arXiv subjects

Weiyao Zhang

Publications and source records attributed to Weiyao Zhang.

10 recordsLinked to original sources

arXiv2026

Unveiling Spectral Mechanisms in Training-Free LLM Text Detection

The rapid advancement of Large Language Models (LLMs) makes it increasingly difficult to distinguish human writing from machine-generated text. Training-free detection offers a scalable solution, yet common confidence-based metrics mainly measure average token probabilities and often miss the signal fluctuations that characterize human writing, which we call "generative vitality". Spectral analysis offers a way to capture this vitality, but its mechanism and practical boundaries remain underexplored. In this paper, we analyze spectral detection from both theoretical and empirical perspectives. We connect spectral energy to variance in proxy log-probability trajectories and explain how broader human token choices create the fluctuations used by frequency-domain indicators. We further show that the strength of this signal depends on text length and sampling range: spectral evidence is clearest for long, continuous, constrained generation, while short, fragmented, mixed, and edited settings require complementary confidence and fluctuation views. These findings clarify when frequency-domain detection works and provide guidance for future multi-dimensional detector design.

cs.CL↗

arXiv2026

Seeing Touch from Motion: A Unified Modality-Aware Visuo-Tactile Policy with Tactile Motion Correlation

Visuo-Tactile policies leveraging optical tactile sensors have shown great promise in contact-rich manipulation. These sensors achieve high spatial resolution and multi-dimensional force sensing by utilizing an internal camera to monitor the deformation of their elastic gel surface, thereby indirectly inferring tactile cues. Despite their advantages, extracting fine-grained contact states necessary for contact-rich manipulation remains an open challenge. Existing methods typically use either raw images or cumulative motion fields to represent tactile cues. However, both are prone to perception ambiguity. Raw tactile images mainly capture appearance changes, while cumulative motion fields only reflect the aggregate gel deformation. Consequently, distinct fine-grained contact states can exhibit highly similar patterns, making it difficult to explicitly distinguish subtle contact variations. To address this issue, we explore the dynamic priors of tactile motion and discover that the correlation between transient and cumulative motion can explicitly distinguish fine-grained contact states. Based on this insight, we propose a motion-aware tactile representation to facilitate contact-rich manipulation. Beyond tactile representation, effective fusion of tactile and visual modalities is also critical. Most existing fusion methods either directly concatenate features from each modality or train modality-specific networks separately and fuse their outputs. However, these strategies struggle to simultaneously model cross-modal interactions and preserve modality-specific characteristics. In this work, we take advantage of the Mixture-of-Transformers architecture and propose a unified modality-aware visuo-tactile policy that captures cross-modal complementarity while maintaining modality-specific properties.

cs.RO↗

arXiv2026

Versatile yet Efficient Network Traffic Analysis: Offloading Network Foundation Model to SmartNIC

Pervasive encryption makes large-scale labeling infeasible for traffic analysis, while security operations demand edge analysis to avert service degradation and further vulnerabilities. These pressures have produced two disjoint research lines: 1) versatile analysis, via network foundation models for low label dependency, and 2) efficient analysis, via hardware offloading for low analysis latency. However, versatility and efficiency have appeared fundamentally incompatible to co-achieve, with prior work consistently sacrificing one for the other, yet we show that this incompatibility is a consequence of polarized design choices across the three components of traffic analysis systems, i.e., traffic processing, model architecture, and analysis execution. In response, we present Nepco, a versatile yet efficient network traffic analysis system that offloads network foundation models to SmartNIC. Our key observation is that discriminative traffic information is concentrated in localized byte regions, motivating versatile yet efficient localized byte-sequence modeling rather than inefficient global modeling. To exploit this without incurring the latency bottlenecks of complex encoding steps, we employ a hardware-friendly processing pipeline that directly embeds raw byte sequences. Crucially, to maintain versatility across diverse tasks, we propose a pattern-aware convolutional architecture equipped with dedicated scoring and gating mechanisms. By exploiting translation invariance, this design dynamically locates and extracts salient semantic signatures. We prototype Nepco on the Nvidia BlueField-3 SmartNIC with multiengine collaborative analysis execution. The experimental results demonstrate that Nepco achieves macro F1 competitive with the best performances achieved by 8 state-of-the-art network foundation models, while reducing end-to-end latency by 328x to the millisecond scale.

