Searcharxiv⌕ Search

arXiv subjects

Yu Pan

Publications and source records attributed to Yu Pan.

At least 37 records · Page 2Linked to original sources
arXiv2026

GPO-V: Jailbreak Diffusion Vision Language Model by Global Probability Optimization

Diffusion Vision-Language Models (dVLMs), built upon the non-causal foundations of Diffusion Large Language Models (dLLMs), have demonstrated remarkable efficacy in multimodal tasks by departing from the traditional autoregressive generation paradigm. While dVLMs appear inherently robust against conventional jailbreak tactics, which we categorize as Fixed Prefix Optimization (FPO) (e.g., anchoring responses with "Sure, here is"), this perceived resilience is deceptive. Our investigation into the safety landscape of dVLMs reveals a unique refusal pattern: Immediate Refusal and Progressive Refusal. We find that while FPO-based attacks often fail by triggering the latter, the progressive refinement process itself uncovers a novel, latent attack surface. To exploit this vulnerability, we propose Global Probability Optimization (GPO), a general jailbreak paradigm designed specifically for the denoising trajectory of masked diffusion models. Unlike prefix-based methods, GPO manipulates the global generative dynamics to bypass guardrails in diffusion language models. Building on this, we introduce GPO-V, the first visual-modality jailbreak framework tailored for dVLMs. Empirical results demonstrate that GPO-V produces stealthy perturbations with exceptional cross-model transferability, revealing a critical security gap in non-sequential generative architectures. Our findings underscore the critical urgency of addressing safety alignment in dVLMs. These results necessitate an immediate and fundamental re-evaluation of current defense paradigms to mitigate the unique risks of diffusion-based generation. Our code is available at: https://anonymous.4open.science/r/GPO-V-0250.

cs.CV↗
arXiv2026

Gungnir: Exploiting Stylistic Features in Images for Backdoor Attacks on Diffusion Models

Diffusion Models (DMs) have achieved remarkable success in image generation, yet recent studies reveal their vulnerability to backdoor attacks, where adversaries manipulate outputs via covert triggers embedded in inputs. Existing defenses, such as backdoor detection and trigger inversion, are largely effective because prior attacks rely on limited input spaces and low-dimensional triggers that are visually conspicuous or easily captured by neural detectors. To broaden the threat landscape, we propose Gungnir, a novel backdoor attack that activates malicious behaviors through style-based triggers embedded in input images. Unlike explicit visual patches or textual cues, stylistic features serve as stealthy, high-level triggers. We introduce Reconstructing-Adversarial Noise (RAN) and Short-Term Timesteps-Retention (STTR) to preserve trigger-consistent diffusion dynamics in image-to-image tasks. The resulting trigger-embedded samples are perceptually indistinguishable from clean images, evading both manual and automated detection. Extensive experiments show that Gungnir bypasses state-of-the-art defenses with an extremely low backdoor detection rate (BDR) and remains effective under fine-tuning-based purification, revealing previously underexplored vulnerabilities in diffusion models.

cs.CV↗
arXiv2026

The Intermediate-Mass Black Hole Reverberation Mapping Project: Stable Optical Continuum Lags of an IMBH in the Dwarf Galaxy NGC 4395 Over Years

NGC 4395 is a nearby dwarf spiral galaxy hosting an active galactic nucleus (AGN) powered by an intermediate-mass black hole (IMBH, $M_{\rm BH} \sim 10^{4}$--$10^{5}\,M_\odot$). Recent optical continuum reverberation mapping studies have suggested potential lag variations between different epochs, offering important clues to the physical mechanisms governing variability in the vicinity of the central black hole. We present continuous intranight multi-band photometric monitoring of NGC 4395 based on five nights of observations, including three nights from the Faulkes Telescope North (two of which are archival) and two new nights from Mephisto. This represents the first systematic investigation of optical continuum lag stability in a galaxy hosting a robustly confirmed IMBH. By applying difference-imaging techniques to both the new observations and the reprocessed archival data, we detect statistically significant optical inter-band lags of $\sim 5$--15 minutes, which increase monotonically with increasing wavelength. No obvious $u$-band lag excess is observed, implying a negligible fractional contribution from diffuse continuum (DC) emission to the optical continuum, in agreement with our spectral decomposition results. The inter-band lags remain stable over multi-year baselines. We suggest that this long-term lag stability may be related to the minor DC contribution, a relatively steady disk-corona structure, and the unusually high X-ray-to-optical luminosity ratio characteristic of low-luminosity AGNs, which likely allows X-ray reprocessing to dominate over other potential variability mechanisms. Future facilities like Gemini/SCORPIO, with its simultaneous optical-to-near-infrared coverage, will be ideally suited to play an important role in advancing this field.

