arXiv subjects

Xin Zheng

Publications and source records attributed to Xin Zheng.

At least 37 records · Page 2Linked to original sources

arXiv2026

Momentum LMS Theory beyond Stationarity: Stability, Tracking, and Regret

In large-scale data processing scenarios, data often arrive in sequential streams generated by complex systems that exhibit drifting distributions and time-varying system parameters. This nonstationarity challenges theoretical analysis, as it violates classical assumptions of i.i.d. (independent and identically distributed) samples, necessitating algorithms capable of real-time updates without expensive retraining. An effective approach should process each sample in a single pass, while maintaining computational and memory complexities independent of the data stream length. Motivated by these challenges, this paper investigates the Momentum Least Mean Squares (MLMS) algorithm as an adaptive identification tool, leveraging its computational simplicity and online processing capabilities. Theoretically, we derive tracking performance and regret bounds for the MLMS in time-varying stochastic linear systems under various practical conditions. Unlike classical LMS, whose stability can be characterized by first-order random vector difference equations, MLMS introduces an additional dynamical state due to momentum, leading to second-order time-varying random vector difference equations whose stability analysis hinges on more complicated products of random matrices, which poses a substantially challenging problem to resolve. Experiments on synthetic and real-world data streams demonstrate that MLMS achieves rapid adaptation and robust tracking, in agreement with our theoretical results especially in nonstationary settings, highlighting its promise for modern streaming and online learning applications.

cs.LG↗

arXiv2026

Apple Intelligence Foundation Language Models

We present foundation language models developed to power Apple Intelligence features, including a ~3 billion parameter model designed to run efficiently on devices and a large server-based language model designed for Private Cloud Compute. These models are designed to perform a wide range of tasks efficiently, accurately, and responsibly. This report describes the model architecture, the data used to train the model, the training process, how the models are optimized for inference, and the evaluation results. We highlight our focus on Responsible AI and how the principles are applied throughout the model development.

cs.AI↗

arXiv2026

ST-TGExplainer: Disentangling Stability and Transition Patterns for Temporal GNN Interpretability

Temporal graph neural networks (TGNNs) have gained significant traction for solving real-world temporal graph tasks. However, their interpretability remains limited, as most TGNNs fail to identify which historical interactions most influence a given prediction. Despite promising progress on interpretable TGNNs, existing methods predominantly focus on previously seen historical interactions, which we term stability patterns, while overlooking newly emerging first-time interactions, which we term transition patterns. Both types of patterns are essential for faithful temporal explanations. To address this limitation, we propose ST-TGExplainer, a self-explainable TGNN that disentangles Stability and Transition patterns in temporal graphs for a more faithful Temporal GNN Explainer. Guided by a disentangled information bottleneck objective, ST-TGExplainer learns a compact explanatory subgraph that remains predictive of the event label while explicitly suppressing label-conditioned redundancy between stability and transition patterns. Extensive experiments demonstrate that ST-TGExplainer achieves strong predictive performance and yields more faithful explanations. Code is available at https://github.com/hjchen-hdu/ST-TGExplainer.

cs.LG↗

arXiv2026

ReCoG: Relational and Compact Context Graph Learning for Few-shot Molecular Property Prediction

Few-shot molecular property prediction (FSMPP) is essential in drug discovery and materials design, where high-quality labeled data are often scarce and expensive to obtain. Despite the promising performance of existing methods, especially context-aware methods, they still face two-fold severe challenges with \textit{insufficient structural context modeling} \& \textit{redundant auxiliary context learning}, leading to inadequate context graph exploration and ineffective information utilization for effective molecule representation learning. To address these, in this paper, we propose a novel framework by learning on \textbf{\underline{Re}}lational and \textbf{\underline{C}}ompact c\textbf{\underline{o}}ntext \textbf{\underline{G}}raph, named \textbf{\method}, to comprehensively exploit the context graph for expressive molecular property prediction. Specifically, the proposed \method contains two core modules: a \textbf{(1) cross-property relational learning module} to better model the structural and relational context information, and a \textbf{(2) context graph information bottleneck module} to adaptively suppress irrelevant auxiliary signals for compact context information utilization, followed by a detailed theoretical demonstration regarding the importance of joint relational and compact knowledge extraction in context graphs.

