Searcharxiv⌕ Search

arXiv subjects

Bin Chen

Publications and source records attributed to Bin Chen.

At least 109 records · Page 6Linked to original sources
arXiv2026

Prompt2Fingerprint: Plug-and-Play LLM Fingerprinting via Text-to-Weight Generation

The widespread deployment and redistribution of large language models (LLMs) have made model provenance tracking a critical challenge. While existing LLM fingerprinting methods, particularly active approaches that embed identity signals via fine-tuning, achieve high accuracy and robustness, they suffer from significant scalability bottlenecks. These methods typically treat fingerprint injection as an independent, one-off optimization task rather than a reusable capability, necessitating separate, resource-intensive training for every new identity. This incurs prohibitive computational costs and deployment delays. To address this, we propose Prompt2Fingerprint (P2F), the first framework that reformulates fingerprinting as a conditional parameter generation task. By leveraging a specialized generator, P2F maps textual descriptions directly to low-rank parameter increments in a single forward pass, enabling plug-and-play LLM fingerprint injection without further model retraining. Our experiments demonstrate that P2F maintains high fingerprint accuracy, harmlessness, and robustness while significantly reducing computational overhead, offering a scalable and instant solution for LLM ownership management.

cs.CR↗
arXiv2026

STEM: Structure-Tracing Evidence Mining for Knowledge Graphs-Driven Retrieval-Augmented Generation

Knowledge Graph-based Question Answering (KGQA) plays a pivotal role in complex reasoning tasks but remains constrained by two persistent challenges: the structural heterogeneity of Knowledge Graphs(KGs) often leads to semantic mismatch during retrieval, while existing reasoning path retrieval methods lack a global structural perspective. To address these issues, we propose Structure-Tracing Evidence Mining (STEM), a novel framework that reframes multi-hop reasoning as a schema-guided graph search task. First, we design a Semantic-to-Structural Projection pipeline that leverages KG structural priors to decompose queries into atomic relational assertions and construct an adaptive query schema graph. Subsequently, we execute globally-aware node anchoring and subgraph retrieval to obtain the final evidence reasoning graph from KG. To more effectively integrate global structural information during the graph construction process, we design a Triple-Dependent GNN (Triple-GNN) to generate a Global Guidance Subgraph (Guidance Graph) that guides the construction. STEM significantly improves both the accuracy and evidence completeness of multi-hop reasoning graph retrieval, and achieves State-of-the-Art performance on multiple multi-hop benchmarks.

cs.CL↗
arXiv2026

Mistletoe: Stealthy Acceleration-Collapse Attacks on Speculative Decoding

Speculative decoding has become a widely adopted technique for accelerating large language model (LLM) inference by drafting multiple candidate tokens and verifying them with a target model in parallel. Its efficiency, however, critically depends on the average accepted length $τ$, i.e., how many draft tokens survive each verification step. In this work, we identify a new mechanism-level vulnerability in model-based speculative decoding: the drafter is trained to approximate the target model distribution, but this approximation is inevitably imperfect. Such a drafter-target mismatch creates a hidden attack surface where small perturbations can preserve the target model's visible behavior while substantially reducing draft-token acceptability. We propose Mistletoe, a stealthy acceleration-collapse attack against speculative decoding. Mistletoe directly targets the acceptance mechanism of speculative decoding. It jointly optimizes a degradation objective that decreases drafter-target agreement and a semantic-preservation objective that constrains the target model's output distribution. To resolve the conflict between these objectives, we introduce a null-space projection mechanism, where degradation gradients are projected away from the local semantic-preserving direction, suppressing draft acceptance while minimizing semantic drift. Experiments on various speculative decoding systems show that Mistletoe substantially reduces average accepted length $τ$, collapses speedup, and lowers averaged token throughput, while preserving output quality and perplexity. Our work highlights that speculative decoding introduces a mechanism-level attack surface beyond existing output robustness, calling for more robust designs of LLM acceleration systems.

cs.CL↗
arXiv2026

Energy Evolution from the Chromosphere to the Heliosphere in the 2021 October 28 Solar Eruption