cs.NI↗

arXiv2026

Nethira: A Heterogeneity-aware Hierarchical Pre-trained Model for Network Traffic Classification

Network traffic classification is vital for network security and management. The pre-training technology has shown promise by learning general traffic representations from raw byte sequences, thereby reducing reliance on labeled data. However, existing pre-trained models struggle with the gap between traffic heterogeneity (i.e., hierarchical traffic structures) and input homogeneity (i.e., flattened byte sequences). To address this gap, we propose Nethira, a heterogeneity-aware pre-trained model based on hierarchical reconstruction and augmentation. In pre-training, Nethira introduces hierarchical reconstruction at multiple levels-byte, protocol, and packet-capturing comprehensive traffic structural information. During fine-tuning, Nethira proposes a consistency-regularized strategy with hierarchical traffic augmentation to reduce label dependence. Experiments on four public datasets demonstrate that Nethira outperforms seven existing pre-trained models, achieving an average F1-score improvement of 9.11%, and reaching comparable performance with only 1% labeled data on high-heterogeneity network tasks.

cs.NI↗

arXiv2026

Aligning the Spectrum: Hybrid Graph Pre-training and Prompt Tuning across Homophily and Heterophily

Graph ``pre-training and prompt-tuning'' aligns downstream tasks with pre-trained objectives to enable efficient knowledge transfer under limited supervision. However, current methods typically rely on single-filter backbones (e.g., low-pass), whereas real-world graphs exhibit inherent spectral diversity. Our theoretical \textit{Spectral Specificity} principle reveals that effective knowledge transfer requires alignment between pre-trained spectral filters and the intrinsic spectrum of downstream graphs. This identifies two fundamental limitations: (1) Knowledge Bottleneck: single-filter models suffer from irreversible information loss by suppressing signals from other frequency bands (e.g., high-frequency); (2) Utilization Bottleneck: spectral mismatches between pre-trained filters and downstream spectra lead to significant underutilization of pre-trained knowledge. To bridge this gap, we propose HS-GPPT. We utilize a hybrid spectral backbone to construct an abundant knowledge basis. Crucially, we introduce Spectral-Aligned Prompt Tuning to actively align the downstream graph's spectrum with diverse pre-trained filters, facilitating comprehensive knowledge utilization across both homophily and heterophily. Extensive experiments validate the effectiveness under both transductive and inductive learning settings.

cs.LG↗

arXiv2026

SpecDetect: Simple, Fast, and Training-Free Detection of LLM-Generated Text via Spectral Analysis

The proliferation of high-quality text from Large Language Models (LLMs) demands reliable and efficient detection methods. While existing training-free approaches show promise, they often rely on surface-level statistics and overlook fundamental signal properties of the text generation process. In this work, we reframe detection as a signal processing problem, introducing a novel paradigm that analyzes the sequence of token log-probabilities in the frequency domain. By systematically analyzing the signal's spectral properties using the global Discrete Fourier Transform (DFT) and the local Short-Time Fourier Transform (STFT), we find that human-written text consistently exhibits significantly higher spectral energy. This higher energy reflects the larger-amplitude fluctuations inherent in human writing compared to the suppressed dynamics of LLM-generated text. Based on this key insight, we construct SpecDetect, a detector built on a single, robust feature from the global DFT: DFT total energy. We also propose an enhanced version, SpecDetect++, which incorporates a sampling discrepancy mechanism to further boost robustness. Extensive experiments show that our approach outperforms the state-of-the-art model while running in nearly half the time. Our work introduces a new, efficient, and interpretable pathway for LLM-generated text detection, showing that classical signal processing techniques offer a surprisingly powerful solution to this modern challenge.

cs.CL↗

arXiv2025

Heterogeneity-Oblivious Robust Federated Learning

Federated Learning (FL) remains highly vulnerable to poisoning attacks, especially under real-world hyper-heterogeneity, where clients differ significantly in data distributions, communication capabilities, and model architectures. Such heterogeneity not only undermines the effectiveness of aggregation strategies but also makes attacks more difficult to detect. Furthermore, high-dimensional models expand the attack surface. To address these challenges, we propose Horus, a heterogeneity-oblivious robust FL framework centered on low-rank adaptations (LoRAs). Rather than aggregating full model parameters, Horus inserts LoRAs into empirically stable layers and aggregates only LoRAs to reduce the attack uncover a key empirical observation that the input projection (LoRA-A) is markedly more stable than the output projection (LoRA-B) under heterogeneity and poisoning. Leveraging this, we design a Heterogeneity-Oblivious Poisoning Score using the features from LoRA-A to filter poisoned clients. For the remaining benign clients, we propose projection-aware aggregation mechanism to preserve collaborative signals while suppressing drifts, which reweights client updates by consistency with the global directions. Extensive experiments across diverse datasets, model architectures, and attacks demonstrate that Horus consistently outperforms state-of-the-art baselines in both robustness and accuracy.

