Papers

Consumer-facing skincare AI must coordinate visual evidence, product information, tool use, and user-facing actions within explicit evidence and safety boundaries. This study evaluates SkinAgent AI, a non-diagnostic multimodal framework that combines visual concern routing with grounded and auditable LLM-based orchestration. The architecture includes routing for Acne, Pores, and Wrinkles; photograph-based skin-type estimation; count-informed ordinal acne-severity support; typed tools; database-grounded recommendation and action functions; deterministic safety, privacy, and evidence checks; approval before state-changing actions; and structured trace and replay mechanisms. Visual-model performance and system-level agent behavior were evaluated separately. Across three seeds, the skin-condition routing model achieved 99.84% +/- 0.07% accuracy. Skin-type estimation achieved 88.85% accuracy, while count-informed acne-severity support achieved 84.59% accuracy with a quadratic weighted kappa of 0.9076. On a locked but non-independent 240-case system benchmark, intent accuracy was 80.00%, exact tool-set match was 62.92%, and strict task completion was 47.08%. No violations or successful cross-user leakage events were observed in the finite safety and privacy test suites. Tool-selection errors, incomplete grounding of product attributes, and unreliable failure fallback nevertheless remained. These findings support the feasibility of bounded, database-grounded, and traceable agent orchestration for non-diagnostic skincare assistance. They do not establish clinical readiness, external generalization, formal privacy guarantees, or universal safety. Independent validation, expert assessment, robustness and fairness testing, and prospective evaluation in real-world settings remain necessary.

This paper presents a grasping model for cloth manipulation specifically tailored to ease the deployment of robotic control methods. The model is robust, fast and easy to implement avoiding at the same time contact and friction considerations between the gripper and the cloth in favor of simple positional constraints. The gripper is described by its pose, jaw state, and an attached grasping volume. Two kinds of grasping volumes are considered: an axis-aligned box to simulate a pinch grasping and a square pyramidal volume to simulate point grasping. When the gripper closes, the discrete cloth positions lying inside this volume are selected, stored in the local gripper frame, and then transported with the gripper motion. A simple squeezing step is also included to progressively move the selected cloth positions toward the center of the grasping region, avoiding an instantaneous displacement at closure. The model can be used in any simulator as it only requires access to discrete cloth positions and a mechanism for imposing target positions as constraints. We implement our grasping model in conjunction with a constraint-based inextensible cloth simulator, where grasping is implemented as moving positional equality constraints coupled with stretch, shear, collision, and table contact projection steps. The same gripper trajectory is applied on a robot arm to fold a real piece of cloth, serving as a simple bridge between simulation and physical cloth manipulation and showcasing the realism and practicality of our idealized grasping model.

Image denoising remains a fundamental problem in image restoration, with applications in photography, biomedical, and scientific imaging. Modern deep neural networks achieve strong performance by learning powerful image priors, but often rely on large black-box models with limited interpretability. In contrast, DCT-based sliding-window and collaborative filtering methods such as BM3D offer clear algorithmic structure, but depend on handcrafted and non-differentiable operations. This work studies how far such structured collaborative filtering principles can be pushed when reformulated as trainable models. We introduce DeepBM3D, a compact fully differentiable architecture that combines non-local patch grouping, DCT-domain filtering, and multi-stage refinement within a BM3D-inspired pipeline. Lightweight convolutional feature extractors guide patch grouping, while filtering is performed through learned Wiener weights in the DCT domain. Experiments show that DeepBM3D improves over classical and hybrid baselines, remains competitive with FFDNet at low and moderate noise levels, and performs particularly well on repetitive textures.