We perform a detailed study of the energetics for a well-observed solar eruption and flare that occurred on 28 October 2021. This event included a GOES class X1.0 flare, a global EUV wave, and a coronal mass ejection that reached speeds of >2000 km/s. The event was observed from a variety of spacecraft in NASA's Heliophysics System Observatory, including multiple missions near Earth, STEREO-A off the Sun-Earth line, and Solar Orbiter, near the Sun-Earth line at about 0.8 au. Using remote sensing, in situ observations, and in some cases scaling laws based on previous observations, we characterize the following quantities: free magnetic energy, energy in non-thermal electrons, energy in non-thermal ions, bolometric energy, energy deposited in the chromosphere, thermal energy radiated in the flare loops, energy dissipated by the EUV wave, CME kinetic and gravitational potential energy, CME energy flux in the heliosphere, and the energy partition in the CME shock. We find that the total energy released during the event is consistent with estimates of the pre-event stored magnetic energy, and the CME kinetic + potential energy dominates the energy partition.

astro-ph.SR↗
arXiv2026

IndustryBench: Probing the Industrial Knowledge Boundaries of LLMs

In industrial procurement, an LLM answer is useful only if it survives a standards check: recommended material must match operating condition, every parameter must respect a regulated threshold, and no procedure may contradict a safety clause. Partial correctness can mask safety-critical contradictions that aggregate LLM benchmarks rarely capture. We introduce IndustryBench, a 2,049-item benchmark for industrial procurement QA in Chinese, grounded in Chinese national standards (GB/T) and structured industrial product records, organized by seven capability dimensions, ten industry categories, and panel-derived difficulty tiers, with item-aligned English, Russian, and Vietnamese renderings. Our construction pipeline rejects 70.3% of LLM-generated candidates at a search-based external-verification stage, calibrating how unreliable industrial QA remains after LLM-only filtering. Our evaluation decouples raw correctness, scored by a Qwen3-Max judge validated at $κ_w = 0.798$ against a domain expert, from a separate safety-violation (SV) check against source texts. Across 17 models in Chinese and an 8-model intersection over four languages, we find: (i) the best system reaches only 2.083 on the 0--3 rubric, leaving substantial headroom; (ii) Standards & Terminology is the most persistent capability weakness and survives item-aligned translation; (iii) extended reasoning lowers safety-adjusted scores for 12 of 13 models, primarily by introducing unsupported safety-critical details into longer final answers; and (iv) safety-violation rates reshuffle the leaderboard -- GPT-5.4 climbs from rank 6 to rank 3 after SV adjustment, while Kimi-k2.5-1T-A32B drops seven positions. Industrial LLM evaluation therefore requires source-grounded, safety-aware diagnosis rather than aggregate accuracy. We release IndustryBench with all prompts, scoring scripts, and dataset documentation.

cs.AI↗
arXiv2026

GLM-5V-Turbo: Toward a Native Foundation Model for Multimodal Agents

We present GLM-5V-Turbo, a step toward native foundation models for multimodal agents. As foundation models are increasingly deployed in real environments, agentic capability depends not only on language reasoning, but also on the ability to perceive, interpret, and act over heterogeneous contexts such as images, videos, webpages, documents, GUIs. GLM-5V-Turbo is built around this objective: multimodal perception is integrated as a core component of reasoning, planning, tool use, and execution, rather than as an auxiliary interface to a language model. This report summarizes the main improvements behind GLM-5V-Turbo across model design, multimodal training, reinforcement learning, toolchain expansion, and integration with agent frameworks. These developments lead to strong performance in multimodal coding, visual tool use, and framework-based agentic tasks, while preserving competitive text-only coding capability. More importantly, our development process offers practical insights for building multimodal agents, highlighting the central role of multimodal perception, hierarchical optimization, and reliable end-to-end verification.

cs.CV↗
arXiv2026

Black Hole Ringdown Seen in Photon Polarization Swings

Light propagating through a perturbed spacetime could imprint the underlying gravitational waveform directly onto electromagnetic observables. In this Letter, we develop a covariant perturbative framework for polarized photon propagation in generic curved spacetimes, and derive a compact expression for the observable polarization-angle (PA) swing during Kerr ringdown, explicitly demonstrating its time-domain locking to the quasi-normal modes. We confirm this behavior using dynamical ray-tracing calculations for a broad class of photon trajectories. Photons grazing the strong-field region exhibit an achromatic, damped PA oscillation that tracks the ringdown, with a phase set by the mode's angular structure. The swing amplitude can reach $\sim 10^{\circ}$ and leaves distinctive signatures in spatially resolved autocorrelations. These results open a new polarimetric window onto black hole mergers and ringdown.

astro-ph.HE↗
arXiv2026

Effects of stimulation frequencies on energy efficiency of a muscle fiber during contraction

