arXiv subjects

Jingwei Zhang

Publications and source records attributed to Jingwei Zhang.

At least 19 recordsLinked to original sources

arXiv2026

NeuPAT: Neuron-aware Plasticity Allocation Tuning for Language-Preserving MLLMs

Multimodal expansion of large language models (LLMs) enables new perceptual capabilities but often compromises the language intelligence acquired during pretraining. In this work, we investigate this phenomenon from the perspective of internal adaptation dynamics and discover that neurons in pretrained LLMs exhibit heterogeneous plasticity during multimodal learning: some neurons are critical for preserving language capabilities, while others are more adaptive to multimodal knowledge. Based on this insight, we propose NeuPAT (Neuron-aware Plasticity Allocation Tuning), a lightweight and architecture-agnostic framework that allocates neuron-wise update constraints during multimodal instruction tuning. NeuPAT uses a small-scale probing stage to estimate neuron adaptation patterns and selectively protects language-sensitive neurons while promoting multimodal adaptation through more plastic neurons. Experiments across diverse LLM families demonstrate that NeuPAT recovers 94.5\% of the language capability degradation caused by vanilla tuning on 11 language benchmarks while maintaining comparable multimodal performance, providing an effective approach for capability-preserving multimodal expansion.

cs.CL↗

arXiv2026

Trajectory-Guided Forget-Recover Network for Continual LLM Unlearning

Machine unlearning aims to eliminate the influence of sensitive data on a model. In the real world, unlearning requests arrive continually, which gives rise to two challenges. First, an unlearning intervention may redistribute target-related computation across remaining pathways, allowing previously forgotten knowledge to re-emerge. Second, repeated unlearning interventions may progressively reduce the model capacity needed to preserve retained utility. To address these challenges, we propose the Trajectory-guided Forget-Recover Network (TFR-Net). TFR-Net tracks channel-level risk across requests. It separates persistent target-related channels from transient hotspots and suppresses only the persistent ones. TFR-Net also recovers model capacity by reactivating dormant channels. These channels make strong contributions to retained utility and show low current and historical forget risk. The recovery is accepted only when retained-utility degradation remains within a predefined tolerance. Experiments on four datasets show that TFR-Net consistently achieves a more favorable trade-off between unlearning effectiveness and retained utility than representative baselines.

cs.LG↗

arXiv2026

Exploring More to Solve More: Boosting Diversity in Text Diffusion Models via Entropy-Based Guidance

Although diffusion models have revolutionized continuous domains like image synthesis through high quality generations and controllable guidance mechanisms, bringing this controllability to the discrete, sequential nature of text remains an open challenge. Meanwhile, current sampling strategies and guidance methods adjust token likelihoods without capturing the broader semantic landscape, leading to a suboptimal balance between fidelity and diversity. In this work, we introduce a novel training-free Semantic-Aware Kernel Entropy (SAKE) guidance method. Our method computes the order-2 R\'enyi entropy over a kernel Gram matrix that captures both cross-token semantic interactions and relative token positions. By linearizing this objective in the embedding space, we derive a tractable guidance signal that dynamically adjusts the sampling distribution, flattening it to encourage exploration during redundancy and sharpening it for fidelity when diverse. Empirical experiments demonstrate that our approach achieves a superior Pareto frontier between fidelity and diversity, and improves multi-sample performance on reasoning-intensive tasks, such as code and mathematics generation, compared to temperature scaling and discrete guidance baselines.

cs.CL↗

arXiv2026

Self-Dual Cyclic Codes with Improved Minimum Distance Estimates via Extending the Chen-Ding Construction

Self-dual cyclic codes have garnered significant interest owing to their rich algebraic structures and wide-ranging applicability. Their construction and the establishment of lower bounds on their minimum distances are fundamental problems in coding theory. Chen and Ding laid an important foundation for the construction of self-dual cyclic codes in the case where the multiplicative order of $q$ module $n$, denoted by $\operatorname{ord}_n(q)$, is odd. Building on their work, we extend the investigation to the case of even order $\operatorname{ord}_n(q)$ and demonstrate that the minimum distances of the resulting self-dual cyclic codes satisfy square-root lower bounds. By examining the consecutive zero segments in the defining set of the dual code, we determine the exact parameters of Euclidean self-dual cyclic codes with even $\operatorname{ord}_n(q)$ and Hermitian self-dual cyclic codes with odd $\operatorname{ord}_n(q)$. Furthermore, for Euclidean self-dual cyclic codes with odd $\operatorname{ord}_n(q)$ and Hermitian self-dual cyclic codes with even $\operatorname{ord}_n(q)$, we introduce a refined parameter selection that leads to larger minimum distances with the same code length and dimension. This approach also yields tighter lower bounds for several families of self-dual cyclic codes. This work enriches the theory of self-dual cyclic codes and offers new insights into estimating lower bounds on their minimum distances.

