Computer Science (arXiv)

Accurate and stable torque rendering is essential for safe and perceptive human--machine interaction. Magnetorheological fluid (MRF)-based actuators offer a compact and rapidly controllable solution for haptic feedback, but their practical implementation requires reliable fluid sealing, low-hysteresis excitation, accurate torque control, and stable long-duration operation. This article presents an integrated MRF haptic system featuring a compact magnetically self-sealed rotary actuator, low-hysteresis PWM operation, high-fidelity model-based torque rendering, and stable performance during long-time operation. Magnetostatic simulation guides the arrangement of magnetic and nonmagnetic materials to focus flux in the multidisk torque and permanent-magnet sealing regions, enabling a maximum 600 N$\cdot$mm/A output. Experiments show that higher PWM frequencies reduce hysteresis and improve repeatability. At 10 kHz, the response is represented by a nonlinear model that varies with the direction and speed of torque change. The real-time controller combines feedforward, hysteresis compensation, PI feedback, and sliding-mode correction. Compared with PID, it reduces square-wave overshoot, undershoot, and steady-state RMSE by 77.4\%, 61.9\%, and 68.3\%, respectively. It tracks sinusoidal and biomechanics-model-based references, and a 1.5-h test shows only a 2.5 $^\circ$C rise near the coil with no clear tracking loss. This high-fidelity torque rendering will fundamentally transform human--robot collaboration by making interactions safer, more efficient, and more intuitive.

We study randomized strategyproof mechanisms for locating multiple facilities on the real line. We introduce the \emph{Product-Gap mechanism}, which selects $k$ reported locations with probability proportional to the product of the consecutive gaps between them and opens facilities at the selected locations. We prove that, for every $k\geq 2$, the mechanism achieves a tight approximation ratio of $2k$ for social cost. We then study its incentive properties and show that it is strategyproof in expectation for $k=2$ and $k=3$, but is not strategyproof for $k\geq 4$. In particular, the mechanism gives strategyproof $4$- and $6$-approximations for two and three facilities, respectively. Finally, for two facilities, we combine Product-Gap with the Proportional mechanism of Lu et al. We show that an optimized report-independent mixture is strategyproof and has a tight approximation ratio of $(74+4\sqrt{3})/23\approx 3.519$ on the line, improving upon the previous factor of $4$.

Physics-informed neural networks (PINNs) embed governing partial differential equations directly into the training loss, offering a promising alternative to costly CFD solvers for unsteady flows. Yet the growing list of techniques proposed to improve PINN training is typically validated one at a time, leaving open whether these techniques actually compose. We study this question in depth on the DFG/Schafer-Turek unsteady cylinder wake benchmark. In isolation, nearly every technique performs no better than an untreated baseline. However, combining periodic (SIREN) activations with causal weighting unlocks a previously inaccessible regime, reconstructing velocity and pressure fields to within 4.1% average relative L2 error against an OpenFOAM reference solution. Adding further techniques instead causes catastrophic performance degradation, demonstrating that individually effective PINN interventions can interact nonlinearly and that more elaborate training recipes are not necessarily better.

Discrete routing protocols in ultra-dense wireless networks are constrained by signaling overhead and transient routing loops that degrade radio-resource efficiency. While continuum modeling provides a scalable alternative, existing scalar density approaches lack the vector geometric structure required to characterize these topological anomalies. This paper introduces a fluid-kinetic framework, vorticity dissipation-based routing (VDR), utilizing the Helmholtz-Hodge decomposition. We demonstrate that the macroscopic traffic flux can be orthogonally decoupled into a demand-driven irrotational component and a loop-induced solenoidal component representing routing vorticity. Building on this insight, we define network vorticity as a macroscopic metric to quantify topological inefficiency. Routing optimization is formulated as a gradient flow on an enstrophy functional, yielding a vorticity dissipation equation as the governing dynamic law. Lyapunov stability analysis proves that this mechanism ensures the monotonic decay of global enstrophy toward an asymptotically loop-free equilibrium. Numerical results validate that VDR suppresses realized forwarding loops, reduces end-to-end delay, maintains robust packet delivery, and exhibits near-linear scaling under fixed-area densification while explicitly accounting for the grid-dependent Poisson-solver cost.