cs.CE↗

arXiv2026

Bridging Sequence and Graph Structure for Epigenetic Age Prediction

Epigenetic clocks based on DNA methylation have emerged as powerful tools for estimating biological age, with broad applications in aging research, age-related disease studies, and longevity science. Despite advances across machine learning approaches to epigenetic age prediction, spanning penalised linear regression, deep feedforward networks, residual architectures, and graph neural networks, no existing method jointly models co-methylation graph structure and site-specific DNA sequence context within a unified framework. We propose a unified sequence--graph integration framework for epigenetic age prediction that addresses this gap, integrating eight-dimensional DNA sequence statistical features through a lightweight gated modulation mechanism that adaptively scales each site's methylation signal according to its sequence-determined biological relevance prior to graph convolution. Evaluated on 3,707 blood methylation samples against a comprehensive set of baselines, our method achieves a test MAE of 3.149 years, a 12.8\% improvement over the strongest graph-based baseline. Biologically informed statistical features outperform CNN-based sequence encoding, demonstrating that handcrafted sequence features are more effective than end-to-end learned representations in this data regime. Post-hoc interpretability analysis identifies CpG density and local adenine frequency as features with age-dependent importance shifts, consistent with known mechanisms of age-related hypermethylation at CpG-dense promoter regions. Our code is at https://github.com/yaoli2022/graphage-seq.

cs.AI↗

arXiv2026

Tight Lower Bound for Approximating Parametrized Maximum Likelihood Decoding under ETH

We present a simple deterministic reduction which, assuming the Exponential Time Hypothesis ($\mathsf{ETH}$), yields tight lower bounds for approximating the parameterized Maximum Likelihood Decoding problem ($\mathsf{MLD}$) and the parameterized Nearest Codeword Problem ($\mathsf{NCP}$) within some fixed constant factor. Our starting point is the ETH-based exponential-time hardness of $(c,s)$-Gap-$\mathsf{MAXLIN}$ established in [BHI+24]. We transform a $(c,s)$-Gap-$\mathsf{MAXLIN}$ instance into an instance of $γ$-Gap $k$-$\mathsf{MLD}$ via a novel combinatorial object that we call a cover family. We provide both a randomized construction of the required cover families and a subsequent derandomization. Prior to our work, $n^{Ω(k)}$ hardness for constant-factor approximation was only shown under the randomized Gap Exponential Time Hypothesis Gap-$\mathsf{ETH}$ [Man20], which is a much stronger assumption than $\mathsf{ETH}$. Under $\mathsf{ETH}$, the strongest known lower bound was $n^{Ω(k/\operatorname{poly} \log k)}$ due to [BKM25]. Unlike previous approaches that rely on reductions from the hardness of approximating $2$-$\mathsf{CSP}$, our reduction provides a more direct and conceptually simpler route to achieving the optimal lower bounds.

cs.CC↗

arXiv2026

ScaleBox: Enabling High-Fidelity and Scalable Code Verification for Large Language Models

Code sandboxes have emerged as a critical infrastructure for advancing the coding capabilities of large language models, providing verifiable feedback for both RL training and evaluation. However, existing systems fail to provide accurate verification and efficiency under high-concurrency workloads. We present ScaleBox, a high-fidelity and scalable system designed to address these limitations in large-scale code training. ScaleBox introduces automated special-judge generation and management, fine-grained parallel execution across test cases with seamless multi-node coordination, and a configuration-driven evaluation suite for reproducible benchmarking. A series of experiments demonstrates that ScaleBox significantly enhances code verification accuracy and efficiency. Our further RLVR experiments show that ScaleBox substantially improves both performance on LiveCodeBench and training stability, significantly outperforming heuristic-matching baselines. By providing a reliable and high-throughput infrastructure, ScaleBox facilitates more effective research and development in large-scale code training.