cs.IT↗

arXiv2026

Mixture-of-Experts Transformer for Automatic Modulation Recognition

Automatic Modulation Recognition (AMR) is a key enabling technology for cognitive radio and intelligent spectrum management in next-generation wireless systems. However, current deep learning-based AMR methods predominantly rely on static multi-scale fusion strategies, which lack the flexibility to adapt to the highly dynamic temporal variations of modulation signals. To address this limitation, we propose MoEformer, an adaptive Multi-Scale Mixture-of-Experts Transformer network that directly processes I/Q signals to preserve their temporal and phase structures. Specifically, MoEformer constructs multi scale expert views through temporal resampling, employs an input-dependent gating mechanism for dynamic expert fusion, and integrates Rotary Position Embeddings (RoPE) within Transformer encoders to capture both local and global tem poral dependencies. Comprehensive evaluations on three widely adopted benchmarks (RadioML2016.10a, RadioML2016.10b, and RadioML2018.01A) demonstrate that MoEformer outperforms the competitive baselines, achieving superior average recognition accuracies of 63.74%, 66.24%, and 64.22%, respectively. In addition, the proposed method strikes an optimal trade-off between recognition performance and model complexity.

eess.SP↗

arXiv2026

U-CAN: Utility-Aware Contrastive Attenuation for Efficient Unlearning in Generative Recommendation

Generative Recommendation (GenRec) typically leverages Large Language Models (LLMs) to redefine personalization as an instruction-driven sequence generation task. However, fine-tuning on user logs inadvertently encodes sensitive attributes into model parameters, raising critical privacy concerns. Existing Machine Unlearning (MU) techniques struggle to navigate this tension due to the Polysemy Dilemma, where neurons superimpose sensitive data with general reasoning patterns, leading to catastrophic utility loss under traditional gradient or pruning methods. To address this, we propose Utility-aware Contrastive AttenuatioN (U-CAN), a precision unlearning framework that operates on low-rank adapters. U-CAN quantifies risk by contrasting activations and focuses on neurons with asymmetric responses that are highly sensitive to the forgetting set but suppressed on the retention set. To safeguard performance, we introduce a utility-aware calibration mechanism that combines weight magnitudes with retention-set activation norms, assigning higher utility scores to dimensions that contribute strongly to retention performance. Unlike binary pruning, which often fragments network structure, U-CAN develop adaptive soft attenuation with a differentiable decay function to selectively down-scale high-risk parameters on LoRA adapters, suppressing sensitive retrieval pathways and preserving the topological connectivity of reasoning circuits. Experiments on two public datasets across seven metrics demonstrate that U-CAN achieves strong privacy forgetting, utility retention, and computational efficiency.

cs.LG↗

arXiv2026

BamaER: A Behavior-Aware Memory-Augmented Model for Exercise Recommendation

Exercise recommendation focuses on personalized exercise selection conditioned on students' learning history, personal interests, and other individualized characteristics. Despite notable progress, most existing methods represent student learning solely as exercise sequences, overlooking rich behavioral interaction information. This limited representation often leads to biased and unreliable estimates of learning progress. Moreover, fixed-length sequence segmentation limits the incorporation of early learning experiences, thereby hindering the modeling of long-term dependencies and the accurate estimation of knowledge mastery. To address these limitations, we propose BamaER, a Behavior-aware memory-augmented Exercise Recommendation framework that comprises three core modules: (i) the learning progress prediction module that captures heterogeneous student interaction behaviors via a tri-directional hybrid encoding scheme; (ii) the memory-augmented knowledge tracing module that maintains a dynamic memory matrix to jointly model historical and current knowledge states for robust mastery estimation; and (iii) the exercise filtering module that formulates candidate selection as a diversity-aware optimization problem, solved via the Hippopotamus Optimization Algorithm to reduce redundancy and improve recommendation coverage. Experiments on five real-world educational datasets show that BamaER consistently outperforms state-of-the-art baselines across a range of evaluation metrics.

cs.LG↗

arXiv2025

A Distributed Hierarchical Spatio-Temporal Edge-Enhanced Graph Neural Network for City-Scale Dynamic Logistics Routing