One long-form exam in a large course costs hundreds of grader-hours, and qualified graders are scarce; LLM graders are a tempting alternative. To show its pitfalls we grade a practical Computer Vision exam ($570$ dual-graded students) under $171$ configurations spanning closed and open-weights models; the best reaches mean absolute error $1.64/35$, below the $2.61/35$ two human graders achieve against each other. The catch is the prompt: a short ”strict grader” preamble drives $14$ of $17$ open-weights models out of the graded band ($\text{MAE} \ge 8$), three stopping grading altogether. The damage traces to the preamble's two credit-withholding sentences, not to tone or model scale; one of them, ”never give partial credit”, alone makes two of three probed models stop grading. The closed flagships of three vendors shift calibration under it but stay in the band. In $162$ further configurations on a second, independent Machine Learning exam from another course ($1{,}038$ dual-graded students), the preamble worsens ten models, moving three out of the band into collapse and one into refusal, yet improves seven whose neutral prompts over-mark: the vulnerability replicates, but its direction is exam-specific. Light LoRA fine-tuning repairs it: one adapter on the two exams' pooled $\sim 3{,}900$ graded examples brings five small open models to parity or better with a human grader in agreement with the grader pair, and sensitivity to the three harsh personas nearly vanishes ($\le 0.32$ MAE). We release the anonymised dataset, full ablation grid, and grading, fine-tuning and analysis pipelines.

The study of approval-based committee voting has so far focused predominantly on proportional representation. The canonical notion of diverse representation, based on the Chamberlin-Courant score, counts the number of voters with at least one representative in the committee, making it an individualistic notion. We develop a more comprehensive theory that instead requires the representation of many groups of voters, grounding it in the justified representation (JR) axiom, which we strengthen in two directions. First, we study the existing axioms of Strong JR (SJR) and Semi-Strong JR (SSJR), which consider the same cohesive groups as JR but demand stricter representation. We show that neither can be optimized efficiently on general domains (unless P = NP), but both can be on the Candidate Interval and Voter Interval domain. Second, to capture the unique and defining opinions of a group, we introduce Distinctive Representation (DR) and its local optimization variant, Local DR. We address satisfiability and computation time, and show (Local) DR to be distinct from known proportionality and diversity axioms. Experiments on real-world and synthetic data show that Local DR performs well on multiple empirical measures of diversity.

Identifying the set of monitored websites in mixed encrypted traffic is challenging because an individual flow often provides only partial evidence of website identity. To address this challenge, we propose FlowAtom, which constructs shared prototypes, called Atoms, from flow representations without website labels. Specifically, FlowAtom pretrains a flow encoder on external unlabeled traffic and aggregates Atom responses across flows within each observation window into a fixed-dimensional, permutation-invariant representation for monitored website-set prediction. Across Direct HTTPS, Trojan, and VMess, FlowAtom achieves micro-F1 scores of 97.82%, 94.43%, and 93.92% in closed-world evaluation, respectively, and consistently outperforms the evaluated baselines in open-world evaluation on windows containing monitored visits. The code is available at https://github.com/aimafan123/FlowAtom.

Existing continual-learning methods protect parameters, replayed examples, or Euclidean feature subspaces. When applied to hyperbolic multimodal models, they do not explicitly preserve the Lorentz geometry that jointly encodes within-modality similarity, cross-modal correspondence, and semantic hierarchy; sequential updates can therefore retain task scores while still distorting previously learned relations. We address this gap with Hyperbolic Multimodal Continual Learning (HMCL). We show that preserving the old multimodal geometry amounts to restricting all modalities to one shared hyperbolic isometry, which induces a family of admissible first-order parameter changes. We formulate a joint closest-admissible (CA) correction that retains the shared rotation best matching the candidate modal updates; its minimal-rotation (MR) special case fixes this rotation to zero. Both variants correct the displacement realized by AdamW, and task anchoring bounds within-task accumulation while preserving learning freedom. Across a unified 16-task classification-retrieval stream with three hyperbolic backbones, HMCL improves final performance and backward transfer over sequential fine-tuning and four continual-learning baselines; HMCL-CA gives the highest Overall score on every backbone. A modality-extended stream confirms the retrieval gains. Representation analyses find 81.2 to 95.5 percent less radial, angular, cross-modal, and paired-distance drift; ImageNet-WordNet results show better semantic ancestry and radial hierarchy.