astro-ph.GA↗
arXiv2026

Comparison of Effective Dissipation Channels in Warm Higgs Inflation from Warm Background Evolution

Within the framework of warm Higgs inflation, a systematic comparison is carried out among seven effective dissipation channels (EDC) constructed from combinations of the three basic dissipation channels, namely the low temperature (LT), high temperature (HT), and threshold (Th) channels. Adopting a unified treatment of warm background evolution, complexity penalization, and boundary consistency checks, the comparison is performed in terms of their distributions of the best fit points in ($n_s$, $r$) plane, relative BIC hierarchy, channel dominance patterns, and warmness indicators. The results show that, except for the pure HT EDC $Υ_{\mathrm{010}}$, the best fit points of the other six EDC are clustered within a small region of the ($n_s$, $r$) plane, around $n_s \approx 0.965$ and $r \approx (3.68 \to 3.74)\times10^{-3}$. In contrast, $Υ_{\mathrm{010}}$ is displaced from this main cluster, with a representative best fit point near $n_s = 0.9552$ and $r = 6.0\times10^{-3}$. Under both the unified scan and the 1200-point refined rescoring, the pure LT EDC $Υ_{\mathrm{100}}$ remains top-ranked, while $Υ_{\mathrm{011}}$ and $Υ_{\mathrm{111}}$ remain disfavored, indicating that the overall hierarchy is stable under the present boundary check criterion. Warmness diagnostics further show that $Υ_{\mathrm{100}}$ corresponds to $Q_* \approx 35.7$ and $T_*/H_* \approx 1.90\times10^{3}$, placing it in the strong warm regime, whereas $Υ_{\mathrm{011}}$ gives $T_*/H_* \approx 0.31$, already below the warmness threshold. The channel fractions, boundary checks, and constrained internal-mixing probes consistently indicate that the best fit points of the multi-channel EDC do not form a stable internally mixed region, but instead lie closer to a single channel dominated regime.

astro-ph.CO↗
arXiv2026

ShapE-GRPO: Shapley-Enhanced Reward Allocation for Multi-Candidate LLM Training

In user-agent interaction scenarios such as recommendation, brainstorming, and code suggestion, Large Language Models (LLMs) often generate sets of candidate recommendations where the objective is to maximize the collective utility of the entire set rather than individual candidates independently. However, existing reinforcement learning post-training paradigms, such as Group Relative Policy Optimization (GRPO), typically assign the same set-level scalar reward to every candidate in the set. This leads to noisy training signals where poor candidates free-ride on the high reward produced by a single strong peer, resulting in suboptimal exploration. To address this, we propose Shapley-Enhanced GRPO (ShapE-GRPO). By leveraging the permutation-invariant nature of set-level utility, we derive a Shapley-enhanced formulation from cooperative game theory to decompose set-level rewards into granular, candidate-specific signals. We show that our formulation preserves the fundamental axioms of the Shapley value while remaining computationally efficient with polynomial-time complexity. Empirically, ShapE-GRPO consistently outperforms standard GRPO across diverse datasets with accelerated convergence during training.

cs.AI↗
arXiv2026

JAL-Turn: Joint Acoustic-Linguistic Modeling for Real-Time and Robust Turn-Taking Detection in Full-Duplex Spoken Dialogue Systems

