Search arXiv⌕ Search

arXiv subjects

Peng Sun

Publications and source records attributed to Peng Sun.

At least 19 recordsLinked to original sources
arXiv2026

Progressive-Observation Physical-Layer Fingerprinting for Efficient Open-Set Ethernet Authentication

Physical-layer fingerprinting (PLF) has emerged as a promising authentication approach for Internet of Things devices. However, most PLF systems acquire and process a fixed-length waveform for every authentication query, burdening resource-constrained edge nodes. To address this issue, we propose a progressive-observation PLF method that adapts the waveform length to each query. Specifically, a lightweight convolutional neural network (CNN) first rejects clearly unknown devices from a short observation. Spectral matching then accepts clearly registered devices from a slightly longer observation. Only ambiguous queries proceed to the same matching on the full-length waveform, reducing both data acquisition and computation. Because fingerprints drift across days, we also design an adaptive calibration algorithm to avoid retraining the CNN. The algorithm realigns the features and decision thresholds using a few current-day references from registered devices. We evaluate the method with 100BASE-TX Ethernet devices of two controller types on held-out days. Compared with fixed-length processing, it keeps the false rejection and false acceptance rates equal or lower for both controller types on both days. It also reduces the waveform data per query by 75.1% and the CPU processing time on a Raspberry Pi 5 by 73.7% on average. With the same monitoring bandwidth, an edge node can therefore serve about four times as many authentication queries.

cs.CR↗
arXiv2026

Extraction of transverse-momentum-dependent helicity distributions of the nucleon

We present an updated global analysis of the transverse-momentum-dependent (TMD) helicity distributions of the nucleon from measurements of the longitudinal double-spin asymmetry $A_{LL}$ in semi-inclusive deep-inelastic scattering (SIDIS). The analysis incorporates target- and detected-hadron-mass corrections, next-to-next-to-leading-order matching in the small-$b_T$ operator product expansion, and an improved treatment of correlated systematic uncertainties. The present data constrain the up- and down-quark helicity TMDs, whereas the sea-quark and gluon distributions remain weakly constrained. After integration over transverse momenta up to the hard scale, the up- and down-quark distributions agree with collinear helicity distributions within uncertainties. Using the extracted TMDs and the Wandzura-Wilczek-type approximation, we predict the $\cosϕ_h$ modulation of $A_{LL}$ and find agreement with the HERMES and CLAS measurements. We also provide predictions for SIDIS $A_{LL}$ at CLAS12 and for longitudinal single-spin asymmetries in $W^\pm$ production with leptonic decays at STAR kinematics. An impact study with pseudodata for the proposed Electron-Ion Collider in China (EicC) indicates substantial potential to improve constraints on helicity TMDs at intermediate $x$, including the presently weakly constrained sea-quark sector.

hep-ph↗
arXiv2026

Extreme Compression of Per-Layer Embeddings via Cross-Layer Latent Vector Quantization

Scaling embedding capacity alongside Transformer computation is an emerging trend in language models. This trend makes high compression rates increasingly useful. Hence, we study extreme compression of Per-Layer Embeddings (PLE), which follow this trend and account for 35.2% of the parameters of Gemma 4 E4B. At such rates, conventional quantizers give each token-layer vector its own code of only a few bytes. We instead develop OPTIC (Output-Path-fitted, Token-shared Indexed Codes), which stores one quantized code tuple per token and decodes it at every layer with a layer-specific affine decoder. Codes and decoders are fitted to the PLE output path rather than to the stored vectors. At about $100\times$ PLE compression of bfloat16 (BF16) E4B, OPTIC retains an aggregate 97.8% of the uncompressed model's performance across seven benchmarks, whereas product quantization, the strongest of six conventional baselines, retains 66.3%. At $594\times$ and $1,002\times$ PLE compression of BF16 E4B, OPTIC retains 90.5% and 84.1%. On the official E4B release with 4-bit linear weights and 16-bit activations (W4A16), OPTIC retains 96.9% at about $100\times$ PLE compression. During inference at about $100\times$ PLE compression, BF16 E4B uses 29.7% less GPU memory than with uncompressed PLE at the same key-value cache size, while running at most 0.40% slower.