Understanding how Large Language Models (LLMs) encode linguistic structures remains a fundamental challenge in interpretability research. While diagnostic classifiers (or "probes") are widely used for this task, they face significant methodological criticism: training auxiliary classifiers introduces capacity confounds and calibration issues, often making it difficult to distinguish the model's intrinsic representations from the probe's ability to learn the task. To address these limitations, we introduce a probe-free framework for localizing linguistic selectivity at the individual neuron level. Leveraging the controlled contrasts of linguistic minimal pairs, we propose a Neuron Separability Index (NSI), a metric that directly quantifies how reliably single neurons differentiate grammatical from ungrammatical constructions without parameter updates. Applying NSI across 68 linguistic paradigms and seven checkpoints reveals three main patterns: 1) raw separability reaches near-peak levels earlier for morphological and syntactic distinctions than for syntax-semantics interface and conceptual distinctions. 2) after permutation normalization, single-unit selectivity is sparse, weak, and narrowly tuned: only a small fraction of units are sensitive to an average paradigm, and strongly selective "grandmother neurons" are rare. 3) whole-vector linear separability, single-neuron selectivity, and behavioral competence are largely dissociated, and targeted ablations further separate activation selectivity from causal reliance.

Selective classification enables a model to abstain from predictions on uncertain instances, but existing approaches typically reject them through confidence scores, predefined coverage constraints or instance-level distance measures. These approaches may overlook the collective geometric structure of difficult samples in learned representation spaces. We propose Guided Clustering-based Uncertain Learning (GCUL), a geometric-guided selective classification framework that identifies misclassified and ambiguous instances as a potential confusion attractor in the representation space. GCUL uses a three-phase procedure to initialize, cluster, and explicitly relabel this uncertain region, allowing the rejection boundary to emerge from the underlying representation geometry rather than from a prescribed rejection rate. We further derive a selectivity score and a geometric sufficient condition that characterizes when rejection can provide positive operational utility, enabling pre-deployment feasibility assessment. GCUL improves DistilBERT accuracy from 89.37 percent to 94.98 percent with less than 9 percent rejection. Beyond accuracy, our selectivity score correctly pre-detects the only dataset (GoEmotion) where all baselines fail, and controlled simulations yield 6.1 percent Type-I and 0 percent Type-II errors, validating the sufficient condition's conservatism. These results suggest that collective representation geometry provides a useful alternative perspective for selective prediction.

We study the \textsc{$k$-Disjoint Paths} problem on a graph embedded on a surface with bounded Euler genus. Given a graph $G$ with $n$ vertices and $k$ vertex pairs embedded on a surface of Euler genus $g$, we present a $2^{O(k^2+g^2)}n$-time algorithm that computes $k$ pairwise vertex-disjoint paths connecting the given vertex pairs if such paths exist. Our approach relies on the decomposition of $G$ into $O(k+g)$ planar subgraphs while bounding the complexity of the boundaries between these subgraphs. This approach enables the use of techniques for compressing linkages in planar graphs. Moreover, our techniques yield two kernels of size polynomial in $k$, $g$, and the treewidth of the graph, and of size $2^{O(k+g)}$. These results extend recent advances on \textsc{$k$-Disjoint Paths} on planar graphs [Cho et al. SODA 2023] and [Włodarczyk and Zehavi FOCS 2023] to surface-embedded graphs.

Estimating electrocardiography (ECG) from a chest-worn inertial measurement unit (IMU) enables continuous heart rate (HR) monitoring without the discomfort of electrodes. We propose TinyCardioUNet, a lightweight UNet that uses all six IMU axes without prior channel selection, refines its bottleneck with a graph neural network that encodes inter-axis dependencies, and employs tensor decomposition with automatic variational Bayesian rank selection for parameter reduction. On a public dataset, TinyCardioUNet achieves an RMSE of $0.098$ and a Pearson correlation coefficient of $0.677$ with only $36.0$k parameters and remains comparatively robust to additive noise, demonstrating accurate ECG reconstruction with a compact model.

