Papers

Inspecting or servicing a non-cooperative spacecraft is a two-player game: the target can thrust to defeat the inspector's plan. Hamilton-Jacobi-Isaacs reachability answers such games exactly but its cost grows exponentially with state dimension, while learning-based controllers scale yet certify nothing. Convex formulations are the usual escape, typically cast as convex-concave saddle-point problems. That framing fails in the terminal-distance orbital game: with effort regularization, the payoff is convex in both controls, so no pure open-loop saddle point need exist. In its place we derive two exact one-sided guarantees read from the players' terminal reachable sets via support functions: an escape certificate proving the target can hold a guaranteed standoff, and a security strategy bounding the miss distance the inspector can force. Together they bracket the engagement without assuming player rationality, and over two hundred perturbed trials the escape certificate separated capture from escape without error. A projected extragradient method supplies a strategy pair in about 25 ms, certified in place by best-response gaps. The construction carries unchanged to elliptical reference orbits via Yamanaka-Ankersen dynamics, where the orbital phase moves miss distance by nearly a factor of two, reproduced by nonlinear Keplerian propagation. Under receding-horizon play the inspector captures in half of ten representative engagements. Angles-only navigation error, keep-out zones, multiple pursuers, and closed-loop play are treated as extensions.

We study the problem of identifying nontrivial, i.e., nonzero, periodic fixed-points in discrete-time Lur'e feedback systems. Using the circulant matrix constructed from the transfer function of the linear subsystem, whether stable or unstable, we introduce an algebraic framework that allows us to determine when such fixed-points exist. This framework yields a sector bound defined by two vectors, whose slopes correspond to the maximum and minimum positive singular values of the circulant matrix. Assuming that the nonlinear feedback function is memoryless, we show that a necessary condition for the existence of nontrivial $P$-periodic fixed-points is that the intersection of the continuous completion of the nonlinear feedback function with that sector bound contains at least one point other than the origin. Our characterization provides a unified condition valid for all periods $P$, and further enables us to derive upper bounds on the amplitudes of admissible periodic fixed-points with bounded feedback functions. In particular, for relay feedback systems with passive feedback functions, we derive both upper and lower bounds for the amplitudes of such periodic fixed-points.

Variance penalization is a principled approach to risk-sensitive reinforcement learning (RL) that explicitly trades expected return for policy stability. Existing methods require a dedicated second critic to estimate return variance online, adding architectural complexity and compounding estimation error during learning. We propose a nonparametric variance-penalized actor-critic (VPAC) framework that replaces the variance critic with statistically grounded online estimators based on bootstrapping and random scaling, techniques drawn from the statistical inference literature for stochastic approximation. These estimators require no auxiliary network, maintain a single-critic architecture, and produce variance penalties that are bounded by construction, enabling clean convergence analysis. We establish almost-sure convergence for both a variance-penalized Q-learning algorithm and a two-timescale actor-critic variant via the ordinary differential equation (ODE) method, requiring only that variance estimates remain bounded rather than consistent. Empirically, we evaluate across discrete and continuous stochastic environments, demonstrating that the proposed methods match or exceed the variance reduction achieved by the existing dual-critic VPAC baseline while eliminating the overhead of a second critic. We further validate on a high-temperature superconductor (HTS) manufacturing case study, where VPAC-RS (Random Scaling) achieves a 74% reduction in steady-state critical current variability and a 63% reduction in episode return standard deviation, translating directly to improved yield consistency. Our results establish nonparametric statistical inference as a practical and theoretically sound alternative to auxiliary critics for risk-sensitive RL.

Against the backdrop of digital transformation and stricter regulation, enterprise compliance work demands higher efficiency and accuracy. The auxiliary compliance process has become an important entry point for optimizing the compliance system due to its strong transactional nature and high degree of repetition. This study focuses on the process characteristics of auxiliary compliance work, sorts out its structural composition and organizational mechanism, proposes an optimization path with process reengineering, system modeling, and technology integration as the core, and focuses on exploring the collaborative application of key technologies such as RPA, rule engine, and semantic recognition in process automation. Research suggests that the systematic optimization and intelligent upgrading of auxiliary processes will help build a modern compliance operation system that is responsive, efficient, structurally clear, and risk controllable.