Contradictory experimental reports on the relationship between efficiency and stimulation frequency have hindered mechanistic understanding in converting neural activity into mechanical work during muscle contraction. To resolve this issue, we develop a biophysical model integrating calcium-mediated excitation with a detailed cross-bridge cycle to enable single-fiber simulations. Our model predictions indicate that the emergent shortening velocity is the primary determinant of cross-bridge efficiency: efficiency peaks at an optimal velocity and declines at higher or lower velocities, while frequency appears to exert secondary influence. Critically, the velocity yielding peak efficiency remains almost consistent across frequencies, with a slight upward shift at higher frequencies in most of our parametric studies. Interestingly, elevated inorganic phosphate ([Pi]) appears to amplify the efficiency disparity between high- and low-frequency regimes in our analysis. Our work suggests that stimulation frequency modulates efficiency predominantly through its regulation of shortening velocity, which primarily governs the kinetics of the myosin power stroke. This work may help clarify neural control of muscle energetics, and provide a quantitative foundation for studying muscle function in physiological and pathological contexts.

physics.bio-ph↗
arXiv2026

Towards Distillation-Resistant Large Language Models: An Information-Theoretic Perspective

Proprietary large language models (LLMs) embody substantial economic value and are generally exposed only as black-box APIs, yet adversaries can still exploit their outputs to extract knowledge via distillation. Existing defenses focus exclusively on text-based distillation, leaving the important logit-based distillation largely unexplored. In this work, we analyze this problem and present an effective solution from an information-theoretic perspective. We characterize distillation-relevant information in teacher outputs using the conditional mutual information (CMI) between teacher logits and input queries conditioned on ground-truth labels. This quantity captures contextual information beneficial for model extraction, motivating us to defend distillation via CMI minimization. Guided by our theoretical analysis, we propose learning a transformation matrix that purifies the original outputs to enhance distillation resistance. We further derive a CMI-inspired anti-distillation objective to optimize this transformation, which effectively removes distillation-relevant information while preserving output utility. Extensive experiments across multiple LLMs and strong distillation algorithms demonstrate that the proposed method significantly degrades distillation performance while preserving task accuracy, effectively protecting models' intellectual property.

cs.CL↗
arXiv2026

Reduced-order Neural Modeling with Differentiable Simulation for High-Detail Tactile Perception

Tactile perception is key to dexterous manipulation, yet simulating high-resolution elastomer deformation remains computationally prohibitive. Finite element methods (FEM) deliver high fidelity but demand costly remeshing, while Material Point Methods (MPM) suffer from heavy particle-memory tradeoffs. We propose a {reduced-order neural simulation framework} that couples coarse-grained MPM dynamics with an implicit neural decoder to reconstruct sub-particle tactile details from compact latent states. The framework learns a continuous deformation manifold from paired high- and low-resolution simulations, enabling physically consistent, differentiable inference. Compared to the TacIPC, our method achieves over 65\% faster simulation and {40\% lower memory usage}, while maintaining better geometric fidelity. In tactile rendering and 3D surface reconstruction, our methods further improve accuracy by 25\% and produce realistic depth images and surface mesh within a faster inference speed. These results demonstrate that the proposed reduced-order neural model enables high-detail, physically grounded tactile simulation with substantial efficiency gains for robotic interaction and optimization.

cs.RO↗
arXiv2026

Fusing Urban Structure and Semantics: A Conditional Diffusion Model for Cross-City OD Matrix Generation

Accurate modeling of commuting flows is important for urban governance, traffic planning, and resource allocation. However, the combined influence of individual intentions, geographic constraints, and social dynamics leads to considerable heterogeneity in commuting patterns, making it difficult to develop generation models that generalize across cities. To address this issue, we propose SEDAN, a Structure-Enhanced Diffusion model conditioned on Attributed Nodes for generalizable OD matrix generation. SEDAN models a city as an attributed graph. Each region is treated as a node with demographic and point-of-interest features, and commuting flows are modeled as weighted edges. Adjacency and distance matrices are incorporated to characterize spatial structure. Based on this representation, we design a fusion mechanism within SEDAN to jointly model semantic information and spatial information. Regional semantic attributes are used to model latent travel demand through graph-transformer-based node interactions, while spatial structure is injected into the generation process as explicit constraints. The adjacency matrix guides attention weights to strengthen interactions between neighboring regions. Meanwhile, the distance matrix serves as a diffusion condition to capture spatial proximity and travel impedance. The fusion of urban semantics and spatial constraints enables SEDAN to generate OD matrices that are both behaviorally plausible and geographically coherent. Experiments on real-world OD datasets from U.S. cities show that SEDAN achieves a 7.38\% improvement in RMSE over the state-of-the-art baseline, WEDAN. It also remains robust across heterogeneous urban scenarios and varying structural patterns. Our work provides an effective and generalizable solution for commuting OD matrix generation. The code is available at https://anonymous.4open.science/r/SEDAN.