City-scale logistics routing has become increasingly challenging as metropolitan road networks grow to tens of millions of edges and traffic conditions evolve rapidly under high-volume mobility demands. Conventional centralized routing algorithms and monolithic graph neural network (GNN) models suffer from limited scalability, high latency, and poor real-time adaptability, which restricts their effectiveness in large urban logistics systems. To address these challenges, this paper proposes a Distributed Hierarchical Spatio-Temporal Edge-Enhanced Graph Neural Network (HSTE-GNN) for dynamic routing over ultra-large road networks. The framework partitions the city-scale graph into regional subgraphs processed in parallel across distributed computing nodes, enabling efficient learning of localized traffic dynamics. Within each region, an edge-enhanced spatio-temporal module jointly models node states, dynamic edge attributes, and short-term temporal dependencies. A hierarchical coordination layer further aggregates cross-region representations through an asynchronous parameter-server mechanism, ensuring global routing coherence under high-frequency traffic updates. This distributed hierarchical design balances local responsiveness with global consistency, significantly improving scalability and inference efficiency. Experiments on real-world large-scale traffic datasets from Beijing and New York demonstrate that HSTE-GNN outperforms strong spatio-temporal baselines such as ST-GRAPH, achieving 34.9% lower routing delay, 14.7% lower MAPE, and 11.8% lower RMSE, while improving global route consistency by 7.3%. These results confirm that the proposed framework provides a scalable, adaptive, and efficient solution for next-generation intelligent transportation systems and large-scale logistics platforms.

eess.SY↗

arXiv2025

Fault Diagnosis and Quantification for Photovoltaic Arrays based on Differentiable Physical Models

Accurate fault diagnosis and quantification are essential for the reliable operation and intelligent maintenance of photovoltaic (PV) arrays. However, existing fault quantification methods often suffer from limited efficiency and interpretability. To address these challenges, this paper proposes a novel fault quantification approach for PV strings based on a differentiable fast fault simulation model (DFFSM). The proposed DFFSM accurately models I-V characteristics under multiple faults and provides analytical gradients with respect to fault parameters. Leveraging this property, a gradient-based fault parameters identification (GFPI) method using the Adahessian optimizer is developed to efficiently quantify partial shading, short-circuit, and series-resistance degradation. Experimental results on both simulated and measured I-V curves demonstrate that the proposed GFPI achieves high quantification accuracy across different faults, with the I-V reconstruction error below 3%, confirming the feasibility and effectiveness of the application of differentiable physical simulators for PV system fault diagnosis.

cs.LG↗

arXiv2025

VST-Pose: A Velocity-Integrated Spatiotem-poral Attention Network for Human WiFi Pose Estimation

WiFi-based human pose estimation has emerged as a promising non-visual alternative approaches due to its pene-trability and privacy advantages. This paper presents VST-Pose, a novel deep learning framework for accurate and continuous pose estimation using WiFi channel state information. The proposed method introduces ViSTA-Former, a spatiotemporal attention backbone with dual-stream architecture that adopts a dual-stream architecture to separately capture temporal dependencies and structural relationships among body joints. To enhance sensitivity to subtle human motions, a velocity modeling branch is integrated into the framework, which learns short-term keypoint dis-placement patterns and improves fine-grained motion representation. We construct a 2D pose dataset specifically designed for smart home care scenarios and demonstrate that our method achieves 92.2% accuracy on the PCK@50 metric, outperforming existing methods by 8.3% in PCK@50 on the self-collected dataset. Further evaluation on the public MMFi dataset confirms the model's robustness and effectiveness in 3D pose estimation tasks. The proposed system provides a reliable and privacy-aware solution for continuous human motion analysis in indoor environments. Our codes are available in https://github.com/CarmenQing/VST-Pose.