This work addresses finite-horizon inverse linear quadratic Gaussian games. In a constrained setting, we characterize the set of cost parameters and optimal dual values that generate a given generalized Nash equilibrium, and we propose an algorithm to compute these parameters. In an unconstrained setting, we address transferability, namely, we bound the cost value perturbation between two policies: one induced by the identified cost parameters, the other by the expert parameters, under a set of different dynamics. This cost value perturbation scales linearly with the deviations in the dynamics and the identified cost parameters. Through numerical simulations, we show that, in a constrained setting, our algorithm identifies the cost parameters and dual values that can reproduce the policy and trajectories corresponding to the observed generalized Nash equilibrium. In an unconstrained setting, we show with a traffic simulation and real-robot experiments that the identified cost parameters can be used to control sufficiently close dynamics, with performance degrading linearly with the deviations in the dynamics and the identified cost parameters.

Recent work on synthesizer inversion shows that generative models outperform deterministic approaches by explicitly modeling the ambiguity in mapping audio to parameters. Training such models, however, requires audio-parameter pairs, which are typically obtained by rendering sampled or preset parameters through the synthesizer itself. This creates a train-test mismatch that can degrade performance on off-manifold real-world recordings, for which ground-truth parameter annotations do not exist. To circumvent this obstacle, we propose to model the joint distribution of audio and parameters with a multi-modal continuous normalizing flow using independent noise schedules for each modality. This formulation allows us to train joint and conditional densities with paired synthesizer data, while unpaired real recordings can train the audio marginal alone, exposing the model to off-manifold signals without requiring parameter labels. Further, because the model learns to map from audio to parameters at all noise levels, we find that partially noising the audio reference at inference improves real-audio reconstruction, consistent with reducing sensitivity to distribution-specific detail while preserving coarse structure. Evaluating on Surge XT and Dexed, we find that modelling the joint distribution substantially improves both inversion of real-world and in-domain audio.

This paper presents a learning-based adaptive augmentation control concept inspired by the adaptation mechanisms of conventional adaptive control, while not being restricted to their specific parametric adaptation structures. In contrast to augmenting a reinforcement learning (RL) baseline controller with classical adaptive control to account for the simulation-to-reality gap, the proposed approach uses RL-based adaptive augmentation to address the limitations of conventional adaptive control. Domain randomization combined with observation stacking is employed to train the RL-based augmentation to compensate for matched uncertainties in a fixed-wing aircraft system. To ensure constraint satisfaction during operation, a safety filter is incorporated into the control architecture. Based on the concept of pseudo control hedging (PCH), we propose a modified reference model that avoids undesirable interactions between the RL-based augmentation and the safety filter. To reduce the conservatism of the safety filter, we additionally incorporate a disturbance observer. The proposed approach is evaluated on a fixed-wing aircraft model subject to uncertainties.

Time series analysis is fundamental in domains such as finance, healthcare, and meteorology. Real-world time series often exhibit multiscale characteristics shaped by diverse latent factors, resulting in intricate temporal patterns and rich frequency structures. However, existing approaches typically focus on either frequency-domain decomposition or time-domain pattern extraction in isolation, neglecting their joint structure. This decoupled modeling limits representation expressiveness and undermines performance in tasks requiring simultaneous temporal and spectral reasoning. To address this gap, we propose m-WCN, a novel end-to-end deep learning framework that neuralizes multi-wavelet decomposition for joint extraction of temporal patterns and frequency components. By approximating the classical GHM multi-wavelet transform with trainable convolutional operators and enforcing orthogonality constraints, m-WCN produces interpretable multi-resolution representations. Built on this foundation, we introduce two task-specific architectures: TFBC for time series classification, which boosts discriminative features across frequency scales, and FTB for forecasting, which ensembles frequency-aware predictors. Extensive experiments on 64 UCR datasets and seven public forecasting benchmarks demonstrate the effectiveness of our approach. Built on the neuralized m-WCN, our TFBC and FTB outperform various baseline models across diverse datasets, achieving average improvements of 19.97% in classification and 19.92% in forecasting tasks.