cs.LG↗
arXiv2026

A2DEPT: Large Language Model-Driven Automated Algorithm Design via Evolutionary Program Trees

Designing heuristics for combinatorial optimization problems (COPs) is a fundamental yet challenging task that traditionally requires extensive domain expertise. Recently, Large Language Model (LLM)-based Automated Heuristic Design (AHD) has shown promise in autonomously generating heuristic components with minimal human intervention. However, most existing LLM-based AHD methods enforce fixed algorithmic templates to ensure executability, which confines the search to component-level tuning and limits system-level algorithmic expressiveness. To enable open-ended solver synthesis beyond rigid templates, we propose Automated Algorithm Design via Evolutionary Program Trees (A2DEPT), which treats LLMs as system-level algorithm architects. A2DEPT explores the vast program space via a tree-structured evolutionary search with hybrid selection and hierarchical operators, enabling iterative refinement of complete algorithms. To make open-ended generation practical, we enforce executability with a lightweight program-maintenance loop that performs feedback-driven repair. In experiments, A2DEPT consistently outperforms representative LLM-based baselines on both standard and highly constrained benchmarks. On the standard benchmarks, it reduces the mean normalized optimality gap by 9.8% relative to the strongest competing AHD baseline.

cs.AI↗
arXiv2026

Multi-Viewpoint Observation of a Failed Prominence Eruption on the Sun

Solar eruptions are sudden ejections of coronal mass and magnetic fields accompanied by intense energy release. The eruptive structure does not always erupt successfully, but sometimes fails to escape the Sun after initiation. The failure of an eruption, however, provides an invaluable opportunity for understanding the intricate mechanism of eruptions. We present a comprehensive observation of a failed prominence eruption on the Sun, taking advantage of multi-viewpoint and multi-messenger imaging. Simultaneous off-limb and on-disk observation gives evidence of magnetic reconnection processes occurring at different sites during the flare. Particularly, in addition to the standard flare reconnection behind the eruption, strong external reconnection occurs on the erupting flux rope, evidenced by a wealth of signatures via multi-wavelength imaging and spectroscopy. The two reconnection processes may play contrasting roles in the flux rope's acceleration and compete in altering the magnetic flux in the rope. As the high rate of external reconnection proceeds, the flux rope and embedded prominence decelerate noticeably and fail to erupt into the heliosphere, under strong magnetic confinement of overlying fields. Our results illustrate a well-defined physical picture for solar eruptive activities and provide insight into the lack of coronal mass ejections found in other solar-type stars.

astro-ph.SR↗
arXiv2026

Tree Training: Accelerating Agentic LLMs Training via Shared Prefix Reuse

Agentic large language model (LLM) training often involves multi-turn interaction trajectories that branch into multiple execution paths due to concurrent tool use, think-mode, sub-agent, context management and other runtime designs. As a result, the tokens produced by a single task naturally form a tree-structured token trajectory with shared prefixes, rather than a linear sequence. Existing training pipelines linearize such trajectories and treat each branch independently, leading to substantial redundant computation in both forward and backward passes. We derive that averaging the loss over all branches independently is algebraically identical to a per-token weighted loss, where each token's weight equals the fraction of branches passing through it. The problem therefore reduces to computing the log-probability of every token in the prefix tree exactly once, with no repeated computation across shared prefixes: we propose DFS serialization of the tree, which visits every token exactly once, and adapt full-attention and SSM layers to ensure the resulting log-probabilities match independent per-branch calculation exactly. In practice, a single trajectory tree can be too large to fit in GPU memory; we therefore propose Redundancy-Free Tree Partitioning, which handles memory-constrained settings with zero redundant computation and peak memory bounded by a single root-to-leaf path. Together, these contributions form Tree Training, an efficient framework for training LLMs on tree-structured trajectories, achieving up to 6.2x end-to-end training speedup on dense and MoE models for both supervised fine-tuning and reinforcement learning.

cs.LG↗
arXiv2026

Testing solitonic boson star interpretations of Sagittarius A* with near-infrared flare astrometry