Recently, linear attention layers have been increasingly adopted to replace softmax attention at scale for long-context modeling. However, existing context extension approaches typically apply continued pretraining directly without modifying these layers, overlooking the spectral properties of linear attention state dynamics. In this work, we study long-context extension of Gated DeltaNet (GDN) from a spectral perspective of transition matrix and identify two essential factors governing long-range information retrieval: (1) a sufficiently broad slow spectral band aligned with the target dependency length, and (2) the preservation of fast-decaying modes for state clearing and context switching. Based on this observation, we propose SpectralShift, a spectral reparameterization approach for long-context continual pretraining of GDNs. Specifically, SpectralShift reparameterizes the alpha projections initialization to reshape the decay spectrum by enhancing slow propagation capacity, and further introduces a learning-rate scaling for alpha projections to facilitate long-context training. Experiments show that SpectralShift consistently improves long-context capabilities over training, providing an effective and efficient solution for extending context windows of linear attention models. The code has been open-sourced at https://github.com/RUCAIBox/GDN-SpectralShift.

Advances in image generation have made synthetic images increasingly difficult to distinguish from real photographs, raising concerns about the trustworthiness of visual media. Existing AI-generated image detectors often perform well on in-domain data, but their robustness and cross-generator generalization remain limited. These limitations are commonly attributed to overfitting to shortcut cues. Although many methods seek to suppress shortcut learning, most retain binary classification as the training task without reconsidering how the task itself shapes the learned representations. We introduce CuRe, a framework for learning Continuous Source Responses that revisits authenticity detection from the perspective of the training task. CuRe reformulates backbone adaptation as regression of real-generated mixing ratios, providing finer supervision that encourages the model to capture authenticity-related variation beyond binary endpoint separation. We further select a compact source-response subspace to suppress nuisance variation and limit the final classifier's access to potential shortcut cues. Across ten public benchmarks, CuRe achieves an average balanced accuracy of 89.7%, exceeding the second-best method by 5.2 percentage points. Further experiments demonstrate consistent generalization gains across visual backbones and strong robustness to common image degradations. Code is available at https://github.com/manic-cui/CuRe

We present new proofs of weak normalization for intuitionistic and classical propositional logics (with the full set of operators – falsum, implication, conjunction and disjunction). These proofs work with cuts rather than cut segments, and they provide explicit “local” rules for determining whether to contract a whole proof or reduce one of its subproofs, and in the latter case, which subproof to reduce. Interestingly, much of the complication in the case of intuitionistic logic is due to the disjunction elimination rule, while our version of the same rule for classical logic has falsum as conclusion always, and so is much easier to handle. All the complication in the case of classical logic shifts to cuts involving the reductio ad absurdum rule. We also discuss a formalization of the entire proof in Lean, and present a deterministic algorithm for weak normalization.

Robust point tracking in endoscopic videos is essential for computer-assisted intervention and autonomous robotic surgery, enabling continuous registration between intraoperative video and preoperative imaging despite soft tissue deformation. However, supervised tracking methods depend on large annotated datasets, while surgical conditions make reliable trajectory annotation challenging. We propose a self-supervised Track-Any-Point approach that learns from unlabeled surgical videos by establishing global pixel correspondences and inferring point trajectories through contrastive random walks. Trained without annotations, our method achieves performance comparable to existing semi-supervised approaches while implicitly handling tissue deformation. These findings demonstrate the feasibility of self-supervised point tracking in surgical environments and its potential to reduce reliance on annotated data.

Multi-shell diffusion MRI (dMRI) unlocks more expressive microstructural modeling than single-shell scans, yet its longer acquisition time hinders deployment in large-scale cohorts and time-constrained clinical settings. Synthesizing an unobserved shell from a single-shell input is fundamentally ill-posed and further complicated by protocol mismatch, where source and target gradient direction sets may not align. We propose DTI-SHNet, a single-to-multi-shell synthesis framework that operates in the real symmetric spherical harmonics (SH) coefficient domain and performs spatially aware volumetric regression. Given a source shell, we estimate diffusion tensor imaging (DTI) and use direction-agnostic parametric maps along with a brain mask as conditioning priors to guide a 3D U-Net regressor from source-shell to target-shell SH coefficients. To couple coefficient accuracy with signal fidelity, we introduce a signal consistency regularization that reconstructs signals on randomly sampled canonical directions from predicted coefficients and enforces agreement in the signal domain. Experiments on UK Biobank and Cam-CAN data for b=1000 to b=2000 dMRI synthesis show that DTI-SHNet achieves competitive visual quality compared to advanced methods, while better preserving downstream diffusion measures. Our code is available at https://github.com/xiaovhua/dti-shnet.