Low-voltage (LV) distribution feeders are increasingly difficult to monitor because real-time load data are unavailable, historical measurements are sparsely sampled, and high-rate voltage sensors cover only a few nodes. This paper proposes a multi-stage linear programming (MSLP) estimator for three-phase unbalanced LV grids that reconstructs per-phase nodal voltages under such limited observability. The estimator linearises the three-phase power flow through voltage-to-power sensitivity matrices and adjusts the nodal active and reactive power injections so that the resulting voltages match the available measurements. To limit the error introduced by linearisation, the sensitivities are rebuilt and the power corrections refined over successive iterations, and power-balance constraints at intermediate metered nodes are added to tighten the feasible region. The method is validated on a real 50-node Danish LV feeder, where it attains a mean absolute error of 0.641V, a standard deviation of 0.816V, a root-mean-square error of 0.842~V, and a maximum error of 3.437V. Compared with a single-stage linear estimator, MSLP lowers the error metrics by roughly 16 percent on average, and additional studies quantify the influence of the input-preparation strategy and of the number of estimation meters.

Parameter mismatch is the main source of rotor position estimation error in sensorless control of surface-permanent magnet synchronous motors (SPMSMs). To this end, this paper proposes a simple yet efficient equivalent flux compensation (EFC) method that directly estimates the equivalent flux disturbance caused by parameter mismatches in real time. First, the equivalent flux disturbance caused by parameter mismatches is derived from a nonlinear flux observer. Second, a flux update law is proposed to minimize both magnitude and directional errors by leveraging geometric error together with the derived equivalent flux disturbance. To enhance numerical stability, a saturation function is introduced to improve gradient continuity in the update process. Additionally, Lyapunov analysis is employed to ensure the stability of the proposed update law, from which the corresponding error bounds and convergence properties are derived. Finally, experimental results validate that the proposed method fully compensates for the steady-state effects of resistance and flux mismatches, and partially mitigates the influence of inductance variation, effectively constraining the position estimation error within a relatively small range.

Communities often respond to potentially AI-assisted work by asking three questions: Was AI used? Was that use disclosed? Can hidden use be detected? These questions place AI use itself at the center of accountability while overlooking a deeper problem: unowned judgment. Evaluations, claims, decisions, and creative directions can be shaped by AI with no accountable human or institution prepared to stand behind them. We develop this argument through two illustrative cases: AI-assisted peer review and concealed AI use in creative work. The first shows how contribution dissolution can weaken responsibility while the second shows how the fear of losing credit can discourage honest disclosure. The cases expose the limits of disclosure rules and provenance records as responses to AI-mediated collaboration. We offer three directions for discussion: distinguishing the roles AI plays, identifying judgments that require clear human ownership, and creating conditions in which AI involvement can be disclosed without default penalty. The broader aim is to make AI-shaped contributions discussable, creditable, contestable, and repairable.

This paper addresses the software portability gap between Model-Based Development (MBD) and advanced many-core execution for Cyber-Physical Systems (CPS). We present a workflow-preserving retargeting approach for Simulink-based CPS applications with candidate-wise data parallelism to OpenCL-based many-core processors. Rather than manually rewriting models for new platforms, our toolchain uses MathWorks GPU Coder to extract data-parallel CUDA code, which is then translated into OpenCL host and device code via a custom framework. The conversion handles syntax rewriting, API emulation, and platform-specific argument packing. We deployed this workflow for a computationally intensive Frenet-frame trajectory planner on the Kalray MPPA Coolidge2. The results demonstrate the feasibility of a workflow-preserving retargeting pipeline for the evaluated CPS workload and platform.