Despite recent advances, efficient and robust turn-taking detection remains a significant challenge in industrial-grade Voice AI agent deployments. Many existing systems rely solely on acoustic or semantic cues, leading to suboptimal accuracy and stability, while recent attempts to endow large language models with full-duplex capabilities require costly full-duplex data and incur substantial training and deployment overheads, limiting real-time performance. In this paper, we propose JAL-Turn, a lightweight and efficient speech-only turn-taking framework that adopts a joint acoustic-linguistic modeling paradigm, in which a cross-attention module adaptively integrates pre-trained acoustic representations with linguistic features to support low-latency prediction of hold vs shift states. By sharing a frozen ASR encoder, JAL-Turn enables turn-taking prediction to run fully in parallel with speech recognition, introducing no additional end-to-end latency or computational overhead. In addition, we introduce a scalable data construction pipeline that automatically derives reliable turn-taking labels from large-scale real-world dialogue corpora. Extensive experiments on public multilingual benchmarks and an in-house Japanese customer-service dataset show that JAL-Turn consistently outperforms strong state-of-the-art baselines in detection accuracy while maintaining superior real-time performance.

cs.CL↗
arXiv2026

SFCoT: Safer Chain-of-Thought via Active Safety Evaluation and Calibration

Large language models (LLMs) have demonstrated remarkable capabilities in complex reasoning tasks. However, they remain highly susceptible to jailbreak attacks that undermine their safety alignment. Existing defense mechanisms typically rely on post hoc filtering applied only to the final output, leaving intermediate reasoning steps unmonitored and vulnerable to adversarial manipulation. To address this gap, this paper proposes a SaFer Chain-of-Thought (SFCoT) framework, which proactively evaluates and calibrates potentially unsafe reasoning steps in real time. SFCoT incorporates a three-tier safety scoring system alongside a multi-perspective consistency verification mechanism, designed to detect potential risks throughout the reasoning process. A dynamic intervention module subsequently performs targeted calibration to redirect reasoning trajectories toward safe outcomes. Experimental results demonstrate that SFCoT reduces the attack success rate from $58.97\%$ to $12.31\%$, demonstrating it as an effective and efficient LLM safety enhancement method without a significant decline in general performance.

cs.CR↗
arXiv2026

New Constraints on Lorentz Invariance Violation at High Redshifts from Multiband of GRBs

In the gravity quantum theory, the quantization of spacetime may lead to the modification of the dispersion relation between the energy and the momentum and the Lorentz invariance violation (LIV). High energy and long-distance gamma-ray bursts (GRBs) observations in the universe provide a unique opportunity to test the possibility of LIV. In this work, we use 88 time delays from GRBs ($0.117 < z < 6.29$), and provide a cosmological model-independent approach based on the luminosity distance data from 174 GRBs to test LIV. Combining the observation data from multiband of GRBs provides us with an opportunity to mitigate the potential systematic errors arising from variations in the physical characteristics among diverse object populations, and to add a higher redshift dataset for testing the energy-dependent velocity caused by the corrected dispersion relationship of photons. These robust limits of the energy scale for the linear and quadratic LIV effects are $E_{\mathrm{QG},1} \ge 1.5\times 10^{15}$ GeV, and $E_{\mathrm{QG},2} \ge 8.5\times 10^{9}$ GeV, respectively. It exhibits a significantly reduced value compared to the energy scale of Planck in both scenarios of linear and quadratic LIV.

astro-ph.HE↗
arXiv2026

Adaptive controllable architecture of analog Ising machine

As a quantum-inspired, non-traditional analog solver architecture, the analog Ising machine (AIM) has emerged as a distinctive computational paradigm to address the rapidly growing demand for computational power. However, the mathematical understanding of its principles, as well as the optimization of its solution speed and accuracy, remain unclear. In this work, we for the first time systematically discuss multiple implementations of AIM and establish a unified mathematical formulation. On this basis, by treating the binarization constraint of AIM (such as injection locking) as a Lagrange multiplier in optimization theory and combining it with a Lyapunov analysis from dynamical systems theory, an analytical framework for evaluating solution speed and accuracy is constructed, and further demonstrate that conventional AIMs possess a theoretical performance upper bound. Subsequently, by elevating the binarization constraint to a control variable, we propose the controllable analog Ising machine (CAIM), which integrates control Lyapunov functions and momentum-based optimization algorithms to realize adaptive sampling-feedback control, thereby surpassing the performance limits of conventional AIMs. In a proof-of-concept CAIM demonstration implemented using an FPGA-controlled LC-oscillator Ising machine, CAIM achieves a twofold speedup and a 7\% improvement in accuracy over AIM on a 50-node all-to-all weighted MaxCut problem, validating both the effectiveness and interpretability of the proposed theoretical framework.