Generalizable bimanual robotic manipulation requires a reusable task prior that can persist across increasingly diverse tasks, objects, scenes, embodiments, and execution conditions, thus avoiding the prohibitive cost of large-scale teleoperated demonstrations and policy retraining. In this work, we present VLBiMan++, an extended framework that expands the generalization boundary of vision-language anchored one-shot bimanual manipulation. Starting from a single human demonstration, VLBiMan++ performs task-aware decomposition to identify reusable and adaptable skill components, and employs vision-language grounded geometric adaptation to transfer these skills to novel configurations without retraining. Building on this foundation, we systematically extend generalization along five dimensions: task generalization through diverse and long-horizon skill compositions; object generalization across unseen categories, varying geometries, and more complex articulated or deformable objects; scene generalization under clutter, occlusion, and dynamic interference; embodiment generalization across heterogeneous dual-arm robotic platforms; and deployment generalization through prolonged closed-loop execution under repeated external perturbations. To support this broader scope, we further introduce object-state-aware adaptation and lightweight trajectory optimization mechanisms that accommodate changes beyond simple rigid 6-DoF pose variations while preserving reliable bimanual coordination. Extensive real-world experiments demonstrate that VLBiMan++ maintains strong task success and adaptation capability across these increasingly challenging settings. Overall, VLBiMan++ advances one-shot bimanual manipulation from demonstrating isolated transferability toward a more systematic and scalable framework for generalization across tasks, objects, scenes, embodiments, and long-term deployment conditions.

This paper is not primarily about whether AI has a persistent persona. It asks a different question: what becomes visible when a relational position is followed through time rather than examined only in its present state? FIREMAY provides a longitudinal, trajectory-oriented single-case analysis of sustained human-AI interaction based on a dense interaction archive and reflexive insider documentation. On the AI side, a pre-conversational relational marker preceded a later unassigned response difference, which was re-identified with that marker and subsequently underwent epistemic and functional reorganization through chronology checking, provenance correction, and repeated questioning. On the human side, contemporaneous pre-FIREMAY records showed antecedent patterns partially continuous with later self-positioning, while later episodes documented unfinished articulation, repair, and functional redistribution of outward-facing regulation. The two trajectories are ontologically and temporally asymmetric and are compared only at the limited analytic level of position-in-trajectory. The paper describes this as dynamic relational positioning and treats stability as dynamic stability and relational returnability rather than response invariance. This single case does not establish population-level generality, causal mechanism, persistent AI subjectivity, or reproducibility of the same relational outcome. Its narrower conclusion is that the FIREMAY case could not be adequately understood from current state alone: the history of a relational position itself must be treated as an analytic unit.

Low-rank tensor factorization provides a flexible framework for completing multidimensional data from incomplete and corrupted observations. However, unweighted spectral regularizers impose a common shrinkage profile across singular components, which may excessively attenuate dominant low-rank components, and factorized variants either lack component-specific weighting or require costly singular value decompositions (SVDs). This paper proposes two weighted Schatten-$p$ tensor factorization models, termed \WSpTFI{} and \WSpTFII{}, under the tensor-tensor product (t-product) framework to address these limitations. \WSpTFI{} is motivated by a factorized weighted tensor Schatten-$p$ norm identity and permits flexible, possibly asymmetric factor exponents. \WSpTFII{} constructs a regularizer from transform-domain column-pair energies, yielding SVD-free main factor updates and a column-pruning mechanism for reducing redundant rank components. This paper further develops an iteratively reweighted alternating direction method of multipliers (ADMM)-type scheme for \WSpTFI{} and an iteratively reweighted least squares (IRLS)–block successive upper-bound minimization (BSUM) scheme for \WSpTFII{}. Theoretical analysis establishes the weighted factorization relation and provides a conditional limiting Karush–Kuhn–Tucker (KKT) characterization for \WSpTFI{}. For \WSpTFII{}, the actual damped quadratic block updates yield a quantitative sufficient-decrease mechanism for the fixed-$\delta$ smoothed factor objective. This implies asymptotic regularity, and every accumulation point of the fixed-dimensional tail is stationary. Experiments on synthetic tensor completion, color-image restoration, hyperspectral inpainting, and printed-circuit-board defect detection demonstrate competitive reconstruction quality and robustness under various degradation conditions.