cs.CV↗

arXiv2025

LBMamba: Locally Bi-directional Mamba

Mamba, a State Space Model (SSM) that accelerates training by recasting recurrence as a parallel scan, has recently emerged as a linearly-scaling alternative to self-attention. Because of its unidirectional nature, each state in Mamba only has information of its previous states and is blind to states after. Current Mamba-based computer-vision methods typically overcome this by augmenting Mamba's global forward scan with a global backward scan, forming a bi-directional scan to restore a full receptive field. However, this operation doubles the computational load, eroding much of the efficiency advantage that originally Mamba have. To eliminate this extra scans, we introduce LBMamba, a locally bi-directional SSM block that embeds a lightweight locally backward scan inside the forward scan and executes it in per-thread registers. Building on LBMamba, we present LBVim, a backbone that alternates scan directions every two layers to recover a global receptive field without extra backward sweeps. We validate our approach on both natural images and whole slide images (WSIs) and show that it constantly offers a superior performance-throughput trade-off. Under the same throughput, LBVim achieves 0.8% to 1.6% higher top-1 accuracy on the ImageNet-1K classification dataset, 0.6% to 2.7% higher mIoU on the ADE20K semantic segmentation dataset, 0.9% higher APb and 1.1% higher APm on the COCO detection dataset. Our method also boosts the accuracy of four SOTA Mamba models, namely VMamba, LocalVim, PlainMamba and Adventurer, by 0.5% to 3.4%. We integrate LBMamba into the SOTA pathology multiple instance learning (MIL) model, MambaMIL, which is unidirectional. Experiments on 3 public WSI classification datasets show that our method achieves a relative improvement of up to 3.06% better AUC, 3.39% better F1, 1.67% better accuracy. Our code is available at https://github.com/cvlab-stonybrook/LBMamba.

cs.CV↗

arXiv2025

When Kernels Multiply, Clusters Unify: Fusing Embeddings with the Kronecker Product

State-of-the-art embeddings often capture distinct yet complementary discriminative features: For instance, one image embedding model may excel at distinguishing fine-grained textures, while another focuses on object-level structure. Motivated by this observation, we propose a principled approach to fuse such complementary representations through kernel multiplication. Multiplying the kernel similarity functions of two embeddings allows their discriminative structures to interact, producing a fused representation whose kernel encodes the union of the clusters identified by each parent embedding. This formulation also provides a natural way to construct joint kernels for paired multi-modal data (e.g., image-text tuples), where the product of modality-specific kernels inherits structure from both domains. We highlight that this kernel product is mathematically realized via the Kronecker product of the embedding feature maps, yielding our proposed KrossFuse framework for embedding fusion. To address the computational cost of the resulting high-dimensional Kronecker space, we further develop RP-KrossFuse, a scalable variant that leverages random projections for efficient approximation. As a key application, we use this framework to bridge the performance gap between cross-modal embeddings (e.g., CLIP, BLIP) and unimodal experts (e.g., DINOv2, E5). Experiments show that RP-KrossFuse effectively integrates these models, enhancing modality-specific performance while preserving cross-modal alignment. The project code is available at https://github.com/yokiwuuu/KrossFuse.

cs.LG↗

arXiv2025

GeoMaNO: Geometric Mamba Neural Operator for Partial Differential Equations

The neural operator (NO) framework has emerged as a powerful tool for solving partial differential equations (PDEs). Recent NOs are dominated by the Transformer architecture, which offers NOs the capability to capture long-range dependencies in PDE dynamics. However, existing Transformer-based NOs suffer from quadratic complexity, lack geometric rigor, and thus suffer from sub-optimal performance on regular grids. As a remedy, we propose the Geometric Mamba Neural Operator (GeoMaNO) framework, which empowers NOs with Mamba's modeling capability, linear complexity, plus geometric rigor. We evaluate GeoMaNO's performance on multiple standard and popularly employed PDE benchmarks, spanning from Darcy flow problems to Navier-Stokes problems. GeoMaNO improves existing baselines in solution operator approximation by as much as 58.9%.

cs.LG↗

arXiv2025

CATCH-FORM-ACTer: Compliance-Aware Tactile Control and Hybrid Deformation Regulation-Based Action Transformer for Viscoelastic Object Manipulation

Automating contact-rich manipulation of viscoelastic objects with rigid robots faces challenges including dynamic parameter mismatches, unstable contact oscillations, and spatiotemporal force-deformation coupling. In our prior work, a Compliance-Aware Tactile Control and Hybrid Deformation Regulation (CATCH-FORM-3D) strategy fulfills robust and effective manipulations of 3D viscoelastic objects, which combines a contact force-driven admittance outer loop and a PDE-stabilized inner loop, achieving sub-millimeter surface deformation accuracy. However, this strategy requires fine-tuning of object-specific parameters and task-specific calibrations, to bridge this gap, a CATCH-FORM-ACTer is proposed, by enhancing CATCH-FORM-3D with a framework of Action Chunking with Transformer (ACT). An intuitive teleoperation system performs Learning from Demonstration (LfD) to build up a long-horizon sensing, decision-making and execution sequences. Unlike conventional ACT methods focused solely on trajectory planning, our approach dynamically adjusts stiffness, damping, and diffusion parameters in real time during multi-phase manipulations, effectively imitating human-like force-deformation modulation. Experiments on single arm/bimanual robots in three tasks show better force fields patterns and thus 10%-20% higher success rates versus conventional methods, enabling precise, safe interactions for industrial, medical or household scenarios.