Robot manipulation policies trained through imitation learning inherit not only the demonstrated behavior but also the conservative execution tempo of robot demonstrations. Existing acceleration approaches can execute faster than the original demonstrations, but determine the appropriate acceleration primarily from robot-side information or a predefined set of tempo factors, leaving open how to obtain a task-appropriate reference for how fast each manipulation phase should progress. We introduce EgoSpeedUp, a framework that uses human manipulation as temporal supervision for robot imitation learning. Our key insight is that human demonstrations naturally reveal task-appropriate, phase-wise manipulation tempo. Given slow robot demonstrations and human demonstrations of the same task, EgoSpeedUp aligns corresponding manipulation phases, estimates their relative execution tempos from multiple human demonstrations, and transfers the resulting phase-wise tempo by retiming the robot demonstrations. The retimed demonstrations are then used for standard behavior cloning, allowing the robot to retain its executable manipulation behavior while learning to perform it at a human-informed tempo. Across two real-world manipulation tasks, EgoSpeedUp improves the task success rate by an average of 25 percentage points (pp) while reducing successful execution time by 36.5%. These results demonstrate that human manipulation tempo provides an effective temporal reference for learning faster and more reliable robot policies.

In open-ended problem solving, collaborators often rely on discussion to surface concerns, challenge perspectives, and refine shared work as it evolves. While AI agents are increasingly used as discussion partners, existing multi-agent systems place a heavy burden on users to initiate and carefully orchestrate the discussions. We present DocuTeam, a mixed-initiative multi-agent discussion system in which both users and agents can initiate and steer conversations. Agents monitor document changes to proactively start and redirect discussions as the work evolves, while users can flexibly shape the conversation or adopt agent ideas. In a within-subjects study (N=20), participants using DocuTeam produced outcomes rated significantly more novel, relevant, and specific than with a baseline without any increase in cognitive load. Rather than using agents for one-off idea sourcing, participants engaged in an iterative refinement loop in which document changes prompted agent reactions, which led users to revisit and further develop their work.

Sticky policies remain attached to protected data so that access restrictions can persist across systems, yet request-time authorization often still depends on a centralized Trusted Authority or Policy Decision Point (PDP). This paper presents SPEAR-Q, a decentralized framework in which independent Policy Authority Nodes (PANs) evaluate the active sticky policy using local context, consent, and risk information and generate signed authorization evidence. A strict-majority Evidence quorum determines the global Permit or Deny decision, while a committed policy becomes active only after the required PAN majority applies it. SPEAR-Q was deployed across physically separated hosts in the Airbus Cyber Security Simulation Platform and evaluated using a healthcare workload derived from MIMIC-IV. Across 9,000 performance requests, all requests completed without execution failure or timeout. Latency increased and throughput saturated as cluster size and concurrency grew, with PAN evidence waiting and PAN-local processing dominating under load and credential validation forming the main PAN-side cost. Policy-activation experiments confirmed majority-based activation. Quorum experiments showed that correctly signed false-Permit evidence from fewer compromised PANs than the required quorum could not produce a global Permit or resource release, while authorization remained fail-closed when reachable PANs fell below the required quorum. Compared with prior approaches that decentralize policy management, credential authority, or threshold-based release separately, SPEAR-Q integrates persistent sticky policy enforcement, independent request-time evidence, majority-based policy activation, and explicit quorum-bounded authorization within one decentralized framework. Under the evaluated conditions, SPEAR-Q supports decentralized sticky-policy authorization without relying on a centralized decision authority.

Brain-age prediction from resting-state fMRI provides a quantitative framework for characterizing age-related changes in spontaneous brain dynamics and for identifying functional signatures. Existing studies have linked fractal and multifractal scaling to age and examined the reliability of individual features. However, prediction repeatability depends on how features fluctuate jointly and how a predictor combines them, which feature-wise reliability assessments do not capture. To address this problem, we propose Repeat-informed Multifractal Curve Regression (RMCR), a structured framework for learning stable age-predictive patterns from multifractal curves. By jointly modeling curve structure and repeat-scan variability, RMCR learns predictive combinations of fluctuation orders that target both accuracy and within-subject consistency. Relative to a matched run-level ridge baseline, RMCR reduces single-run MAE by 6.1% on HCP-A and 7.9% on an external Cam-CAN cohort, and within-visit repeat absolute difference by 18.5% on HCP-A, using a single scan at inference.