We use GRAVITY near-infrared (NIR) flare astrometry to test whether Sagittarius A* could be a solitonic boson star. We consider five spherically symmetric solitonic boson-star models with different effective radii, together with the Schwarzschild black hole. Treating the flares as hot spots on circular equatorial orbits, we analyze their centroid motions and images in these spacetimes and use them for parameter fitting. We perform the fitting using both $χ^2$ analysis and Markov Chain Monte Carlo (MCMC) methods, which yield consistent results: the inferred masses of boson-star models are systematically larger than the established value of $4.3\times10^6M_\odot$. Notably, more diffusive boson stars exhibit imaging properties closer to those of a black hole, leading to mass estimates that are correspondingly closer to the established value. Overall, our results place stringent constraints on solitonic boson star interpretations of Sagittarius A*, although do not completely rule them out.

astro-ph.HE↗
arXiv2026

From Verbatim to Gist: Distilling Pyramidal Multimodal Memory via Semantic Information Bottleneck for Long-Horizon Video Agents

While multimodal large language models have demonstrated impressive short-term reasoning, they struggle with long-horizon video understanding due to limited context windows and static memory mechanisms that fail to mirror human cognitive efficiency. Existing paradigms typically fall into two extremes: vision-centric methods that incur high latency and redundancy through dense visual accumulation, or text-centric approaches that suffer from detail loss and hallucination via aggressive captioning. To bridge this gap, we propose MM-Mem, a pyramidal multimodal memory architecture grounded in Fuzzy-Trace Theory. MM-Mem structures memory hierarchically into a Sensory Buffer, Episodic Stream, and Symbolic Schema, enabling the progressive distillation of fine-grained perceptual traces (verbatim) into high-level semantic schemas (gist). Furthermore, to govern the dynamic construction of memory, we derive a Semantic Information Bottleneck objective and introduce SIB-GRPO to optimize the trade-off between memory compression and task-relevant information retention. In inference, we design an entropy-driven top-down memory retrieval strategy. Extensive experiments across 4 benchmarks confirm that MM-Mem achieves state-of-the-art performance on both offline and streaming tasks, demonstrating robust generalization and validating the effectiveness of cognition-inspired memory organization. Code and associated configurations are publicly available at https://github.com/EliSpectre/MM-Mem.

cs.CV↗
arXiv2026

HiPrune: Hierarchical Attention for Efficient Token Pruning in Vision-Language Models

Vision-Language Models (VLMs) encode images and videos into abundant tokens, which contain substantial redundancy and computation cost. While visual token pruning mitigates the issue, most existing methods lack insight into the intrinsic property of the vision encoder itself. In this work, we dive into the vision encoder and prove that the middle layers pay more attention to the main objects of the image qualitatively and quantitatively, while the deep layers to tokens with rich global information. Utilizing this Hierarchical attention pattern, we propose HiPrune, a training-free and model-agnostic token Pruning method. HiPrune identifies three types of visual tokens according to their attention in different phases of the vision encoder, which preserves different levels of information. By coupling with the similarity of text tokens, we propose a prompt-aware variance, HiPrune++, which further improves instruction following performance under a very low token budget. Extensive experiments across four representative VLMs show that HiPrune achieves up to 99.3% of task accuracy with only 1/3 of the tokens, while reducing inference FLOPs by 58.7%. HiPrune++ maintains up to 99.7% accuracy with 2/9 tokens, highlighting robustness under high-resolution. Our code is available at https://github.com/Danielement321/HiPrune.

cs.CV↗
arXiv2026

Electrochemical reactions under reverse bias create additional mobile ions that enable hole tunneling in metal halide perovskite diodes

Gradual reverse-bias breakdown in metal-halide perovskite diodes and solar cells is thought to originate from hole tunneling through steep bands in an ionic depletion region near the electron transport layer after positively charged iodine vacancies accumulate near the hole-transport layer (HTL). However, typical reported mobile ion concentrations near $1\times10^{17}$ cm$^{-3}$ are too small to quantitatively explain significant tunneling current densities and (Zener) breakdown observed near $-5$ V. Here, we show that inferred mobile ion concentrations increase by more than 100$\times$, to over $1\times10^{18}$ cm$^{-3}$, within just three minutes of reverse bias at $-6.0$ V in p-i-n perovskite diodes. We attribute the increase in mobile ion concentration to iodide oxidation and the resulting iodine vacancy creation which must be balanced by reduction reactions near the HTL. Thin and sub-optimal HTL coverage leads to direct contact between the transparent conducting electrode and perovskite and facilitates electron transfer and reduction, enabling the creation of even larger inferred mobile ion concentrations ($\sim1\times10^{19}$ cm$^{-3}$) and leading to faster degradation under reverse bias. This explains previous work that showed increased breakdown voltages and improved reverse-bias stability by implementing thick, uniform HTLs.

physics.app-ph↗