cs.LG↗
arXiv2026

Metadata-Guided Adaptable Frequency Scaling across Heterogeneous Applications and Devices

Dynamic Voltage and Frequency Scaling is essential for enhancing energy efficiency in mobile platforms. However, traditional heuristic-based governors are increasingly inadequate for managing the complexity of heterogeneous System-on-Chip designs and diverse application workloads. Although reinforcement learning approaches offer improved performance, their poor generalization capability and reliance on extensive retraining for each hardware and application combination leads to significant deployment costs. In this work, we observe that device and application metadata inherently encapsulate valuable knowledge for DVFS, presenting an opportunity to overcome these limitations. We formulate DVFS for heterogeneous devices and applications as a multi-task reinforcement learning problem. We introduce MetaDVFS, which is a metadata-guided framework that systematically leverages metadata to discover and transfer shared knowledge across DVFS tasks. MetaDVFS can output a set of DVFS models with significant generalization capability for various applications of heterogeneous devices. Evaluations on five Google Pixel devices running six applications show that MetaDVFS achieves up to 17% improvement in Performance-Power Ratio and up to 26% improvement in Quality of Experience. Compared to state-of-the-art methods, MetaDVFS delivers 70.8% faster adaptation and 5.8-27.6% higher performance over standalone device-application specific training, while avoiding negative transfer effects. These results establish MetaDVFS as an effective and scalable solution for DVFS deployment in heterogeneous mobile environments.

cs.DC↗
arXiv2026

UltraText Bench: A Comprehensive Bilingual Benchmark for Evaluating Visual Text Rendering in Image Generation

Dense visual text requires image generators to reproduce long strings across multiple regions with correct placement and legibility. As short-string rendering improves, evaluation must test sustained performance across more demanding scenes. We introduce UltraText Bench, a bilingual benchmark for prompt-only generation of dense visual text. It contains 432 prompts spanning 24 real-world scene categories and three difficulty levels, split equally between English and Chinese. Each human-reviewed prompt supplies exact strings for four to twelve text regions, paired with structured references for their content, placement, and visual attributes. We use the Q-Judger vision-language model to assess each image against the complete reference, reporting text fidelity, text clarity, spatial quality, and scene quality. Across 24 model configurations, these dimensions reveal different strengths: Z-Image-Turbo gains 3.81 clarity points over Z-Image-Base while losing 14.76 fidelity points under the reported settings. Performance also varies with workload; Qwen-Image-2512's English composite falls from 86.50 at L1 to 42.86 at L3. Ten participants took part in human evaluation of the automatic scores. Repository: https://github.com/LINs-lab/UltraText_Bench.

cs.CV↗
arXiv2026

VenusRL: A Fully Disaggregated Agentic RL System with Priority Scheduling and Scalable Interaction

Agentic Reinforcement Learning (RL) trains LLM agents through multi-turn interactions with external tool environments. Its multi-turn nature exposes two system-level bottlenecks unaddressed by existing agentic RL frameworks. First, end-to-end training throughput is constrained by the slowest trajectories to complete, yet optimizing per-GPU utilization alone scatters rollout progress across many groups, delaying the completion of enough groups to unblock the next training step. Second, tool sandboxes are statically over-provisioned by their declared memory ceilings, leaving most physical memory stranded while replicating near-identical state across sandboxes launched from the same prompt. We present VenusRL, a fully disaggregated agentic RL system that addresses both bottlenecks. VenusRL's priority-aware action scheduler uses length-prediction heuristics to identify sample groups whose completion is most likely to unblock the next training step, and pushes them ahead of others across batch admission, KV Cache residency, and cross-worker request orchestration. VenusRL's environment resource manager combines a memory-aware admission threshold with a template-keyed page-sharing pool, packing more sandboxes per node while preserving strict memory isolation via write-protected page table entry aliasing and copy-on-write. Across representative agentic RL workloads, VenusRL achieves up to 4.24x end-to-end training speedup over state-of-the-art baselines and reduces environment cost by up to 89%.