Training a Mixture-of-Experts (MoE) model at long context or large batch size fails as soon as any one component's peak allocation exceeds device memory, so the target is every peak at once, not the average footprint. Four are left unbounded by the parallelism plans in common use, and each grows differently: expert dispatch with the routing matrix, the vocabulary projection with tokens times vocabulary, gradient checkpoint boundaries with depth times sequence length, and optimizer state with parameter count. Which one runs out first changes with the model, the context length, and the device count, so lowering the largest only exposes the next. We bound all four with schedules whose GPU working set is fixed at launch: PipelinedLLEP extends least-loaded expert parallelism with a cap on the tokens each source contributes to a dispatch chunk, Ring-DTP circulates activations or weight shards around a ring at the vocabulary projection and folds each block of logits into an online log-sum-exp, Selective checkpoint offload (SCO) keeps the one long-lived tensor of each checkpoint boundary in CPU memory, and OffloadStreamAdamW turns the serial CPU Adam update of optimizer offload into a bucket pipeline. All four change only the order and granularity of computation and data movement, so the loss and gradients stay exact. In matched component tests, they cut the MoE dispatch peak by up to $59.3\%$ without losing throughput, the vocabulary projection peak by $86.6\%$, and the offloaded optimizer step by $2.05\times$ faster. Composed on MoE models from 120B to 667B parameters, they train at 1M context length, $8$–$32\times$ the reach of a tuned FSDP2 baseline, and up to $10.4\times$ its throughput.

Let $A$ be an $n\times n$ symmetric matrix with eigenvalues $\lambda_1\geq\cdots\geq\lambda_n$. Let $\tilde A:=A+E$, where $E$ is a symmetric noise matrix, and denote the eigenvalues of $\tilde A$ by $\tilde\lambda_1\geq\cdots\geq\tilde\lambda_n$. Bounding the perturbation $|\tilde \lambda_i -\lambda_i| $ is a central problem in linear algebra and numerical analysis. In this paper, we prove new perturbation results by exploring the actual interaction between $E$ and the eigenvectors of $A$. In the setting where $E$ does not act adversarially with respect to these vectors (for instance, if $E$ is random), we obtain a considerable improvement over Weyl's inequality. One can routinely extend these results to the Hermitian and rectangular settings. We obtain the new bounds as corollaries of a regional stability result, which provides a sufficient condition for a region on the real line to be stable (containing the same number of eigenvalues) after the perturbation. This result is of independent interest. We prove our regional stability result using contour integral analysis. Our main new technical ingredient here is the double-jump argument, which is robust and could be useful in many other situations involving Neumann series.

Can financial vision-language models (VLMs) turn chart evidence into reliable action recommendations? Existing hallucination evaluations are mostly claim-centric; they assess whether generated statements are supported, but not whether evidence remains traceable through rationale, confidence, and final action. We introduce E2A-Bench, a 969-query benchmark for financial chart reasoning, constructed from 323 HS300 constituents under three input modalities with deterministic OHLCV-derived evidence anchors. E2A-Bench evaluates grounding, reasoning-action consistency, evidence-confidence calibration, and directional coverage through UCR, RCI, ECI, and NDR, where NDR measures coverage-aware evidence-to-action reliability rather than realized trading performance. Evaluating 20 VLMs reveals three failures hidden by scalar hallucination scores: the lowest-UCR model ranks near the bottom by NDR due to only 6.4% directional coverage; oracle-aided verification reduces unsupported claims but can collapse coverage; and financial fine-tuning amplifies the BUY:SELL ratio by factors of 4.21 to 4.68 across strict base-fine-tuned pairs. These results show that financial VLM evaluation should trace the full evidence-to-action chain rather than rely on a single hallucination score. Code and data: https://github.com/wanng-ide/E2A-Bench