quant-ph↗
arXiv2026

Learning Shortest Paths When Data is Scarce

Digital twins and other simulators are increasingly used to support routing decisions in large-scale networks. However, simulator outputs often exhibit systematic bias, while ground-truth measurements are costly and scarce. We study a stochastic shortest-path problem in which a planner has access to abundant synthetic samples, limited real-world observations, and an edge-similarity structure capturing expected behavioral similarity across links. We model the simulator-to-reality discrepancy as an unknown, edge-specific bias that varies smoothly over the similarity graph, and estimate it using Laplacian-regularized least squares. This approach yields calibrated edge cost estimates even in data-scarce regimes. We establish finite-sample error bounds, translate estimation error into path-level suboptimality guarantees, and propose a computable, data-driven certificate that verifies near-optimality of a candidate route. For cold-start settings without initial real data, we develop a bias-aware active learning algorithm that leverages the simulator and adaptively selects edges to measure until a prescribed accuracy is met. Numerical experiments on multiple road networks and traffic graphs further demonstrate the effectiveness of our methods.

cs.LG↗
arXiv2026

S2ST-Omni: Hierarchical Language-Aware SpeechLLM Adaptation for Multilingual Speech-to-Speech Translation

Despite recent advances in speech-to-speech translation (S2ST), it remains difficult to achieve both high translation accuracy and practical flexibility. In this paper, we present S2ST-Omni, a compositional S2ST framework that integrates a high-accuracy speech-to-text translation (S2TT) frontend with a modular, plug-and-play text-to-speech (TTS) backend, enabling independent optimization of translation and synthesis. On the S2TT side, we introduce a hybrid adapter that follows a "local-then-global" strategy to bridge a pretrained Whisper encoder and a Qwen3 LLM, yielding a hierarchical acoustic-to-semantic abstraction. Building on this bridge, we further propose a hierarchical language-aware architecture that injects source-language information at two complementary levels. At the acoustic level, Language-Aware Dual-CTC operates on intermediate adapter features and employs FiLM-style feature modulation with a learnable gate, encouraging the model to learn language-specific but content-faithful acoustic representations. At the linguistic level, Language-Aware Prompting dynamically constructs source-language-conditioned prompts that activate language-specific translation knowledge in the LLM. To enable efficient optimization, we design a task-specific progressive fine-tuning strategy that first stabilizes speech-text alignment and then improves translation via LoRA on top of this converged foundation. The TTS backend remains fully modular and can be instantiated with any state-of-the-art synthesizer without retraining the S2TT frontend. Experiments on CVSS-C show that S2ST-Omni consistently achieves the best BLEU and ASR-BLEU across French, German, and Spanish to English directions, outperforming strong recent S2ST baselines.

eess.AS↗
arXiv2026

Wearable-informed generative digital avatars predict task-conditioned post-stroke locomotion

Dynamic prediction of locomotor capacity after stroke could enable more individualized rehabilitation, yet current assessments largely provide static impairment scores and do not indicate whether patients can perform specific tasks such as slope walking or stair climbing. Here, we present a wearable-informed data-physics hybrid generative framework that reconstructs a stroke survivor's locomotor control from wearable inertial sensing and predicts task-conditioned post-stroke locomotion in new environments. From a single 20 m level-ground walking trial recorded by five IMUs, the framework personalizes a physics-based digital avatar using a healthy-motion prior and hybrid imitation learning, generating dynamically feasible, patient-specific movements for inclined walking and stair negotiation. Across 11 stroke inpatients, predicted postures reached 82.2% similarity for slopes and 69.9% for stairs, substantially exceeding a physics-only baseline. In a multicentre pilot randomized study (n = 21; 28 days), access to scenario-specific locomotion predictions to support task selection and difficulty titration was associated with larger gains in Fugl-Meyer lower-extremity scores than standard care (mean change 6.0 vs 3.7 points; $p < 0.05$). These results suggest that wearable-informed generative digital avatars may augment individualized gait rehabilitation planning and provide a pathway toward dynamically personalized post-stroke motor recovery strategies.