cs.LG↗

arXiv2025

Distillation-Enhanced Clustering Acceleration for Encrypted Traffic Classification

Traffic classification plays a significant role in network service management. The advancement of deep learning has established pretrained models as a robust approach for this task. However, contemporary encrypted traffic classification systems face dual limitations. Firstly, pretrained models typically exhibit large-scale architectures, where their extensive parameterization results in slow inference speeds and high computational latency. Secondly, reliance on labeled data for fine-tuning restricts these models to predefined supervised classes, creating a bottleneck when novel traffic types emerge in the evolving Internet landscape. To address these challenges, we propose NetClus, a novel framework integrating pretrained models with distillation-enhanced clustering acceleration. During fine-tuning, NetClus first introduces a cluster-friendly loss to jointly reshape the latent space for both classification and clustering. With the fine-tuned model, it distills the model into a lightweight Feed-Forward Neural Network model to retain semantics. During inference, NetClus performs heuristic merge with near-linear runtime, and valid the cluster purity with newly proposed metrics ASI to identify emergent traffic types while expediting classification. Benchmarked against existing pretrained methods, NetClus achieves up to 6.2x acceleration while maintaining classification degradation below 1%.

cs.NI↗

arXiv2025

TraGe: A Generic Packet Representation for Traffic Classification Based on Header-Payload Differences

Traffic classification has a significant impact on maintaining the Quality of Service (QoS) of the network. Since traditional methods heavily rely on feature extraction and large scale labeled data, some recent pre-trained models manage to reduce the dependency by utilizing different pre-training tasks to train generic representations for network packets. However, existing pre-trained models typically adopt pre-training tasks developed for image or text data, which are not tailored to traffic data. As a result, the obtained traffic representations fail to fully reflect the information contained in the traffic, and may even disrupt the protocol information. To address this, we propose TraGe, a novel generic packet representation model for traffic classification. Based on the differences between the header and payload-the two fundamental components of a network packet-we perform differentiated pre-training according to the byte sequence variations (continuous in the header vs. discontinuous in the payload). A dynamic masking strategy is further introduced to prevent overfitting to fixed byte positions. Once the generic packet representation is obtained, TraGe can be finetuned for diverse traffic classification tasks using limited labeled data. Experimental results demonstrate that TraGe significantly outperforms state-of-the-art methods on two traffic classification tasks, with up to a 6.97% performance improvement. Moreover, TraGe exhibits superior robustness under parameter fluctuations and variations in sampling configurations.

cs.NI↗

arXiv2024

Spectral-Based Graph Neural Networks for Complementary Item Recommendation

Modeling complementary relationships greatly helps recommender systems to accurately and promptly recommend the subsequent items when one item is purchased. Unlike traditional similar relationships, items with complementary relationships may be purchased successively (such as iPhone and Airpods Pro), and they not only share relevance but also exhibit dissimilarity. Since the two attributes are opposites, modeling complementary relationships is challenging. Previous attempts to exploit these relationships have either ignored or oversimplified the dissimilarity attribute, resulting in ineffective modeling and an inability to balance the two attributes. Since Graph Neural Networks (GNNs) can capture the relevance and dissimilarity between nodes in the spectral domain, we can leverage spectral-based GNNs to effectively understand and model complementary relationships. In this study, we present a novel approach called Spectral-based Complementary Graph Neural Networks (SComGNN) that utilizes the spectral properties of complementary item graphs. We make the first observation that complementary relationships consist of low-frequency and mid-frequency components, corresponding to the relevance and dissimilarity attributes, respectively. Based on this spectral observation, we design spectral graph convolutional networks with low-pass and mid-pass filters to capture the low-frequency and mid-frequency components. Additionally, we propose a two-stage attention mechanism to adaptively integrate and balance the two attributes. Experimental results on four e-commerce datasets demonstrate the effectiveness of our model, with SComGNN significantly outperforming existing baseline models.

cs.IR↗