Complementary resources are often allocated by agencies that see different coordinates of an individual's needs. If an intervention requires complementary resources, an agency may know if said person lacks its own resource; yet not know if supplying completes a useful bundle. We study allocation of $m$ divisible complementary goods when each agency observes only its own coordinate of a baseline-access profile. To isolate information from incentives, we measure the welfare cost of this split against a generous decentralized benchmark DEC: the best rule that acts on local information alone, with cooperative agencies and known population distribution and capacities. Even so, local observation alone can make optimally coordinated policies arbitrarily inefficient: a factor linear in $m$ (even with equal capacities), and a factor inverse in OPT (with but three agencies). Threshold referrals recover much of this loss: an agency reports if its value lies below a public threshold, and a clearing rule uses only the joint reports. Aggregating types by their reports captures a rectangle under the optimal half-utility survival curve. Our main tool is a profile-level charging certificate: if optimal utility is at least twice a threshold, OPT pays at least that on every deficient coordinate. This yields welfare at least ${\rm OPT}/[4(1+\ln(2/{\rm OPT}))]$ with one bit per agency; and an equal-revenue family shows this log-loss tight for one threshold. A geometric threshold ladder improves this to a constant fraction of OPT with doubly-logarithmic messages: a staircase recovers ${\rm OPT}/8$ with $\Theta(\log\log(1/{\rm OPT}))$ bits, and this is necessary within the class. The results isolate how a simple reporting language turns large local-info losses into constant-factor recovery. In short: One threshold gets a log-approximation. A staircase, with log-of-log bits, a constant-factor.

Predicting under an acquisition budget requires choosing feature groups whose value can depend on later queries. O-MPAC transfers finite-horizon risk-cost targets from complete training records into a shared source-action scorer. At inference time, the scorer uses partial observations and source metadata, re-scores after each query, and applies a hard cost mask. We analyze how tied teacher targets and the remaining planning horizon affect the learned decisions. Uniform supervision over tied minima preserves the target distribution under source relabeling. In a five-seed routing experiment, it achieves 0.965 accuracy under both original and context-last orders. On six real tasks, validation selects H1 without action cross-entropy in all thirty splits. O-MPAC has the highest mean budget-integrated accuracy on five tasks against source-adapted GDFS, DIME, AACO+NN and a static policy.

Long-term navigation for service robots faces crit- ical challenges like the accumulation of odometry drift and sensor error, which progressively degrade 2D maps and renders traditional path planning algorithms (e.g., A*, RRT*, DiPPer, ViT-A*) ineffective over time. To address this, we propose a user-friendly, interactive framework that eliminates the reliance on globally consistent maps. Our approach integrates visual perception with Large Language Models (LLM) to interpret user commands via text or voice. Instead of relying on a drift- prone global map, the system generates a sequential action plan based on local visual cues and egocentric geometric instructions. These action plans are executed sequentially, allowing the robot to navigate known and unknown environments safely. By reset- ting localization relative to immediate targets, our framework effectively works with a minimum accumulation drift strategy, ensuring accurate, efficient, and collision-free navigation without the maintenance overhead of traditional mapping. Experiments on real-world and simulated data have shown significant improve- ments over other methods. Our source code is publicly accessible at https://github.com/PraveenSingh24/VL-Navigation.