Fast state-machine replication (SMR) protocols in the crash-fault setting have attracted significant interest in both academia and industry. This interest stems from their advantages over leader-based protocols, including low execution latency for non-conflicting commands, high throughput, improved availability, and increased fairness. While Fast SMR is well studied in the crash-fault setting, the Byzantine fault-tolerant (BFT) setting remains much less understood, with only a small number of existing Fast BFT SMR protocols. In this paper, we present tight upper and lower bounds on the replication factor required for Fast BFT SMR. We also present a suboptimal protocol that illustrates a trade-off between replication factor and recovery efficiency.

Modern GPUs increasingly integrate Tensor Cores into the execution pipeline. Although aggregate tensor throughput continues to grow, aided by an operand supply that has evolved from register-based in Ampere to redundancy-free, memory-based in Hopper and Blackwell, efficiently orchestrating the complete tensor compute pipeline for the modern AI workloads remains challenging. We identify the fundamental bottlenecks as fixed parallelism and coarse-grained scheduling, both of which are exposed by modern AI workloads that interleave diverse non-GEMM operations with GEMM. To orchestrate tensor computation efficiently, we propose FIBER, a new architecture that extends the GPU SIMT (single instruction, multiple thread) model. Its basic execution instance, the \emph{fiber}, is decoupled from private register ownership, carrying only minimal control state while accessing an SM's registers through a shared view. This enables dynamic parallelism scaling, fine-grained register-level dataflow scheduling, and offers a redundancy-free alternative for matrix operand supply. We extend the ISA, microarchitecture, and compiler to realize shared-register addressing, conflict-free operand delivery, and fiber-based program mapping. Under a typical mixed-precision LLM serving scenario, FIBER achieves a 2.25x end-to-end speedup on Ampere (1.15x for the original FP16 computation), with 1.8x and 2.09x on Hopper and Blackwell respectively, and kernel-level gains up to 2.49x.

Execution feedback is often treated as a self-verifying signal for improving LLM-generated tests. However, when generated inputs are executed on a single accepted program and its outputs are used as ground truth, invalid or underspecified inputs can create spurious fault detections and apparent evolutionary gains. We audit this failure mode in feedback-driven test generation using 142 development tasks, 114 locked external tasks, and 138 held-out tasks, with two code models, three seeds, and fault-cross-fitted real submissions. On external inputs for which three accepted implementations agree, generated outputs match the panel on only 27.79% and 50.12% of cases. A single-reference oracle inflates the measured gain from evolution by 9.46-14.85 percentage points; after auditing, equal-budget independent resampling outperforms mutation-based evolution by 6.01-18.83 points. We further compare a genuine three-round feedback loop with a density-matched placebo. External Real-Placebo differences are +0.13 and -0.50 points, while held-out differences are +1.99 and +0.28 points and do not provide robust evidence of fine-grained feedback benefit. A blinded semantic audit by two software engineering doctoral students classifies 94.41% of panel-disconfirmed inputs as invalid but 3.60% as valid, showing that panel disagreement is informative but not semantic proof. We propose an audit-and-placebo protocol that separates verifier artifacts, interaction scaffolding, and grounded feedback credit in evaluations of self-evolving test generators.