cs.CE↗
arXiv2025

Fine-grained Preference Optimization Improves Zero-shot Text-to-Speech

Integrating human feedback to align text-to-speech (TTS) system outputs with human preferences has proven to be an effective approach for enhancing the robustness of language model-based TTS systems. Current approaches primarily focus on using preference data annotated at the utterance level. However, frequent issues that affect the listening experience often only arise in specific segments of audio samples, while other segments are well-generated. In this study, we propose a fine-grained preference optimization approach (FPO) to enhance the robustness of TTS systems. FPO focuses on addressing localized issues in generated samples rather than uniformly optimizing the entire utterance. Specifically, we first analyze the types of issues in generated samples, categorize them into two groups, and propose a selective training loss strategy to optimize preferences based on fine-grained labels for each issue type. Experimental results show that FPO enhances the robustness of zero-shot TTS systems by effectively addressing local issues, significantly reducing the bad case ratio, and improving intelligibility. Furthermore, FPO exhibits superior data efficiency compared with baseline systems, achieving similar performance with fewer training samples.

eess.AS↗
arXiv2025

Cosmological Model Independent Constraints on Lorentz Invariance Violation with Updated Gamma-Ray Burst Observations: An Artificial Neural Network Approach

Searching for Lorentz invariance violation (LIV) using astrophysical sources such as gamma-ray bursts (GRBs) is crucial for probing quantum gravity. However, the dependence of LIV constraints on assumed cosmological models has been largely overlooked. In this work, we present a model-independent reconstruction of the cosmic expansion history using artificial neural networks (ANN), thereby avoiding biases from specific cosmological priors. We analyze 74 GRB time delays, including 37 measurements from GRB~160625B across multiple energy bands at $z = 1.41$, and 37 additional bursts spanning redshifts $0.117 \leq z \leq 1.99$. Our analysis yields stringent constraints on both linear and quadratic LIV, with $E_{\mathrm{QG},1} \geq 2.60 \times 10^{15}~\mathrm{GeV}$ and $E_{\mathrm{QG},2} \geq 1.21 \times 10^{10}~\mathrm{GeV}$. The linear limit is within four orders of magnitude of the Planck scale. By leveraging a large sample of GRBs, our approach significantly enhances the robustness of LIV constraints, providing a powerful, cosmological-independent framework for future tests of quantum gravity.

astro-ph.HE↗
arXiv2025

SN 2024aecx: A double-peaked rapidly evolving Type IIb supernova at 11 Mpc

We present the results of low-resolution spectroscopic and densely sampled multi-band photometric follow-up of supernova (SN) 2024aecx. The SN was discovered in the spiral galaxy NGC 3521 (distance $\sim$11 Mpc) within a day after the explosion. The early spectra of SN 2024aecx show a weak signature of hydrogen lines, which disappeared in $\sim$30 days after the explosion. Light curves in all bands show a distinct feature of two peaks, and the first peak is likely due to the shock cooling emission. The early phase light curve evolution of SN 2024aecx has similarity with the typical Type IIb events, but the decay rate in different bands (e.g., $\rm Δm_{15}$ = 1.60 $\pm$ 0.05 mag, $g$-band) is significantly faster in the post-peak phase. It attained the secondary maximum in $\sim$19 days ($g$-band) with a peak absolute magnitude of M$_{g}$ = -17.94 $\pm$ 0.10 mag. SN 2024aecx colors trend redder in early epochs ($<$8 days), followed by a duration in which it grows bluer, then later gets redder again $>$20 days after explosion. The analytical model fitting to the light curves reveals an envelope mass and progenitor radii in the range of $\sim$0.03 - 0.24 $M_\odot$ and $\sim$169 - 200 $R_\odot$, respectively. Modeling of the pseudo-bolometric light curve suggests that synthesized $^{56}$Ni in the explosion was $\sim$0.15 M$_{\odot}$ with ejecta mass and kinetic energy of $\sim$0.7 M$_{\odot}$ and $\sim$0.16 $\times$ 10$^{51}$ erg, respectively. The observational properties and modeling indicate that the SN~2024aecx progenitor belongs to the extended progenitor category.

astro-ph.HE↗
arXiv2025

What Matters in Data for DPO?

