Accurate cardiac output (CO) estimation from photoplethysmography (PPG) is promising for unobtrusive hemodynamic monitoring, but remains difficult since CO is jointly determined by cardiac function and vascular tone. Conventional feature-based models use physiologically meaningful PPG descriptors, yet depend on accurate pulse detection and may miss latent temporal relationships. In contrast, fully end-to-end deep learning models learn directly from raw PPG but often underuse established PPG-derived prior information. Here, we introduce the Cross-View Attention Fusion Network (CVAF-Net), a prior-guided dual-view deep learning model for CO estimation from short, fixed-length PPG segments. CVAF-Net processes raw PPG as a temporal view and a feature sequence map (FSM) as a structured prior-guided view, and fuses the two representations through cross-view attention. The model was independently evaluated using 5-, 15-, and 30-s segments from three datasets: simulated pulse waves (3323 subjects), vasoconstriction provocation (79 subjects), and resting/cycling activities (10 subjects), and was compared with multiple machine learning and deep learning benchmarks. CVAF-Net outperformed most benchmark methods and achieved performance comparable to a state-of-the-art Transformer-based model, with a mean absolute error (MAE) of 0.19 L/min (MAPE: 3.95%) on simulated data and high accuracy in real-world settings (minimum MAE: 1.20 L/min). Importantly, CVAF-Net reduced FLOPs by twelvefold compared with the leading Transformer-based model. Plausibility analysis showed physiologically consistent CO estimates, with expected correlations with age ($ρ= -0.274$), heart rate ($ρ= 0.894$), and systemic vascular resistance ($ρ= -0.740$). These findings indicate that CVAF-Net provides an accurate, computationally efficient, and generalizable approach for continuous wearable-based CO monitoring.

Large Reasoning Models (LRMs) have demonstrated remarkable capabilities in solving complex problems by generating structured, step-by-step reasoning content. However, exposing a model's internal reasoning process introduces additional safety risks; for example, recent studies show that LRMs are more vulnerable to jailbreak attacks than standard LLMs. In this paper, we investigate jailbreak attacks on LRMs and reveal that the attack success rate (ASR) is closely correlated with LRMs' attention patterns. Specifically, successful jailbreaks tend to assign lower attention to harmful tokens in the input prompt, while allocating higher attention to those tokens in the reasoning content. Motivated by this finding, we propose a novel jailbreak method for LRMs that leverages reinforcement learning (RL) to enhance attack effectiveness, explicitly incorporating attention signals into the reward function design. In addition, we introduce diverse persuasion strategies to enrich the RL action space, which consistently improves the ASR. Extensive experiments on five open-source and closed-source LRMs across three benchmarks demonstrate that our method achieves substantially higher ASR, outperforming existing approaches in terms of effectiveness, efficiency, and transferability.

We formulate quantum geometry for non-Hermitian systems under open boundary conditions. By defining quantum-geometric quantities in both real-space and non-Bloch representations, we establish a unified framework beyond conventional Bloch band theory. Our central result is an exact equivalence between the real-space integrated quantum metric and a non-Bloch integrated quantum metric defined on the generalized Brillouin zone. We further introduce localized non-Bloch Wannier functions in the presence of the non-Hermitian skin effect and show that the non-Bloch integrated quantum metric gives the gauge-invariant part of their spread functional. These results establish quantum geometry as a natural framework for characterizing open-boundary non-Hermitian band structures and the localization properties encoded in skin modes.

Abstract Hippocampal time representation can flexibly scale in response to changes in the temporal structure of the environment, a mechanism known as temporal scaling. However, how neural circuits can learn these scalable representations of time remains unclear. To address these questions, we trained a recurrent neural network (RNN) agent using deep reinforcement learning (DRL) to perform a temporal bisection task in a 2D maze. In this task, short and long delays during an initial waiting phase corresponded to distinct reward locations in the subsequent navigation phase. When short and long delay durations were systematically scaled across different blocks, temporal scaling, at the single-neuron level, naturally emerged in the activity of trained RNN units during the waiting phase. At the population level, neural activity followed low-dimensional trajectories that preserved their geometry across scaled blocks by modulating trajectory speed to reach similar terminal states. In contrast, navigation-phase activity showed two mutually inhibitory neural ensembles that remained stable across scaled contexts, maintaining a persistent memory of the perceived long or short delay duration. Extending the task to three delays yielded three stable attractor states while preserving the scalability property of the network dynamics. These results reveal how reinforcement learning shapes recurrent circuits to generate scalable temporal representations, providing a mechanistic account for experimentally observed hippocampal dynamics.