Scientific data are increasingly used by AI agents, yet existing dataset representations provide limited support for autonomous discovery, interpretation, and invocation. This limitation stems from the fragmentation of scientific data across heterogeneous repositories and from dataset representations designed primarily for human use. To address this limitation, we introduce the Scientific Data Skill (SciDSK), an agent-ready representation that packages dataset-specific knowledge and operational guidance as a reusable agent skill. A SciDSK integrates dataset descriptions, scientific context, file organization, usage procedures, quality checks, and provenance information while retaining the underlying data in its original repository. We define a structured SciDSK specification and develop a systematic construction pipeline that grounds each SciDSK in authoritative dataset records and associated supporting materials. We further establish the Scientific Data Skill Bank, a unified platform that publishes SciDSK resources across six scientific disciplines and supports package access, persistent identification, and traceability to source datasets. We evaluate SciDSK through a retrieval benchmark for dataset discovery and controlled cases for dataset interpretation. The results show that SciDSK improves agent-driven dataset discovery and provides more precise and actionable support for dataset interpretation. These findings support the value of organizing dataset-specific knowledge in an agent-ready representation.

We prove a palette sparsification theorem for general $r$-uniform hypergraphs. For all sufficiently large $n$, every $r\ge 3$, and every $α\ge 7.1$, we show that an $n$-vertex $r$-uniform hypergraph of maximum degree $Δ$ is w.h.p. colorable from independently sampled lists of size $O(\sqrt{\log n})$ drawn from an ambient palette of size $\lceil αΔ^{1/(r-1)}\rceil$. The $\sqrt{\log n}$ dependence is asymptotically tight.

Large language models (LLMs) offer a scalable approach to social simulation, but their credibility depends on how agents are constructed. Existing methods can partially reproduce population-level patterns, yet often fail to capture human-like diversity. Our analysis shows that static-profile agents exhibit stronger demographic separation and within-group compression than humans, a pattern consistent with identity essentialism: demographic labels can encourage models to treat group-average tendencies as individual traits, homogenizing responses within groups. We argue that this limitation arises from two related factors: sparse, static agent representations and the limited ability of prompt-only memory to persistently integrate experience. Inspired by complementary memory systems, we propose LifeMem, a longitudinal memory framework that combines structured life-event retrieval with agent-specific parametric memory for experience integration. Experiments on Add Health and Understanding Society with three LLMs show that LifeMem improves alignment with human data in terms of response distributions, overall and within-group diversity, and patterns of within-person response change across life stages. These findings highlight the value of longitudinal life-event memory for constructing more faithful and dynamically evolving social agents.

Whether AI assistance helps or harms a population depends less on the model's accuracy than on whether people rely on it appropriately trusting it when it is right and checking it when it is not. Yet reliance is usually studied one user at a time. We model it as a population process: agents repeatedly solve a task alone, accept an AI answer, or verify it, updating a Bayesian belief about AI quality and, when networked, learning from peers. Four results form one story. The environment sets the baseline: task difficulty and AI quality fix both overreliance and calibration regret. Social learning creates consensus, not overreliance: a mean-preservation theorem, confirmed by a 2*2 topology*tagging design, shows connectivity moves the aggregate only when influence transmits beliefs. Social proof turns reliance into a feedback cascade: visible unverified use suppresses verification and tips the population into collective overreliance. Feedback design can prevent collapse: making verification visible or dampening social proof reverses it. Together, the results frame AI reliance as a computational social dynamics problem, where individual learning, peer observation, and feedback exposure jointly shape whether a population remains calibrated.

Much of the text produced throughout the lifetime of a document is impermanent. In this paper, we explore how writing activity traces can be made visible and interactive to help writers navigate their document histories and understand their writing processes. Using the Flower and Hayes cognitive process model of writing, IRIS infers writing process states from keystroke logs and presents them using an AI-enhanced version history. IRIS provides three primary interactions: revision highlighting that shows local process histories in-situ, conceptual filters that constrain the version history by process type or topic, and natural language inquiry that lets writers pose reflective questions about their writing and process. Following a formative and a longitudinal study, we find that writers use the interfaces to locate specific revisions and understand the progression of their writing. They use system outputs as interpretive material, relating them to pre-existing beliefs and confirming, challenging, and deepening their understanding of their writing.