cs.DC↗
arXiv2026

Learning Normal Diffusion Dynamics for Backdoor Defense in Text-to-Image Models

Backdoor attacks pose a serious threat to the secure deployment of text-to-image (T2I) diffusion models. Existing defenses typically detect backdoors from specific abnormal patterns in internal representations, which may limit their generalizability with the emergence of increasingly diverse attack mechanisms. In this paper, we study backdoor defense of T2I diffusion models from a transition-dynamics perspective. We observe that benign diffusion trajectories exhibit structured and timestep-dependent transition patterns from cross-attention, latent and noise spaces, whereas backdoor attacks tend to induce deviations from such normal evolution. Motivated by these observations, we propose Normal Diffusion Dynamics Learning (NDDL), a novel backdoor defense framework that learns the normal transition dynamics of diffusion trajectories utilizing only benign samples. NDDL constructs compact multi-space trajectory representations and trains a timestep-conditioned dynamics model to predict the diffusion evolution. In the inference phase, deviations between the observed and predicted transitions are exploited to quantify dynamics inconsistency for backdoor detection. NDDL further enables trigger localization without any prior knowledge of the embedded backdoor by performing substitution with low-semantic words. Extensive experiments for diverse backdoor attacks demonstrate the effectiveness and generalizability of our proposed NDDL.

cs.CV↗
arXiv2026

Efficient Agentic LLM Serving over SSD-based Sparse KV Storage

Agentic sessions driven by Large language models (LLMs) often alternate between model inference and tool use, accumulating long histories across successive rounds. Serving these sessions efficiently requires reducing attention computation and retaining history key-value (KV) caches to avoid recomputation. Recently, frontier open-source LLMs adopt sparse attention to reduce computation by selecting only part of the history, while SSDs provide a cheaper alternative to CPU DRAM for storing KV caches. However, sparse KV selection depends on the ad hoc intermediate values during model inference, so it forces SSD reads to lie on the inference critical path. These reads are further slowed by fragmented accesses and read-write interference in SSDs. To address these challenges, we present Janus, an agentic serving framework for sparse attention LLMs with SSD-centric KV storage. Janus focuses on append prefill, which processes each round's newly added inputs and accounts for most history KV loading. To move SSD reads out of the critical path, Janus runs the model's own KV selection module on earlier intermediate values, predicting KV demand without additional training. The predicted reads overlap with model computation, and any prediction misses are fetched before attention executes to preserve model outputs. To improve SSD efficiency, Janus coalesces adjacent reads, packs scattered KV pages into sequential writes on the CPU, and limits background writes while reads are active. Across three models and three agentic traces, Janus outperforms existing works by up to 1.57-3.69 times (1.22-1.85 times on average) in terms of the time to first token latency, while maintaining decode efficiency.

cs.DC↗
arXiv2026

GeoForward: Geometry-Guided Cross-View Alignment for Feedforward 3D Gaussian Splatting