cs.SE↗

arXiv2026

Graph Neural Networks for Graphs with Heterophily: A Survey

Recent years have witnessed fast developments of graph neural networks (GNNs) that have benefited myriad graph analytic tasks and applications. Most GNNs rely on the homophily assumption that nodes belonging to the same class are more likely to be connected. However, as a ubiquitous graph property in numerous real-world scenarios, heterophily, i.e., nodes with different labels tend to be linked, significantly limits the performance of tailor-made homophilic GNNs. Hence, GNNs for heterophilic graphs are gaining increasing research attention to enhance graph learning with heterophily. In this paper, we provide a comprehensive review of GNNs for heterophilic graphs. Specifically, we propose a systematic taxonomy that governs existing heterophilic GNN models, along with general summaries and detailed analyses. Furthermore, we discuss the relationship between heterophily and various graph research domains, aiming to facilitate the development of more effective GNNs across a spectrum of practical applications and learning tasks in the graph research community. In the end, we point out potential directions to advance and inspire future research and applications on heterophilic graph learning with GNNs.

cs.LG↗

arXiv2026

GoAgent: Group-of-Agents Communication Topology Generation for LLM-based Multi-Agent Systems

Large language model (LLM)-based multi-agent systems (MAS) have demonstrated exceptional capabilities in solving complex tasks, yet their effectiveness depends heavily on the underlying communication topology that coordinates agent interactions. Within these systems, successful problem-solving often necessitates task-specific group structures to divide and conquer subtasks. However, most existing approaches generate communication topologies in a node-centric manner, leaving group structures to emerge implicitly from local connectivity decisions rather than modeling them explicitly, often leading to suboptimal coordination and unnecessary communication overhead. To address this limitation, we propose GoAgent (Group-of-Agents), a communication topology generation method that explicitly treats collaborative groups as the atomic units of MAS construction. Specifically, GoAgent first enumerates task-relevant candidate groups through an LLM and then autoregressively selects and connects these groups as atomic units to construct the final communication graph, jointly capturing intra-group cohesion and inter-group coordination. To mitigate communication redundancy and noise propagation inherent in expanding topologies, we further introduce a conditional information bottleneck (CIB) objective that compresses inter-group communication, preserving task-relevant signals while filtering out redundant historical noise. Extensive experiments on six benchmarks demonstrate the state-of-the-art performance of GoAgent with 93.84% average accuracy while reducing token consumption by about 17%.

cs.LG↗

arXiv2026

Spiking Graph Predictive Coding for Reliable OOD Generalization

Graphs provide a powerful basis for modeling Web-based relational data, with expressive GNNs to support the effective learning in dynamic web environments. However, real-world deployment is hindered by pervasive out-of-distribution (OOD) shifts, where evolving user activity and changing content semantics alter feature distributions and labeling criteria. These shifts often lead to unstable or overconfident predictions, undermining the trustworthiness required for Web4Good applications. Achieving reliable OOD generalization demands principled and interpretable uncertainty estimation; however, existing methods are largely post-hoc, insensitive to distribution shifts, and unable to explain where uncertainty arises especially in high-stakes settings. To address these limitations, we introduce SpIking GrapH predicTive coding (SIGHT), an uncertainty-aware plug-in graph learning module for reliable OOD Generalization. SIGHT performs iterative, error-driven correction over spiking graph states, enabling models to expose internal mismatch signals that reveal where predictions become unreliable. Across multiple graph benchmarks and diverse OOD scenarios, SIGHT consistently enhances predictive accuracy, uncertainty estimation, and interpretability when integrated with GNNs.

cs.LG↗

arXiv2026

Gradient-Based Adaptive Prediction and Control for Nonlinear Dynamical Systems

This paper investigates gradient-based adaptive prediction and control for nonlinear stochastic dynamical systems under a weak convexity condition on the prediction-based loss. This condition accommodates a broad range of nonlinear models in control and machine learning such as saturation functions, sigmoid, ReLU and tanh activation functions, and standard classification models. Without requiring any persistent excitation of the data, we establish global convergence of the proposed adaptive predictor and derive explicit rates for its asymptotic performance. Furthermore, under a classical nonlinear minimum-phase condition and with a linear growth bound on the nonlinearities, we establish the convergence rate of the resulting closed-loop control error. Finally, we demonstrate the effectiveness of the proposed adaptive prediction algorithm on a real-world judicial sentencing dataset. The adaptive control performance will also be evaluated via a numerical simulation.