cs.RO↗

arXiv2025

Controllable and Continuous Quantum Phase Transitions in Intrinsic Magnetic Topological Insulator

The intrinsic magnetic topological material MnBi2Te4 has demonstrated great potential to investigate the interplay between topology and magnetism, which opens up new avenues for manipulating non-trivial electronic states and designing quantum devices. However, challenges and controversies remain due to its inevitable n-type antisite defects, hindering the experimental realization of intrinsic magnetic topological phenomena and rendering the precise control of topological phase transitions (TPTs) unachievable. Here, we study a candidate material family, Mn(1-x)GexBi2Te4, in which the heavy n-type doping features are strongly suppressed when the Ge content reaches 0.46, and multiple topological phases are well maintained with the surface Dirac point located near the Fermi level. Based on angle-resolved photoemission spectroscopy, transport measurements, and first-principles calculations, we reveal two magnetism-induced TPTs: the first is antiferromagnetic-ordering-induced transition from strong topological insulator to a magnetic topological insulator as revealed by gap opening of topological surface states; the second is external-magnetic-field-dependent transition from magnetic topological insulator to a Weyl semimetal with the gap reclosed. Our work paves the way for the realization of intrinsic magnetic topological states in MnBi2Te4 family and provides an ideal platform for achieving controllable and continuous TPTs towards future spintronic applications.

cond-mat.mtrl-sci↗

arXiv2025

SeizeIT2: Wearable Dataset Of Patients With Focal Epilepsy

The increasing technological advancements towards miniaturized physiological measuring devices have enabled continuous monitoring of epileptic patients outside of specialized environments. The large amounts of data that can be recorded with such devices holds significant potential for developing automated seizure detection frameworks. In this work, we present SeizeIT2, the first open dataset of wearable data recorded in patients with focal epilepsy. The dataset comprises more than 11,000 hours of multimodal data, including behind-the-ear electroencephalography, electrocardiography, electromyography and movement (accelerometer and gyroscope) data. The dataset contains 886 focal seizures recorded from 125 patients across five different European Epileptic Monitoring Centers. We present a suggestive training/validation split to propel the development of AI methodologies for seizure detection, as well as two benchmark approaches and evaluation metrics. The dataset can be accessed on OpenNeuro and is stored in Brain Imaging Data Structure (BIDS) format.

eess.SP↗

arXiv2025

Crop Yield Time-Series Data Prediction Based on Multiple Hybrid Machine Learning Models

Agriculture plays a crucial role in the global economy and social stability, and accurate crop yield prediction is essential for rational planting planning and decision-making. This study focuses on crop yield Time-Series Data prediction. Considering the crucial significance of agriculture in the global economy and social stability and the importance of accurate crop yield prediction for rational planting planning and decision-making, this research uses a dataset containing multiple crops, multiple regions, and data over many years to deeply explore the relationships between climatic factors (average rainfall, average temperature) and agricultural inputs (pesticide usage) and crop yield. Multiple hybrid machine learning models such as Linear Regression, Random Forest, Gradient Boost, XGBoost, KNN, Decision Tree, and Bagging Regressor are adopted for yield prediction. After evaluation, it is found that the Random Forest and Bagging Regressor models perform excellently in predicting crop yield with high accuracy and low error.As agricultural data becomes increasingly rich and time-series prediction techniques continue to evolve, the results of this study contribute to advancing the practical application of crop yield prediction in agricultural production management. The integration of time-series analysis allows for more dynamic, data-driven decision-making, enhancing the accuracy and reliability of crop yield forecasts over time.

cs.CY↗

arXiv2024

A Super-pixel-based Approach to the Stable Interpretation of Neural Networks

Saliency maps are widely used in the computer vision community for interpreting neural network classifiers. However, due to the randomness of training samples and optimization algorithms, the resulting saliency maps suffer from a significant level of stochasticity, making it difficult for domain experts to capture the intrinsic factors that influence the neural network's decision. In this work, we propose a novel pixel partitioning strategy to boost the stability and generalizability of gradient-based saliency maps. Through both theoretical analysis and numerical experiments, we demonstrate that the grouping of pixels reduces the variance of the saliency map and improves the generalization behavior of the interpretation method. Furthermore, we propose a sensible grouping strategy based on super-pixels which cluster pixels into groups that align well with the semantic meaning of the images. We perform several numerical experiments on CIFAR-10 and ImageNet. Our empirical results suggest that the super-pixel-based interpretation maps consistently improve the stability and quality over the pixel-based saliency maps.

cs.CV↗