Feedforward 3D reconstruction has largely followed an {appearance-driven paradigm}: multi-view images are encoded into visual representations, cross-view correspondence is implicitly inferred from learned features, and 3D primitives are subsequently predicted from these representations. However, calibrated cameras and sparse metric measurements already provide explicit constraints on where the same physical point should appear across views. We propose a novel paradigm, \textbf{Geometry-Conditioned Feedforward Reconstruction}, in which metric geometry establishes cross-view correspondence before the 3D representation is formed, while learned visual features refine rather than discover this correspondence. Based on this paradigm, we introduce \textbf{GeoForward}. Sparse LiDAR points are unprojected into 3D and analytically projected across calibrated cameras to provide metric correspondence anchors. A bounded continuous local search then refines these anchors using visual features, and the resulting geometry-conditioned representation is fused with global multi-view context from VGGT. The shared representation drives separate metric-depth and Gaussian-attribute prediction, establishing a direct path from explicit geometry to both Gaussian centers and spatial attributes. A sparse 3D refinement module subsequently improves local attribute compatibility while preserving the geometrically determined centers. For temporal reconstruction, endpoint Gaussians are composed using vehicle masks and motion priors. On nuScenes, GeoForward achieves the lowest cross-camera depth-consistency error and Chamfer distance among the compared methods while maintaining competitive reconstruction quality. These results demonstrate that explicitly conditioning correspondence formation on metric geometry provides an effective alternative to purely appearance-driven feedforward 3D reconstruction.

cs.CV↗
arXiv2026

AI-driven ionic liquid discovery with unified chemical intelligence

The rational discovery of functional solvents remains challenging because macroscopic solvent behaviors emerge from complex many-body interactions across vast chemical spaces. Ionic Liquids (ILs) constitute one of the most chemically diverse solvent families, yet their exceptional structural tunability remains only partially explored owing to the limitations of existing experimental and computational approaches. Here we introduce AIonopedia, a large-language-model-orchestrated agentic framework for IL research. Built around a domain-specific multimodal foundation model, it integrates information retrieval, chemical reasoning and autonomous tool execution within a unified architecture. Pretrained on large-scale unlabeled corpora and fine-tuned using over 80,000 curated measurements, AIonopedia learns expressive joint representations of multicomponent ionic systems, enabling accurate prediction across diverse physicochemical properties. Through a series of literature-grounded evaluations, AIonopedia exhibits robust out-of-distribution generalization, while many-body-expansion-based analysis provides physically grounded explanations of structure-property relationships. By coupling molecular screening and optimization with wet-lab validation in a closed-loop workflow, AIonopedia identifies high-performance ILs and reveals transferable design principles for the capture of various volatile organic compounds. This work establishes an end-to-end AI-driven paradigm for understanding and exploring complex ionic systems, demonstrating the promise of chemically specialized foundation models as a generalizable strategy for scientific discovery.

cs.AI↗
arXiv2026

Server-Enforced Watermarking in U-Shaped Split Federated Learning

U-shaped split federated learning (U-SFL) enables resource-constrained Internet of Things devices to collaboratively train models with an edge server while retaining raw data and labels locally. We consider model-service deployments in which a provider supplies proprietary models whose client-side segments reside on participating devices, creating risks of unauthorized copying and redistribution. Protecting these segments is challenging because the server cannot directly access client data, labels, or model parameters, and potentially malicious clients may refuse to perform watermark embedding. To address this challenge, we propose Sigil, a server-enforced watermarking framework for U-SFL. Sigil defines a secret watermark constraint in the server-visible activation space and embeds the watermark into client-side models by injecting a watermark gradient into the gradients returned during training. This mechanism requires neither access to clients' raw data and labels nor client-side watermarking operations. To limit interference with the main task and reduce detectability by gradient anomaly detectors, Sigil adaptively clips the watermark gradient relative to the main-task gradient. Experiments on two datasets and four model architectures demonstrate high watermark detection rates with limited impact on task accuracy, robustness against the evaluated removal attacks, and stealthiness against the evaluated gradient anomaly detector.