Latent Action Models (LAMs) have emerged as a promising paradigm for enabling robot learning to leverage large-scale unlabeled videos through latent actions that serve as compact surrogates for physical actions. Despite rapid progress, research on LAM remains highly fragmented, with existing methods evaluating different design choices in isolation under inconsistent experimental settings, making it difficult to identify the factors that truly determine downstream robotic manipulation performance. In this work, we present the first comprehensive empirical study of latent action learning for robotic manipulation. We unify representative LAM methods within a common autoencoding framework and systematically investigate 41 LAM design choices across three dimensions, including latent action modeling paradigms, learning objectives and regularization methods, and latent action integration strategies. We further examine four proxy metrics for evaluating latent action quality and assess their ability to reliably predict downstream robotic manipulation performance. Extensive experiments on three widely used benchmarks provide strong empirical evidence that fine-tuning vision-language model (VLM) backbones with latent actions provides a stronger initialization for downstream policy learning, with further validation on real-world robot manipulation tasks.

LLM reasoning is stochastic, and so understanding a model requires grappling with the distribution of reasoning chains that it might produce for a given question, i.e., its uncertainty. Resampling-based analyses characterize this distribution, revealing which steps of a rollout determine how the model arrives at its answer. However, a major limitation of these approaches is that resampling text sequences at every token or sentence in a reasoning chain is very costly. Our work strives to make resampling analysis more computationally efficient, while also shedding light on an important scientific question: what is the right statistical model for explaining uncertainty dynamics in text generation? We show that when resampling many reasoning chains, uncertainty dynamics converge to stable patterns, and noise is largely an artifact of sampling rather than an LLM's sensitivity to each individual token or reasoning step. We develop a statistical model for smoothing noisy low-sample rollout data to better approximate high-sample data, allowing us to significantly cut sampling costs.

Direct Preference Optimization (DPO) has emerged as an effective approach for aligning large language models (LLMs) with human preferences. However, its adaptation to multimodal settings remains unexplored. Through representational analysis, we identify a key limitation in multimodal preference optimization, which we term visual insensitivity: models often fail to distinguish between images and those with critical visual context removed. Our theoretical analysis further uncovers two manifestations of this problem, namely Across-Image Insensitivity and Within-Image Insensitivity. To address these challenges, we propose Perception-Enhanced Alignment DPO (PEA-DPO), a framework for multimodal LLMs alignment, which explicitly leverages visual preference signals to overcome visual insensitivity. We further provide a theoretical analysis demonstrating that PEA-DPO provably mitigates both failure modes. Empirical results demonstrate that PEA-DPO enhances sensitivity to visual context while preserving the language modeling capacity of the base model. Evaluations across three hallucination benchmarks using MLLMs of varying scales show that PEA-DPO effectively mitigates visual insensitivity, achieves stronger multimodal alignment, and substantially reduces hallucinations.

Embedding-based retrieval (EBR) is pivotal in e-commerce search but often struggles with complex semantics. While recent methods often fine-tune large language models (LLMs) for representation learning, they typically lack robust mechanisms for handling complex and implicit semantics. While Retrieval-GRPO (R-GRPO) recently introduced reinforcement learning to dense retrieval, it suffers from noisy top-K candidates due to limited batch sampling and biased relevance assessments caused by using similarly trained LLMs as reward models. To tackle these issues, we propose Supervised Retrieval-GRPO with Semantic Identifiers (SSR-GRPO). Specifically, our method first proposes a dual-perspective framework for relevance assessment. It leverages both Semantic Identifiers (SIDs) produced by quantization learning and dense representation vectors to generate more unbiased relevance scores. Furthermore, leveraging the hierarchical similarity relationships of the generated SIDs, we mine a set of hard negative samples that serve two purposes: (1) to design a masking function integrated into R-GRPO, effectively filtering intra-group noisy samples; and (2) to construct a Retrieval-DPO task composed of positive and negative sample pairs, enabling the model to capture fine-grained semantic distinctions from a pair-wise perspective. By integrating these optimization strategies, we propose SSR-GRPO. Extensive offline and online experiments validate SSR-GRPO's effectiveness, and it has been deployed on a large-scale e-commerce platform.