eess.SY↗

arXiv2025

Judicial Sentencing Prediction Based on Hybrid Models and Two-Stage Learning Algorithms

The investigation of legal judgment prediction (LJP), such as sentencing prediction, has attracted broad attention for its potential to promote judicial fairness, making the accuracy and reliability of its computation result an increasingly critical concern. In view of this, we present a new sentencing model that shares both legal logic interpretability and strong prediction capability by introducing a two-stage learning algorithm. Specifically, we first construct a hybrid model that synthesizes a mechanism model based on the main factors for sentencing with a neural network modeling possible uncertain features. We then propose a two-stage learning algorithm: First, an adaptive stochastic gradient (ASG) algorithm is used to get good estimates for the unknown parameters in the mechanistic component of the hybrid model. Then, the Adam optimizer tunes all parameters to enhance the predictive performance of the entire hybrid model. The asymptotic convergence of the ASG-based adaptive predictor is established without requiring any excitation data conditions, thereby providing a good initial parameter estimate for prediction. Based on this, the fast-converging Adam optimizer further refines the parameters to enhance overall prediction accuracy. Experiments on a real-world dataset of intentional injury cases in China show that our new hybrid model combined with our two-stage ASG-Adam algorithm, outperforms the existing related methods in sentencing prediction performance, including those based on neural networks and saturated mechanism models.

math.DS↗

arXiv2025

Graph Neural Architecture Search with GPT-4

Graph Neural Architecture Search (GNAS) has shown promising results in finding the best graph neural network architecture on a given graph dataset. However, existing GNAS methods still require intensive human labor and rich domain knowledge when designing the search space and search strategy. To this end, we integrate Large Language Models (LLMs) into GNAS and present a new GNAS model based on LLMs (GNAS-LLM for short). The basic idea of GNAS-LLM is to design a new class of GNAS prompts for LLMs to guide LLMs towards understanding the generative task of graph neural architectures. The prompts consist of descriptions of the search space, search strategy, and search feedback of GNAS. By iteratively running LLMs with the prompts, GNAS-LLM generates more accurate graph neural network architectures with fast convergence. Experimental results show that GNAS-LLM outperforms the state-of-the-art GNAS methods on four benchmark graph datasets, with an average improvement of 0.7% on the validation sets and 0.3% on the test sets. Besides, GNAS-LLM achieves an average improvement of 1.0% on the test sets based on the search space from AutoGEL.

cs.LG↗

arXiv2025

Enhancing optical lattice clock coherence times with erasure conversion

Increasing coherent interrogation times is central to advancing the precision of optical clocks. Synchronous differential optical clock comparisons have now demonstrated atomic coherence times that far exceed the coherence time of the clock laser. While atom coherence times are then primarily limited by errors induced by lattice Raman scattering, excited clock state radiative decay, and broadening from two-body collisions, many of these errors take the atoms out of the clock transition subspace, and can therefore be converted into "erasure" errors if the appropriate readout scheme is employed. Here we experimentally demonstrate a hyperfine-resolved readout technique for ${}^{87}$Sr optical lattice clocks that mitigates decoherence from Raman scattering induced by the lattice as well as radiative decay. By employing hyperfine-resolved readout in synchronous differential comparisons between ${}^{87}$Sr ensembles with both Ramsey and spin echo spectroscopy sequences, we achieve enhanced atomic coherence times exceeding 100 s and 150 s, respectively, enabling longer coherent measurements without a reduction in performance. We anticipate that this hyperfine-resolved readout technique will benefit applications of state-of-the-art optical lattice clock comparisons in which the coherence times are constrained by Raman scattering or radiative decay.