cs.CR↗
arXiv2026

Stateful Agent Backdoors: Constructing Cross-Session Attack Programs

Multi-step attacks on large language model agents may depend on opportunities distributed across sessions, such as access to target information or the availability of required tools. To combine these opportunities, an attack needs to retain its state and intermediate results, and choose actions based on current conditions. In this work, we study such attacks as cross-session attack programs. We focus on their shared control structures, including sequential execution with conditional waiting, branching and merging, looping, and condition accumulation. We represent these programs as Mealy machines and construct a sub-backdoor for each transition. We build single-session training trajectories for each sub-backdoor, combine them into a training dataset, and fine-tune the model to learn the local behaviors jointly. After a single injection of the initial trigger, the agent uses persistent memory to connect local behaviors into a complete cross-session attack program. We evaluate four instantiations of these control structures across four models in a LangChain-based agent environment. The primary instantiation achieves mean complete-program success rates of 71.7\%--94.7\% in LangChain. In the official OpenClaw runtime under a controlled configuration, it achieves a complete-program success rate of 85\% for each of two evaluated models. These results demonstrate the feasibility of end-to-end execution of cross-session attack programs with different control structures.

cs.CR↗
arXiv2026

Sequential Detection-Based Iterative Blind Separation for Single-Channel Co-Frequency Signals

Existing single-channel co-frequency signal blind separation (SCSBS) algorithms struggle to balance separation accuracy, computational complexity, and robustness, while current channel state information (CSI) estimation methods lack precision. To address these limitations, we propose a sequential detection (SD)-based iterative separation (SDIS) algorithm. SDIS incorporates a delayed unscented Kalman filter (DUKF) into an iterative decision feedback framework, jointly enhancing signal separation and CSI estimation. Simulation results show that SDIS outperforms benchmarks in separation accuracy, CSI estimation accuracy, computational efficiency, and robustness. Notably, when the mean bit error rate (MBER) drops below $10^{-4}$, SDIS can tolerate at least $0.8$ dB more noise than the benchmarks.

eess.SP↗
arXiv2026

Realization of all-to-all fermion propagator for the first principle high accuracy strong interaction prediction

We propose a ``blending" algorithm that projects the all-to-all fermion propagator onto spatial low-frequency modes (LFM) combines the projection with a stochastic estimate of spatial high-frequency modes (SHFM) at each time slice. This approach enables the calculation of correlation functions at arbitrary points for arbitrary hadron states in strongly interacting quantum field theories (QFT) with fermions, such as quantum chromodynamics (QCD). Specifically, LFM allows the construction of spatially extended hadron states below a certain energy threshold by diagonalizing multi-fermion interpolation fields. Meanwhile, the local interactions required for N-point correlation functions in QFT can be approximated in an unbiased manner through a reweighted summation of both LFM and SHFM contributions. To demonstrate the efficiency of this algorithm, we obtained $g_A^u=0.8408(86)$, $g_A^d= -0.3929(86)$, $g_A^s=-0.0381(57)$, $g_A^{u+d+s}=0.410(20)$ and $g_A^{u-d}=1.2337(84)$ for the nucleon at $m_π=135$ MeV and $a=0.077$ fm using 41 configurations. We also provide a consistency check of the pion electric form factor and charge radius derived from 3-point and 4-point correlation functions is also provided.

hep-lat↗
arXiv2026

FlowCPO: A Unified Divergence View of Preference Alignment for Flow Models

Preference alignment for flow and diffusion models now spans online reinforcement learning and offline preference optimization, but the relation between these methods remains unclear. In particular, existing forward-process alignment methods require fresh samples from the current model, while offline methods based on fixed preference pairs rely primarily on positive-only fine-tuning or DPO-style likelihood-ratio surrogates. We organize these approaches through a divergence-based framework and introduce FlowCPO, an offline forward-KL objective that uses both preferred and dispreferred samples without online rollouts. For linear interpolation, we show under explicit regularity conditions that the forward-KL objective is bounded by a contrastive flow matching loss, yielding a tractable surrogate on fixed data. We further show that this loss is nonnegative, whereas the signed regression loss of simplified FlowDPO can be unbounded below. In the in-domain setting, FlowCPO achieves higher mean GenEval and OCR scores than the evaluated baselines, reaching 0.84 and 0.87 versus 0.81 and 0.74 for FlowDPO at CFG 3.0. In the out-of-domain setting, the results are mixed, with the best GenEval result but lower reward scores than RFT on several metrics.