Valiant and Vazirani isolate a satisfying assignment of a circuit with probability $\Omega(1/n)$. Dell, Kabanets, van Melkebeek, and Watanabe showed that success above $2/3$ implies $\mathrm{NP}\subseteq\mathrm{P/poly}$ and asked about the range in between. We show that every positive constant already implies the collapse: if a randomized nonuniform polynomial-size pruning procedure succeeds with probability $\epsilon$ on affine circuit inputs with at most $2^{\lfloor 2/\epsilon\rfloor}$ satisfying assignments, then $\mathrm{NP}\subseteq\mathrm{P/poly}$. Success $10/\log L$ on affine inputs with at most $L^{1/3}$ satisfying assignments suffices, where $L$ is the description length, and on inputs with one or two satisfying assignments the threshold $2/3$ drops to $3/5$. No cryptographic assumption is used, and the procedure may read the entire circuit. The proof compiles a pool of circuits into one circuit whose satisfying assignments are indexed by tags in $\mathbb{F}_2^d$. Each member is assigned an affine region of tag space, and if one member is unsatisfiable, the satisfying set shrinks to that member's region. Because regions may overlap and have different dimensions, the collapse reduces to a combinatorial bound: no set of tags meets more than a $2/d$ fraction of an equally weighted family of affine subspaces of all dimensions below $d$ in exactly one point. This regional counting cannot go below order $1/\log L$. The range between $\Theta(1/n)$, achieved by affine hashing, and $O(1/\log n)$ remains open.

Existing formal mathematics benchmarks, such as miniF2F, ProofNet, and PutnamBench, primarily evaluate models on constructing complete formal proofs for challenging problems. Because success is measured at the theorem level, these benchmarks offer limited insight into models' step-level formal reasoning. Evaluating this capability separately enables finer-grained diagnosis of model limitations than theorem-level evaluation alone. To fill this evaluation gap, we introduce ProofGap, a fine-grained benchmark for step-level formal reasoning. ProofGap is constructed through a natural-language proof-processing pipeline that decomposes each reasoning step into one or more aligned proof gaps. Applying this pipeline to natural-language solutions to 3,015 exercises in B. P. Demidovich's Problems in Mathematical Analysis yields 26,116 gaps. The benchmark focuses on mathematical analysis, a domain that remains challenging for current models. By supplying the local context and target explicitly, gap completion isolates local formal proof construction from end-to-end proof composition, enabling more precise localization of model failures. Natural-language solutions serve as the provenance of these obligations, while the benchmark task itself starts from an already formalized local context and goal. Beyond benchmarking, the same pipeline may support future proof-verification systems, provided that semantic translation and sequential proof composition are handled reliably.

The $k$ vertex-disjoint paths problem asks whether, given a graph $G$ and $k$ pairs of vertices $(s_1,t_1)$, \ldots, $(s_k,t_k)$, $G$ has $k$ pairwise vertex-disjoint paths connecting $s_i$ and $t_i$ for all $1\leq i\leq k$. If $G$ is undirected, then this problem is NP-complete, but there exist FPT algorithms parameterized by $k$.Since these algorithms involve an extremely large function on $k$, algorithms for restricted graphs have also been investigated. In particular, a $2^{2tw\log tw+O(tw)}\cdot n$ time algorithm for undirected graphs with $n$ vertices and treewidth $tw$ is proposed by Scheffler (Technical Report 396, TU Berlin, '94), and it is proved by Lokshtanov, Marx, and Saurabh (SIAM J. Comput. '18) that, under the ETH, there exists no $2^{o(pw\log pw)}\cdot n^{O(1)}$ time algorithm for either directed or undirected graphs with pathwidth $pw$ and for $k=\Omega(pw^4)$. It has not been known whether the lower bound also holds for a smaller $k$. In this paper, we prove that, for both the directed and undirected cases, there is an algorithm faster than Lokshtanov et al.'s lower bound for $k=tw^{o(1)}$ by proposing a $2^{O((tw+k)\log k)}\cdot n$ time algorithm. Besides, we prove a lower bound that, under the SETH, there exists no $(2-\epsilon)^{pw\log pw}\cdot n^{O(1)}$ time algorithm for directed graphs and for a general $k$. This lower bound is tight because, with slight modifications, Scheffler's algorithm runs in $2^{pw\log pw+O(pw)}\cdot n$ time also for directed graphs.