physics.atom-ph↗

arXiv2025

Few-shot Molecular Property Prediction: A Survey

AI-assisted molecular property prediction has become a promising technique in early-stage drug discovery and materials design in recent years. However, due to high-cost and complex wet-lab experiments, real-world molecules usually experience the issue of scarce annotations, leading to limited labeled data for effective supervised AI model learning. In light of this, few-shot molecular property prediction (FSMPP) has emerged as an expressive paradigm that enables learning from only a few labeled examples. Despite rapidly growing attention, existing FSMPP studies remain fragmented, without a coherent framework to capture methodological advances and domain-specific challenges. In this work, we present the first comprehensive and systematic survey of few-shot molecular property prediction. We begin by analyzing the few-shot phenomenon in molecular datasets and highlighting two core challenges: (1) cross-property generalization under distribution shifts, where each task corresponding to each property, may follow a different data distribution or even be inherently weakly related to others from a biochemical perspective, requiring the model to transfer knowledge across heterogeneous prediction tasks, and (2) cross-molecule generalization under structural heterogeneity, where molecules involved in different or same properties may exhibit significant structural diversity, making model difficult to achieve generalization. Then, we introduce a unified taxonomy that organizes existing methods into data, model, and learning paradigm levels, reflecting their strategies for extracting knowledge from scarce supervision in few-shot molecular property prediction. Next, we compare representative methods, summarize benchmark datasets and evaluation protocols. In the end, we identify key trends and future directions for advancing the continued research on FSMPP.

cs.CE↗

arXiv2025

Test-time GNN Model Evaluation on Dynamic Graphs

Dynamic graph neural networks (DGNNs) have emerged as a leading paradigm for learning from dynamic graphs, which are commonly used to model real-world systems and applications. However, due to the evolving nature of dynamic graph data distributions over time, well-trained DGNNs often face significant performance uncertainty when inferring on unseen and unlabeled test graphs in practical deployment. In this case, evaluating the performance of deployed DGNNs at test time is crucial to determine whether a well-trained DGNN is suited for inference on an unseen dynamic test graph. In this work, we introduce a new research problem: DGNN model evaluation, which aims to assess the performance of a specific DGNN model trained on observed dynamic graphs by estimating its performance on unseen dynamic graphs during test time. Specifically, we propose a Dynamic Graph neural network Evaluator, dubbed DyGEval, to address this new problem. The proposed DyGEval involves a two-stage framework: (1) test-time dynamic graph simulation, which captures the training-test distributional differences as supervision signals and trains an evaluator; and (2) DyGEval development and training, which accurately estimates the performance of the well-trained DGNN model on the test-time dynamic graphs. Extensive experiments demonstrate that the proposed DyGEval serves as an effective evaluator for assessing various DGNN backbones across different dynamic graphs under distribution shifts.

cs.LG↗

arXiv2025

An S-matrix Formalism for the Nonclassical Optical Response of Plasmonic Sphere Aggregates

A computational method for the scattering of light by multiple nonclassical plasmonic nanospheres, each of which has multiple (non-)concentric dielectric or metallic layers, is presented. The electromagnetic (EM) response of the free electrons in the metals is described by three popular mesoscopic models: the nonlocal hydrodynamic Drude model (NLHDM) and its diffusive variant, namely the generalized nonlocal optical response (GNOR) model, as well as the surface response model (SRM). The main equation behind the method is set up by detailing the evaluation of the S-matrix for each individual spherical interface and the interactions amongst the interfaces. The algorithm is numerically validated by comparing with an in-house boundary element method (BEM) solver for a spherical NP with two smaller embedded spheres, and physically checked on a trimer configuration, where the responses from the NLHDM and SRM are contrasted with the local response model (LRM). In both cases a very good agreement is seen regarding frequency shifts and field enhancements.

physics.comp-ph↗

arXiv2025

OASIS: Harnessing Diffusion Adversarial Network for Ocean Salinity Imputation using Sparse Drifter Trajectories

Ocean salinity plays a vital role in circulation, climate, and marine ecosystems, yet its measurement is often sparse, irregular, and noisy, especially in drifter-based datasets. Traditional approaches, such as remote sensing and optimal interpolation, rely on linearity and stationarity, and are limited by cloud cover, sensor drift, and low satellite revisit rates. While machine learning models offer flexibility, they often fail under severe sparsity and lack principled ways to incorporate physical covariates without specialized sensors. In this paper, we introduce the OceAn Salinity Imputation System (OASIS), a novel diffusion adversarial framework designed to address these challenges.

cs.LG↗