A task-effective proposal is not yet a justified physical action. In networked Physical AI, a proposal may be understood while valid, timely, proposal-bound evidence or the authority required to finalize an action remains unavailable. We call this mismatch the verification gap and introduce a Post-Semantic Communication Framework for the systems interface between proposal formation and physical execution. The framework begins with application-declared evidence requirements, represents qualifying observations as evidence records, validates supporting and conflicting records through one path, and separates evidence sufficiency from authorized finalization and a downstream runtime gate. It further distinguishes evidence transfer, which can enlarge the record set reachable by a finalizer, from evidence coordination, which can suppress transmission around records already held at the finalization endpoint. Finite-state framework checks verify that the evaluator implements the declared distinctions consistently. Under the declared model, the controlled communication study exposes a finalizer-dependent asymmetry: sender-finalized Feedback uses evidence transfer to expand evidence reachability throughout the feasible plotted region, whereas receiver-finalized Feedback uses coordination to suppress redundant payload until loss, latency, freshness, and deadline costs shift selection to One-way. Finally, an episode-level reporting schema defines common denominators for future measured Physical-AI studies.

Pretrained Vision-Language-Action models provide a strong foundation for robot learning, but sequentially adapting them to diverse skills can perturb the representations and velocity mappings used by previous skills, leading to catastrophic forgetting. Architecture-based approaches improve retention by isolating skills but lead to increased inference footprint. Recent subspace-constrained methods restrict parameter updates in an orthogonal subspace to minimize interference but impose a unified constraint on the entire model. We analyze the distinct roles of internal VLA components and identify two VLA-specific challenges. First, the VLM maintains broad semantic representations, making it vulnerable to capacity exhaustion, whereas the ActionHead refines semantics into localized velocity patterns that are highly sensitive to perturbations. Second, the final velocity decoder serves as a readout layer. Freezing it forms an output-stage expressivity bottleneck, while updating it risks overwriting previous velocity mappings. To this end, we propose OrthoSkillVLA, a parameter-efficient framework for continual skill learning in pretrained VLA models without demonstration replay. Given the representation heterogeneity, we impose separate subspace constraints on the VLM and ActionHead, preserving reusable semantic capacity while protecting localized velocity patterns. For the output layer, we introduce a lightweight feature-aware MoE decoder, where each skill is allocated a compact expert and a training-free router selects the expert according to feature-space affinity. Extensive simulated and real-world evaluations, together with ablations, demonstrate that OrthoSkillVLA better preserves prior skills while acquiring new ones.

Background: Ecological momentary assessment (EMA) is increasingly used in mental health research, but research-grade deployment requires platforms supporting protocol configuration, automated delivery, participant management, and data export. In China, these requirements are not consistently supported. Objective: We aimed to identify workflow gaps affecting localized EMA deployment, develop an implementation-oriented framework for platform assessment, and assess Huixin EMAI. Methods: We reviewed EMA platforms reported in Chinese mental health studies in CNKI and Wanfang. A multidisciplinary panel of 6 experts developed the Multi-dimensional EMA Platform Evaluation Framework (MEPEF) and benchmarked 7 platforms across 43 indicators in 6 domains. MEPEF was then applied to Huixin EMAI using deployment logs from 48 participants, questionnaires from 44 participants, and semistructured interviews with 6 researchers. Results: We identified 66 empirical studies. Most relied on instant-messaging-based workflows (36/66, 54.5%), whereas specialized EMA platforms were less common (14/66, 21.2%). MEPEF provided a 6-domain framework for cross-platform benchmarking and highlighted a trade-off between localized deployability and advanced research functions. In a Huixin EMAI deployment, 1893 of 2472 expected prompts were completed (76.6%), with a median response latency of 4.0 minutes (Q1-Q3 0.0-13.0). Participant feedback indicated favorable acceptability; researchers reported support for core workflows but gaps in control, delivery monitoring, and data readiness. Conclusions: The main challenge for EMA in Chinese mental health research appears to lie less in feasibility than in recurring workflow gaps affecting localized deployment. This study translates these gaps into structured evaluation and design targets, providing an implementation-oriented pathway for advancing localized EMA platforms.