Direct Preference Optimization (DPO) has emerged as a simple and effective approach for aligning large language models (LLMs) with human preferences, bypassing the need for a learned reward model. Despite its growing adoption, a fundamental question remains open: what characteristics of preference data are most critical for DPO performance? In this work, we provide a systematic study of how preference data distribution influences DPO, from both theoretical and empirical perspectives. We show that the quality of chosen responses plays a dominant role in optimizing the DPO objective, while the quality of rejected responses may have relatively limited impact. Our theoretical analysis characterizes the optimal response distribution under DPO and reveals how contrastiveness between responses helps primarily by improving the chosen samples. We further study an online DPO setting and show it effectively reduces to supervised fine-tuning on the chosen responses. Extensive experiments across diverse tasks confirm our findings: improving the quality of chosen responses consistently boosts performance regardless of the quality of the rejected responses. We also investigate the benefit of mixing the on-policy data. Our results interpret the mechanism behind some widely adopted strategies and offer practical insights for constructing high-impact preference datasets for LLM alignment.

cs.LG↗
arXiv2025

QiNN-QJ: A Quantum-inspired Neural Network with Quantum Jump for Multimodal Sentiment Analysis

Quantum theory provides non-classical principles, such as superposition and entanglement, that inspires promising paradigms in machine learning. However, most existing quantum-inspired fusion models rely solely on unitary or unitary-like transformations to generate quantum entanglement. While theoretically expressive, such approaches often suffer from training instability and limited generalizability. In this work, we propose a Quantum-inspired Neural Network with Quantum Jump (QiNN-QJ) for multimodal entanglement modelling. Each modality is firstly encoded as a quantum pure state, after which a differentiable module simulating the QJ operator transforms the separable product state into the entangled representation. By jointly learning Hamiltonian and Lindblad operators, QiNN-QJ generates controllable cross-modal entanglement among modalities with dissipative dynamics, where structured stochasticity and steady-state attractor properties serve to stabilize training and constrain entanglement shaping. The resulting entangled states are projected onto trainable measurement vectors to produce predictions. In addition to achieving superior performance over the state-of-the-art models on benchmark datasets, including CMU-MOSI, CMU-MOSEI, and CH-SIMS, QiNN-QJ facilitates enhanced post-hoc interpretability through von-Neumann entanglement entropy. This work establishes a principled framework for entangled multimodal fusion and paves the way for quantum-inspired approaches in modelling complex cross-modal correlations.

cs.LG↗
arXiv2025

SN 2024gy: Multi-epoch Spectroscopic Features Suggestive of Delayed Detonation in a Type Ia Supernova

We present photometric and spectroscopic observations of SN 2024gy, a Type Ia supernova (SN Ia) exhibiting high-velocity features (HVFs) in its early-time spectra. This SN reaches a peak $B$-band magnitude of $-19.25 \pm 0.29$ mag and subsequently declines by $Δm_{15}(B) \approx 1.12$ mag, consistent with the luminosity-width relation characteristic of normal SNe Ia. Based on the peak thermal luminosity of $(1.2 \pm 0.3) \times 10^{43}$ erg s$^{-1}$, we estimate that $0.57 \pm 0.14~\rm M_{\odot}$ of $^{56}$Ni was synthesized during the explosion. Our dense early spectral monitoring revealed significant velocity disparities within the ejecta. Notably, absorption features from the Ca II near-infrared triplet were observed at velocities exceeding 25,000 km s$^{-1}$, while the Si II $λ$6355 line velocity at the same epoch was significantly lower at $\sim$ 16,000 km s$^{-1}$. This velocity disparity likely reflects distinct ionization states of intermediate-mass elements in the outermost layers. The prominent Ca II HVFs may originate from ionization suppression within the highest-velocity ejecta, potentially indicative of minimal hydrogen mixing in a delayed-detonation explosion scenario. Additionally, the Ni/Fe ratio derived from the nebular spectrum of SN 2024gy provides further support for this model.

astro-ph.HE↗