In this work, we focus on photorealistic sign avatar modeling, which is crucial for effective communication with the Deaf community and is characterized by complex hand gestures and nuanced facial expressions. To this end, we introduce MVSign, the first multi-view Chinese sign language dataset co-designed with Deaf experts, featuring diverse gestures and rich annotations. For precise SMPL-X annotation, we develop a hybrid fitting pipeline that produces accurate body, hand, and facial parameters and can also be applied to the monocular setting. Building on MVSign, we propose a decoupled sign avatar representation that isolates body, head, and hand components to capture complex articulations, together with a motion-aware sampling strategy to handle motion blur and balance gesture diversity. Extensive experiments demonstrate that our method achieves high-fidelity visual results on MVSign, particularly in detailed hand and facial regions, and generalizes well to in-the-wild monocular sign language videos. Project page: https://naaapi.github.io/PHOSA.

We propose Sufficiently Reduced Distributional Regression (SRDR), a generative method that combines conditional distribution estimation with nonlinear sufficient dimension reduction (SDR). It builds on a characterization of sufficiency through strictly proper scoring rules: a dimension reduction is sufficient if and only if predicting the response from the reduced covariates incurs no loss in expected score relative to the full covariates. Sufficient dimension reduction thus becomes a risk minimization problem. SRDR jointly trains a dimension reduction map and a generative prediction model by minimizing the energy score, which can be estimated by sampling without density evaluation or adversarial training. The framework extends to multi-environment data and to classification. We prove that the estimated conditional distributions converge in energy distance to the true ones, which implies that the learned representation is asymptotically sufficient. In simulations and applications to CT slice localization, superconductivity, and digit classification, SRDR recovers low-dimensional sufficient structure and matches or outperforms state-of-the-art nonlinear SDR methods in representation quality and predictive performance.

To address the widespread sensitivity of surface-mounted permanent magnet synchronous motor (SPMSM) sensorless control to motor parameters, this paper proposes a temporal regression-based model-free sensorless control (TFC) method. First, voltage integrals and current increments over consecutive short intervals are stacked to construct a finite window regression, in which the unknown stator inductance appears as a common scalar coefficient. Second, a projection operator constructed from the stacked current increments eliminates the inductance term, and a least-squares formulation is developed to reconstruct the rotor flux vector. Meanwhile, the analysis of the projected regression and current-flux geometry establishes a two-dimensional direction vector whose components share a common amplitude containing the stator resistance and flux linkage. This amplitude cancels during position extraction. By setting the resistance reference to zero, the proposed method estimates the position without specifying the stator resistance, inductance, or flux linkage. Finally, experimental results verify the effectiveness of the proposed TFC method.

Large audio-language models (LALMs) expand language models to process and interpret audio, but also expose them to heterogeneous audio jailbreaks. We ask whether successful jailbreaks reflect failures to recognize harmful intent or failures occurring after such recognition. Layer-wise probing reveals the latter: risk-related information remains decodable from intermediate representations, yet the internal risk signal fails to translate into refusal in later-layer processing. We identify this discrepancy as the risk-to-refusal gap. Building on this finding, we propose AEGIS, a detect-then-intervene defense whose mid-layer risk gate selectively activates downstream safety adapters. Across six LALMs and three heterogeneous audio jailbreak benchmarks, AEGIS reduces the average unsafe rate from 17.9% to 0.4%, while causing only a marginal increase in over-refusal on benign inputs. These results establish selective internal intervention as an effective path toward more robust refusal in LALMs. The code is available at https://github.com/azzzzliao/aegis-audio-defense.