While entropy regularization is widely used to stabilize and accelerate Natural Policy Gradient methods, its ability to yield faster convergence rates for the unregularized objective remains underexplored. Existing analyses often rely on double-loop architectures and invoke a linear entropy penalty. To bridge the gap between theory and practice, we analyze a single-loop, entropy-regularized Natural Actor-Critic algorithm under compatible linear function approximation. By training an uncentered critic, our critic tracking can remain stable even as the training policy approaches determinism and the Fisher information matrix degenerates. We focus on two primary regimes for the optimization landscape: a Stochastic Regime, where we fuse coupled actor-critic updates into a joint Lyapunov recurrence, and a Deterministic Regime, where we pivot to a Policy Mirror Descent framework to circumvent the collapse of Euclidean geometry. By exploiting a positive Minimal Action Gap in the unregularized Markov decision process, we introduce an Exponential Translation mechanism that maps the regularized gap to the unregularized one up to an exponentially decaying tail. By tuning the fixed temperature, our algorithm achieves accelerated unregularized convergence rates, up to approximation-error terms: $\tilde{\mathcal{O}}(T_{total}^{-1})$ in the Stochastic Regime, and $\tilde{\mathcal{O}}(T_{total}^{-2/3})$ for the average iterate alongside $\tilde{\mathcal{O}}(T_{total}^{-1/3})$ for the last iterate in the Deterministic Regime. Here, $T_{total}$ denotes the total number of stochastic critic updates (or Monte Carlo rollouts). Furthermore, in the tabular setting, our positive-action-gap analysis yields a $\tilde{\mathcal{O}}(T_{total}^{-2/3})$ average-iterate rate, surpassing the $\mathcal{O}(T_{total}^{-1/2})$ worst-case statistical barrier that applies without a positive action margin.

We study landscapes for complex-parameterized networks. Our approach is motivated with an information-theoretic manifold perspective of the parameter and via classical optimization guarantees although of complex geometric variety such as through Dolbeault asymptotics. The descent path admits a Kähler information metric under a cross-entropy via the Wirtinger Hessian on the log-likelihood potential. We restrict attention to a descent update rule with natural gradient descent via a differentiated loss scaled by the inverse metric, so the descent path remains in the holomorphic tangent bundle. We emphasize Calabi-Yau information manifolds which profane theoretical guarantees via an ill-curvature-conditioned landscape. Under a Calabi-Yau metric, specifically in a non-compact setting with a global potential so defined geometrically rather than invoking the topological requirements of the Calabi conjecture, a wedged nowhere-vanishing holomorphic form is the top exterior product of the Kähler form up to constants, yielding a constant determinant condition. Under a fixed determinant, a metric almost low rank up to an eigenvalue tolerance implies a blow-up effect. Moreover, it has been discovered that negative curvature subverts the loss landscape, specifically sectional curvature, so we expand on this and draw interconnections to negative-definite Ricci curvature. Our arguments primarily exist in a geometric analytic modality, although we establish roots in deep learning theory such as through asymptotics at initialization and connections through failure modes of neural network guarantees under vanishing and negative Ricci curvature.