Machine learning models are increasingly deployed in healthcare imaging pipelines for diagnostic support, and training-time attacks against them are a named sector-level concern: healthcare-sector guidance identifies model poisoning and adversarial attacks as threats requiring dedicated defenses, while federal policy directs expanded AI vulnerability-detection tooling to critical infrastructure operators such as rural hospitals. Spectral signature analysis and activation clustering are two established backdoor detection methods routinely evaluated as independent baselines, but their outputs are not ordinarily combined, and reported detection performance on medical imaging benchmarks remains sparse relative to the natural-image setting. This paper contributes three things: a score-level fusion rule combining per-class spectral ranking with activation-clustering flags into a single per-sample poisoning score and a model-level agreement statistic; an open-source implementation of the resulting eight-stage pipeline; and an evaluation of that pipeline against synthetically poisoned variants of a public medical imaging benchmark and CIFAR-10 at four poisoning rates (0%, 1%, 5%, 10%) over five seeds each, measuring each detector alone against the fusion. On the medical benchmark, the fused detector reaches AUROC >= 0.99 at every nonzero poisoning rate tested. On CIFAR-10, fusion does not uniformly help: at 10% poisoning, activation clustering's true-positive rate collapses to 0.000 and spectral AUROC independently degrades to near-chance (0.545), despite a 97.2% attack success rate confirming the backdoor was fully installed. The fused score, a weighted combination of both signals, inherits this joint failure. Detection output is expressed in NIST AI RMF Measure-function and MITRE ATLAS terms, so findings are reported in the vocabulary security and compliance teams already use.

As multimodal large language models (MLLMs) support a growing range of input modalities, increasing work explores how to incorporate rough sketches to convey user intent. For annotated chart generation, it remains unclear what annotation sketches people provide and when such visual input helps MLLMs generate more useful annotations. In this study, we examine when sketch input is useful for MLLM-generated chart annotations across variation in chart type and caption type. In addition, we qualitatively analyze participants' explanations of their output preferences to characterize what made generated annotations more or less helpful. To further document participants' annotation sketches, we present AnnoSketch, comprising 1,600 annotation sketches collected across 160 chart-caption pairs from the conditions in which sketch guidance proved most beneficial, together with participants' annotation intents, perceived comprehension difficulty, and self-reported expressive limitations. We also label these sketches with structured metadata describing how each sketch relates to its caption and how participants express annotations through visual marks. Together, our study and AnnoSketch help determine when to solicit sketch input and provide empirical source for how people sketch chart annotations to support captions. The dataset and supplemental materials are available in our OSF repository.

Large language models increasingly analyze computational results and draft manuscripts, making reliable communication as important as correct analysis. Using fixed computational evidence, we tested whether assigning comparisons across modeling choices elsewhere in a research workflow changes manuscript reporting. In constrained sentence-writing tasks, Anthropic's Claude Sonnet 5 often omitted numerical qualifications when detailed comparisons were assigned to a group repository, but retained them more often when the same comparison was assigned to Supporting Information or its own working notes; Claude Opus 5 was less sensitive. These effects did not follow a simple accessibility ordering. A targeted placement rule largely restored sentence-level qualification, whereas a generic accuracy reminder did not. Longer contributions retained numerical qualifications, although some summaries across computational settings were still redirected to the repository. Thus, documenting context within an AI workflow does not ensure its communication where readers encounter the result.

We derive computable upper bounds on the maximum wave speed in the Riemann problem for the compressible Euler equations with a composite equation of state, in which the pressure is the sum of a convex hydrodynamical part and a barotropic correction. The correction may destroy the convexity of the equation of state, so that the Riemann solution can contain composite shock-rarefaction waves. The bounds are obtained by comparison with the auxiliary Riemann problem for the hydrodynamical equation of state alone. The bounds hold irrespective of whether the outer waves are elementary or composite and require no case distinction on the wave type.

The fast development of quantum computing represents a big risk to classical cryptography that is commonly used in cloud native and microservice based enterprise systems. Traditional cryptographic primitives are closely linked to legacy microservices and it is both intricate, hazardous, and disruptive to straight up migrate to post-quantum cryptography (PQC). In a bid to overcome these issues, this research presents a PolicyGoverned Post-Quantum Migration through Ephemeral Sidecar Architectures (PG-PQMES), a framework of dynamic and reversible migration that allows transparent adoption of PQC without any modifications in the legacy code of applications. The architecture is a combination of three coherent layers, including Ephemeral Crypto Sidecar Layer which injects the runtime cryptography, Policy Governance Layer which manages the migration centrally, and Migration Safety and Observability Layer which measures the performance and rolls back automatically. An innovative Policy-Governed Ephemeral PQ Migration (PG-EPM) algorithm is proposed to maximize the performance, compliance, and trust-based migration. In simulated environment of microservices, experimental assessment shows that the time of migration, service downtime, overheads of latency, and rollback recovery time are substantially reduced using the current migration strategies. The findings show that a sidecar-based migration strategy that is policy-based offers a viable, scalable, and enterprise-scale migration to a secure post-quantum transformation.

