Finally, we identify opportunities to improve next-generation geocoding systems in light of recent technological advances. We envision that this review provides a technical foundation and research agenda for advancing the design, assessment, and deployment of next-generation geocoding systems.

cs.SE↗

arXiv2025

CoTDeceptor:Adversarial Code Obfuscation Against CoT-Enhanced LLM Code Agents

LLM-based code agents(e.g., ChatGPT Codex) are increasingly deployed as detector for code review and security auditing tasks. Although CoT-enhanced LLM vulnerability detectors are believed to provide improved robustness against obfuscated malicious code, we find that their reasoning chains and semantic abstraction processes exhibit exploitable systematic weaknesses.This allows attackers to covertly embed malicious logic, bypass code review, and propagate backdoored components throughout real-world software supply chains.To investigate this issue, we present CoTDeceptor, the first adversarial code obfuscation framework targeting CoT-enhanced LLM detectors. CoTDeceptor autonomously constructs evolving, hard-to-reverse multi-stage obfuscation strategy chains that effectively disrupt CoT-driven detection logic.We obtained malicious code provided by security enterprise, experimental results demonstrate that CoTDeceptor achieves stable and transferable evasion performance against state-of-the-art LLMs and vulnerability detection agents. CoTDeceptor bypasses 14 out of 15 vulnerability categories, compared to only 2 bypassed by prior methods. Our findings highlight potential risks in real-world software supply chains and underscore the need for more robust and interpretable LLM-powered security analysis systems.

cs.CR↗

arXiv2025

Timely Parameter Updating in Over-the-Air Federated Learning

Incorporating over-the-air computations (OAC) into the model training process of federated learning (FL) is an effective approach to alleviating the communication bottleneck in FL systems. Under OAC-FL, every client modulates its intermediate parameters, such as gradient, onto the same set of orthogonal waveforms and simultaneously transmits the radio signal to the edge server. By exploiting the superposition property of multiple-access channels, the edge server can obtain an automatically aggregated global gradient from the received signal. However, the limited number of orthogonal waveforms available in practical systems is fundamentally mismatched with the high dimensionality of modern deep learning models. To address this issue, we propose Freshness Freshness-mAgnItude awaRe top-k (FAIR-k), an algorithm that selects, in each communication round, the most impactful subset of gradients to be updated over the air. In essence, FAIR-k combines the complementary strengths of the Round-Robin and Top-k algorithms, striking a delicate balance between timeliness (freshness of parameter updates) and importance (gradient magnitude). Leveraging tools from Markov analysis, we characterize the distribution of parameter staleness under FAIR-k. Building on this, we establish the convergence rate of OAC-FL with FAIR-k, which discloses the joint effect of data heterogeneity, channel noise, and parameter staleness on the training efficiency. Notably, as opposed to conventional analyses that assume a universal Lipschitz constant across all the clients, our framework adopts a finer-grained model of the data heterogeneity. The analysis demonstrates that since FAIR-k promotes fresh (and fair) parameter updates, it not only accelerates convergence but also enhances communication efficiency by enabling an extended period of local training without significantly affecting overall training efficiency.

cs.LG↗

arXiv2025

A Systematic Study of Code Obfuscation Against LLM-based Vulnerability Detection

As large language models (LLMs) are increasingly adopted for code vulnerability detection, their reliability and robustness across diverse vulnerability types have become a pressing concern. In traditional adversarial settings, code obfuscation has long been used as a general strategy to bypass auditing tools, preserving exploitability without tampering with the tools themselves. Numerous efforts have explored obfuscation methods and tools, yet their capabilities differ in terms of supported techniques, granularity, and programming languages, making it difficult to systematically assess their impact on LLM-based vulnerability detection. To address this gap, we provide a structured systematization of obfuscation techniques and evaluate them under a unified framework. Specifically, we categorize existing obfuscation methods into three major classes (layout, data flow, and control flow) covering 11 subcategories and 19 concrete techniques. We implement these techniques across four programming languages (Solidity, C, C++, and Python) using a consistent LLM-driven approach, and evaluate their effects on 15 LLMs spanning four model families (DeepSeek, OpenAI, Qwen, and LLaMA), as well as on two coding agents (GitHub Copilot and Codex). Our findings reveal both positive and negative impacts of code obfuscation on LLM-based vulnerability detection, highlighting conditions under which obfuscation leads to performance improvements or degradations. We further analyze these outcomes with respect to vulnerability characteristics, code properties, and model attributes. Finally, we outline several open problems and propose future directions to enhance the robustness of LLMs for real-world vulnerability detection.

cs.CR↗

arXiv2025

Power Consumption and Energy Efficiency of Mid-Band XL-MIMO: Modeling, Scaling Laws, and Performance Insights