Recent studies suggest that video generation models can exhibit certain forms of zero-shot visual reasoning through generated frames. Yet reliable evaluation remains challenging: benchmarks should adopt inputs aligned with the visual priors of current video models, require valid evolving processes rather than only plausible final states, and calibrate task difficulty to remain challenging yet partly feasible. To this end, we introduce VGI-bench, containing 27 tasks and 810 instances, organized by a two-level taxonomy of task domains and skill tags for fine-grained evaluation of visual reasoning capabilities of video generation models. Our evaluations show that current generative systems can solve a subset of visually grounded reasoning tasks, but remain far from reliable, with even the strongest model, Seedance~2.0, achieving only 51.0% under our evaluation criteria. Our analysis further explore the output failure modes, input condition sensitivity, performance transfer boundary from synthetic fine-tuning, and internal denoising perspective revealing limited self-correction, where later steps mainly refine early hypotheses rather than correct reasoning errors. We hope VGI-bench will help stimulate the development of next-generation video generation models. We will release our code and data.

Vessel trajectory prediction is critical for maritime safety and accident prevention. While most existing trajectory prediction models rely on Automatic Identification System (AIS) data due to its precision and availability, small vessels mostly operate without AIS, resulting in a significant monitoring gap. To address this, we propose Mix&Fix-Net, a dual-stage mixer-based trajectory prediction model designed to handle vessel trajectory time-series data derived from both AIS and (non-AIS) vision data. Our architecture integrates a Primary Trajectory Predictor with a Residual Trajectory Adjuster, enabling more refined trajectory prediction. Additionally, we introduce a new video-based dataset derived from webcam streams, from which vessel trajectories are extracted to represent non-AIS data. Extensive evaluations on both AIS and non-AIS datasets across six metrics (mean squared error, mean absolute error, symmetric mean absolute percentage error, final displacement error, Frechet distance, and average Euclidean distance) demonstrate that Mix&Fix-Net consistently outperforms existing baselines across most metrics and datasets.

Large Language Models (LLMs) are increasingly deployed in high-stakes applications, yet their tendency to generate toxic, harmful, or policy-violating content poses significant risks. Detecting these unsafe outputs efficiently in a black-box manner remains an open challenge. In this paper, we extend a recently proposed dynamical systems framework designed for hallucination detection to LLM safety classification. By projecting both prompts and responses into high-dimensional embedding spaces and fitting separate Koopman-based predictive models for safe and unsafe regimes, we classify new outputs using a new differential residual score that compares prediction errors of the safe and unsafe regimes. A key contribution is the incorporation of the prompt and response embedding dynamics, yielding fitted Koopman operators that capture crucial interaction patterns. We evaluate our black-box method across three safety benchmarks using three embedding models. Our results show that incorporating prompt embeddings yields consistent improvements, particularly for interaction-dependent violations when paired with causal decoders (e.g., in Llama-3), while response-only violations benefit more from dense semantic embedding representations. These findings opens the door for using dynamical systems to analyze AI systems rather than the dominant paradigm of using AI to model dynamical systems.

Acute exacerbation of chronic obstructive pulmonary disease (AECOPD) can worsen rapidly, making timely prediction a clinical priority. Most existing machine learning approaches rely on episodically collected clinical variables, introducing delays that limit their practical utility in home monitoring settings. Home ventilators offer a lower-latency alternative, producing a near-continuous record of respiratory status during daily use. However existing ventilator-based approaches either compress the waveform into handcrafted features or focus primarily on binary risk classification, leaving the timing of an impending event unresolved. In this paper, we present a two-stage framework that operates directly on raw pressure and flow waveforms from the most recent seven days of home ventilator use. The first-stage classification model identifies patients at high risk of a severe exacerbation. The second-stage regression model then estimates how many days remain before the event occurs. Our experimental results demonstrate that the two-stage model outperforms traditional baseline models on both risk classification and time-to-event estimation, with our selected Stage 1 classifier achieving F1 = 0.91 and our Stage 2 regression model achieving RMSE = 1.00 days and R^2 = 0.76, giving clinicians both an early warning and actionable lead time before a severe exacerbation occurs.