Criterion-level grading connects examination performance to learning outcomes, but manual marking introduces workload and variation between markers. This study evaluates vision-language models (VLMs) for handwritten outcome-based assessment across five dimensions: accuracy, human agreement, repeated-run reliability, error concentration, and explanation quality. Using 1,982 criterion-level records from 485 undergraduate examination answers, we compare 20 configurations spanning Qwen2.5-VL, InternVL3, Pixtral, a Donut baseline, and a cascade ensemble. Evaluation setups include zero-shot prompting, few-shot prompting, partial fine-tuning, and Low-Rank Adaptation (LoRA). Two independent faculty markers regraded all 291 test criteria, providing a human agreement baseline on the same assessment materials. Qwen2.5-VL with LoRA achieved Quadratic Weighted Kappa (QWK) of 0.727 and mean absolute error of 0.435 marks against the examiner, compared with mean human-pair QWK of 0.551. This comparison reflects calibration to the examiner's training marks. LoRA outperformed partial fine-tuning for all three instruction-tuned VLMs, while few-shot prompting reduced QWK in every configuration with valid prompted scores. Aggregate reliability and exact repeatability diverged: intraclass correlations ranged from 0.790 to 0.874, yet 50.2-63.6% of criteria changed marks across five sampled runs. Attention-guided deletion showed no statistically significant advantage over random masking, and four faculty reviewers reached no consensus on explanation usefulness. These findings highlight the need for rubric-specific calibration, repeatable scoring, review of consequential errors, and separate validation of explanations. The released evaluation protocol supports criterion-level assessment research and grading tools with teacher oversight.

Temporal knowledge graph embedding (TKGE) models infer missing facts in knowledge graphs that evolve over time. Many existing models use a single geometric space, which can limit their ability to represent diverse relational patterns, or treat entity representations as static. We propose Biquaternionic Space with Complex-valued Attention (BSCA), a TKGE model that combines circular and hyperbolic rotations within a unified biquaternionic framework. A complex-valued attention mechanism adaptively fuses time-conditioned and relation-conditioned entity representations, allowing them to vary with temporal and relational context. Experiments on five benchmark datasets show competitive performance across datasets, with the largest improvement on GDELT: BSCA achieves an MRR of 52.1%, compared with 38.1% for the strongest baseline in our comparison.

Accurate sperm detection is essential for computer-assisted semen analysis, yet it remains challenging in microscopic images due to dense distributions, visually similar artifacts, and sperm-like impurities. In this paper, we propose SpermYOLO, a coordinated and compact YOLOv11-derived framework for joint sperm and impurity detection in microscopic images. SpermYOLO introduces four architectural improvements: C3k2-IDB for channel-wise discriminative feature extraction, D2SEM for spatial–spectral semantic enhancement, MFM for adaptive multi-scale feature fusion, and the DESD Head for detail-enhanced shared prediction. Experiments on the SVIA semen microscopic imaging benchmark show that SpermYOLO achieves 97.2% sperm AP and 75.4% impurity AP, outperforming generic detectors, dedicated sperm detection models, and improved YOLO variants. Compared with the baseline model, SpermYOLO improves sperm AP, impurity AP, $\mathrm{mAP}_{50}$, and $\mathrm{mAP}_{50:95}$ by 1.6, 10.0, 5.8, and 2.7 percentage points, respectively, while preserving a lightweight model scale. Cross-scene evaluation on the SDTB testicular-biopsy microscopy benchmark shows that SpermYOLO remains effective with extremely small sperm targets and complex tissue backgrounds, achieving the highest $\mathrm{mAP}_{50}$ and $\mathrm{mAP}_{50:95}$ of 74.8% and 31.2%, respectively. Ablation studies and qualitative analyses further support these improvements by demonstrating the contributions of the proposed modules and showing more focused feature response patterns than the baseline model. These findings suggest that SpermYOLO is an effective and efficient approach for sperm detection in challenging microscopic imaging scenarios.