stat.ML↗
arXiv2026

Vision-Guided Text Prompt Tuning for Multimodal Sentiment Analysis

Multimodal sentiment analysis requires effective modeling of both verbal semantics and non-verbal affective cues. A central challenge is to calibrate text-centered sentiment understanding with visual facial evidence in a controlled, adaptive, and parameter-efficient manner. Text usually serves as the semantic anchor, whereas visual cues provide complementary evidence for ambiguous or implicit expressions; however, indiscriminate fusion may introduce visual noise and distort textual semantics. Moreover, fully fine-tuning large visual and textual encoders is costly and prone to overfitting on limited and scenario-dependent MSA benchmarks. To address these issues, we propose Vision-Guided Text Prompt Tuning (VG-TPT), which formulates visual-text sentiment modeling as controllable visual calibration of frozen text representations. VG-TPT injects visual affective cues into a frozen BERT encoder through layer-wise adaptive prompts, rather than relying on late-stage feature fusion or full backbone tuning. A co-guided router composes prompts from a trainable prompt bank according to both the evolving text state and the visual guidance feature, enabling sample-specific and layer-specific modulation. Experiments on CMU-MOSEI and CMU-MOSI show that VG-TPT consistently improves over text-only baselines and achieves competitive or superior performance compared with several full-modality methods, while updating only 2.4M trainable parameters. The code is available at https://github.com/ma-tubu/VG-TPT.

cs.MM↗
arXiv2026

LLaDA-Image: Building Strong Image Generators with Fully Open Training Recipes

We introduce LLaDA-Image, a unified framework that pairs a 6B Diffusion Transformer (DiT) trained from scratch with a frozen vision-language understanding module built on the LLaDA2.0-Mini diffusion language model backbone. Instead of relying heavily on paired image-text data from the beginning, we first build a strong visual generative prior through image-only pre-training and mid-training. The generation pipeline comprises 220M samples, 98 of which are real images. For efficient and scalable optimization, we use parameter-free RMSNorm throughout the DiT together with the Muon optimizer. The resulting unified model produces highly photorealistic images while accurately following fine-grained editing instructions. We further distill LLaDA-Image into LLaDA-Image-Turbo, enabling fast inference in 2-4 sampling steps. On Qwen-Image-Bench, LLaDA-Image achieves overall scores of 53.53 and 53.38 on the English and Chinese tracks, respectively, setting a new state-of-the-art among open-source models on both tracks. To support further research on capable and efficient generative models, we release our model weights, training code, and detailed recipes.

cs.CV↗
arXiv2026

Low-Complexity Sequential Detection Framework for Single-Channel Co-Frequency Signal Separation

Practical separation of single-channel co-frequency signals (SCCFSs) is hindered by the prohibitive computational complexity of benchmark algorithms. To address this issue, we propose a low-complexity separation framework based on sequential detection (SD), in which signal separation is cast as a sequential path search over a trellis. To support both hard-decision detection and log-likelihood ratio (LLR) extraction, we develop two algorithms within this framework: the SD-based separation (SDS) algorithm and its soft-output variant (SO-SDS). Furthermore, SDS employs a windowing strategy combined with dynamic pruning to concentrate computational resources on high-probability paths, thereby enabling efficient detection of transmitted symbol sequences. Building upon SDS, SO-SDS further incorporates a state completeness verification mechanism (SCVM) to estimate bit LLRs, thus facilitating subsequent soft decoding. Numerical results show that, compared to benchmark algorithms, SDS achieves significant complexity reduction without degrading separation performance, while SO-SDS offers notable computational savings with only modest LLR accuracy loss. Notably, the computational complexity advantage of the proposed algorithms over benchmark algorithms grows substantially with increasing modulation order.

eess.SP↗