Mid-band extra-large-scale multiple-input multiple-output (XL-MIMO), emerging as a critical enabler for future communication systems, is expected to deliver significantly higher throughput by leveraging the extended bandwidth and enlarged antenna aperture. However, power consumption remains a significant concern due to the enlarged system dimension, underscoring the need for thorough investigations into efficient system design and deployment. To this end, an in-depth study is conducted on mid-band XL-MIMO systems. Specifically, a comprehensive power consumption model is proposed, encompassing the power consumption of major hardware components and signal processing procedures, while capturing the influence of key system parameters. Considering typical near-field propagation characteristics, closed-form approximations of throughput are derived, providing an analytical framework for assessing energy efficiency (EE). Based on the proposed framework, the scaling law of EE with respect to key system configurations is derived, offering valuable insights for system design. Subsequently, extensions and comparisons are conducted among representative multi-antenna technologies, demonstrating the superiority of mid-band XL-MIMO in EE. Extensive numerical results not only verify the tightness of the throughput analysis but also validate the EE evaluations, unveiling the potential of energy-efficient mid-band XL-MIMO systems.

eess.SP↗

arXiv2025

Large and Small Model Collaboration for Air Interface

Large artificial intelligence models (LAMs) have shown strong capability in wireless communications, yet existing works mainly rely on their generalized knowledge across environments while overlooking the potential gains of environment-specific adaptation. Directly fine-tuning LAMs for adaptation is often impractical due to prohibitive training costs, low inference efficiency in multi-user scenarios, and the risk of catastrophic forgetting, in addition to the limited accessibility of model parameters. To address these limitations, we establish a collaborative framework for air interface. In this framework, unlike prior approaches that either depend solely on LAMs or require direct fine-tuning, LAMs are exploited as a universal channel knowledge base while small artificial intelligence models (SAMs) are employed as lightweight plugins to capture environment-specific knowledge, facilitating efficient environment-specific adaptation of LAMs. Subsequently, we instantiate this framework for CSI feedback tasks, and develop a large and small collaboration framework for CSI feedback, referred to as LASCO. LASCO operates by letting the base LAM produce an initial CSI reconstruction, learning the environment-induced reconstruction shift through a reference SAM and a proxy SAM, and transferring this shift back to the LAM. To further enhance adaptability, we introduce elastic-LASCO (E-LASCO), which augments LASCO with learnable collaboration coefficients that control the contribution of LAMs and SAMs across different environments. Numerical results demonstrate that LASCO and E-LASCO enables LAMs to achieve environment-specific performance gains with significantly reduced training costs, lower data collection requirements, and faster adaptation speed.

cs.IT↗

arXiv2025

ESS: An Offload-Centric Latent-Cache Management Architecture for DeepSeek-V3.2-Exp

DeepSeek-V3.2-Exp introduces a sparse attention mechanism that significantly reduces inference latency in long-context scenarios. Although the overall throughput has improved greatly, the Decode-stage of PD disaggregation remains to be a major bottleneck. This bottleneck primarily stems from the conflict between linear growth of Latent-Cache with sequence length and the limited GPU memory capacity, which constrains the feasible batch-size and thereby suppresses Decode-stage throughput. To address this challenge, we propose ESS (Extended Sparse Server), an offload-centric system design tailored for DeepSeek-V3.2-Exp. ESS selectively offloads Latent-Cache to CPU memory while preserving latency-critical components on GPU. By freeing up GPU memory, ESS effectively decoupling batch-size scaling from GPU memory constraints. This design significantly improves Decode-stage throughput, thereby reducing deployment costs in real-world settings. Our high-fidelity simulations show that ESS delivers 69.4\% throughput improvement at 32K context length and up to 123\% throughput improvement at 128K, demonstrating its effectiveness for large-context inference workloads. These results highlight ESS as a practical and scalable solution for long-context LLM serving.

cs.DC↗

arXiv2025

Financial Fraud Identification and Interpretability Study for Listed Companies Based on Convolutional Neural Network

Since the emergence of joint-stock companies, financial fraud by listed firms has repeatedly undermined capital markets. Fraud is difficult to detect because of covert tactics and the high labor and time costs of audits. Traditional statistical models are interpretable but struggle with nonlinear feature interactions, while machine learning models are powerful but often opaque. In addition, most existing methods judge fraud only for the current year based on current year data, limiting timeliness. This paper proposes a financial fraud detection framework for Chinese A-share listed companies based on convolutional neural networks (CNNs). We design a feature engineering scheme that transforms firm-year panel data into image like representations, enabling the CNN to capture cross-sectional and temporal patterns and to predict fraud in advance. Experiments show that the CNN outperforms logistic regression and LightGBM in accuracy, robustness, and early-warning performance, and that proper tuning of the classification threshold is crucial in high-risk settings. To address interpretability, we analyze the model along the dimensions of entity, feature, and time using local explanation techniques. We find that solvency, ratio structure, governance structure, and internal control are general predictors of fraud, while environmental indicators matter mainly in high-pollution industries. Non-fraud firms share stable feature patterns, whereas fraud firms exhibit heterogeneous patterns concentrated in short time windows. A case study of Guanong Shares in 2022 shows that cash flow analysis, social responsibility, governance structure, and per-share indicators are the main drivers of the model's fraud prediction, consistent with the company's documented misconduct.

cs.LG↗