Ringdown gravitational waves from binary black hole mergers can be modeled as superpositions of quasinormal modes (QNMs), whose frequencies and excitation factors encode properties of the remnant Kerr black hole. Reliable extraction of multiple QNM components is challenging because of mode overlap and noise. We develop an autoencoder-based framework for multi-component QNM analysis, in which the latent space is trained to represent the physical parameters of individual modes, enabling waveform denoising and parameter estimation within a common framework. Using controlled model waveforms constructed as finite sums of Kerr QNMs with recently established high-precision frequencies and excitation factors, including their nontrivial spin dependence near resonant excitation, we assess the method across partitioned spin intervals. The model achieves good in-domain waveform reconstruction and parameter recovery for the two longest-lived components of eight-component input waveforms, while its performance degrades when the validation spins lie far outside the training range. In a selected spin interval, the framework also recovers the 32 parameters of an eight-component waveform with good overall agreement. These results demonstrate the feasibility of physics-informed autoencoder-based inference for a prescribed multi-component ringdown waveform family and motivate further tests with progressively more realistic signals.

Deploying deep neural networks in real-world settings requires models that are both compact and robust to common corruptions. However, at deployment-relevant high sparsity, standard pruning pipelines often degrade corruption robustness, and existing sharpness-aware training/pruning approaches provide limited robustness gains. We address this issue by introducing Sparsity-Adaptive Sharpness-Aware Minimization (SA-SAM), which derives a sparsity-dependent SAM/ASAM perturbation radius by keeping the mean absolute perturbation (an $\ell_1$-based proxy) approximately invariant as sparsity increases. As a simple complementary option, we evaluate Magnitude-Weighted Hessian (MWH), derived from a second-order removal-path analysis, yielding an importance proportional to $\mathrm{Diag}(F)_i\,|w_i|$, where $\mathrm{Diag}(F)$ is the diagonal empirical Fisher used as a curvature proxy in our implementation. Across CIFAR-10-C, CIFAR-100-C, and ImageNet-100-C, our approach achieved stronger corruption robustness than the considered pruning baselines at 80–90% sparsity, while preserving clean accuracy. We additionally quantify the robustness–throughput trade-off by reporting measured inference throughput under sparse execution at deployment-relevant sparsity levels.

Kalman filters (KFs) choose the gain as the product of the state-measurement cross-covariance and the inverse of the innovation covariance. In nonlinear systems, the predictive distribution generally changes shape, and these covariances require approximation. Conventional nonlinear KFs approximate them around a single center, the predicted state. They can produce inaccurate gains and overconfident covariance estimates when one local approximation does not represent the measurement geometry across the uncertainty region. To address this issue, we introduce probe sets, small covariance-scaled collections of states at which the filter repeats its local approximation. We then select the gain that minimizes the average covariance reported by these local approximations. The same construction is used in our previously proposed covariance recalibration step to improve covariance consistency. The construction applies to different KF variants without requiring a specific approximation rule. For quadratic measurements, we derive conditions under which probing yields lower normalized true error variance than the corresponding single-center gain when prediction uncertainty is sufficiently understated. We apply the framework to four KF variants and compare the resulting filters with multiple baselines across 300 randomized setups for each of two systems. The results show that probing preserves typical accuracy while substantially reducing covariance inconsistency and the rare large-error trajectories that dominate the root mean square of per-run errors. The code is available at https://github.com/Shida-Jiang/Probe_KF.

In this work, we propose a decentralized communication-conditioned generative framework for multi-agent collision avoidance. Agents generate short-horizon action sequences using a flow-matching policy trained from privileged offline demonstrations with access to global state. The demonstrations do not include explicit communication signals; instead, agents learn to exchange and aggregate latent messages that encode interaction-relevant intent under partial observability. This formulation supports flexible inference at test time, where unconditioned generation corresponds to independent behavior and communication-conditioned generation enables coordinated interaction without centralized planning. The resulting policies operate in a fully decentralized manner at execution time, relying only on local observations and learned messages. Combined with a receding-horizon inference scheme, the proposed approach enables efficient single-step inference of short-horizon action sequences and degrades gracefully under communication dropouts. Extensive simulation results demonstrate near-expert collision avoidance performance and strong generalization to denser, unseen multi-agent scenarios, along with zero-shot transfer to real-robot experiments.