We extend the CCSDTQ implementation in CFOUR to UHF and ROHF references and demonstrate its efficiency. We apply it to basis set convergence of post-CCSDT(Q) corrections for the W4-08 thermochemical dataset. Convergence of (Q)$_Λ$--(Q) is relatively rapid. For difficult species (e.g., B2, O3), CCSDTQ--CCSDT(Q)$_Λ$ may converge more slowly than (5)$_Λ$, but the effects and and basis-set trends oppose each other. Consequently, a single-shot CCCSDTQ(5)$_Λ$-CCSDT(Q)$_Λ$ correction appears most efficient. For radicals with bifurcating UHF solutions, energetics of the `less spin-contaminated' solution are clearly more well-behaved. Our best computed adiabatic electron affinity of ozone is in excellent agreement with experiment.

Deep learning models detect pneumonia from chest X-rays with high accuracy, but the performance declines under domain shifts caused by differences in devices, patients, or institutions. We present PneumoNet, a domain-incremental learning method for point-of-care pneumonia diagnosis in resource-limited settings. PneumoNet combines a lightweight CNN for on-device prediction, a dual-stage balanced buffer for class-balanced replay, and a dynamic class-weighted loss to correct training-batch imbalances. Evaluated on a domain-shifted PneumoniaMNIST dataset simulating five realistic domain change scenarios, PneumoNet achieves 86.6% accuracy with 1.4% forgetting while being smaller and faster than existing baselines. These results highlight PneumoNet's potential to enable adaptive, privacy-preserving diagnostic AI directly on point-of-care medical devices in real-world and pandemic-ready healthcare.

Automatic report labeling facilitates the identification of clinical findings from unstructured text and enables large-scale annotation for medical imaging research. Existing rule-based labelers struggle with the diverse descriptions in clinical reports, while fine-tuning pre-trained language models (PLMs) requires large amounts of labeled data that are often unavailable in clinical settings. In this paper, we propose PromptRad, a knowledge-enhanced multi-label \textbf{prompt}-tuning approach for \textbf{rad}iology report labeling under low-resource settings. PromptRad reformulates multi-label classification as masked language modeling and incorporates synonyms from the UMLS Metathesaurus into a multi-word verbalizer to enrich category representations. By fine-tuning the PLM without additional classification layers, PromptRad requires substantially less labeled data than conventional fine-tuning. Experiments on liver CT reports show that PromptRad outperforms dictionary-based and fine-tuning baselines with only 32 labeled training examples, and achieves competitive performance with GPT-4 despite using a much smaller model. Further analysis demonstrates that PromptRad captures complex negation patterns more effectively than existing methods, making it a promising solution for report labeling in data-scarce clinical scenarios. Our code is available at https://github.com/ila-lab/PromptRad.

We adapt split conformal prediction and adaptive conformal inference (ACI) to continuous AI agent evaluation, providing distribution-free coverage guarantees for forecasted quality scores. Conformal intervals achieve calibration error below 0.02 across all nominal levels at the 24h horizon, while ACI correctly widens intervals by 35% following agent releases then reconverges. We further develop compositional uncertainty bounds for multi-agent pipelines (validated via simulation across inter-stage correlations rho in [-0.5, 0.9]), a conformal abstention rule for pairwise rankings with controlled false-ranking rate, and FDR-corrected abstention for leaderboard-scale multiple testing. Evaluating 50 agents via 18 real-time signals collected hourly, we show that per-agent conditional coverage is well-concentrated around the nominal level (mean 80.4%, 90% of agents within [72%, 90%]), and that cross-source sentiment divergence predicts ranking instability (r=0.64, p<0.01). A circularity-controlled validation confirms the framework captures signal beyond benchmarks (rho_s=0.52, p<0.01, n=35). Code and data are released under CC BY 4.0.

In this article, we characterize the holomorphic mappings from $B_{\ell_p^n}\times\mathbb{D}^{m}$ into $\mathbb{D}^{m}$ for $p\in \{2,\infty\}$. In addition, we give a simple proof for the boundary Schwarz lemma for vector valued holomorphic functions, which also extends the existing result. Also, we obtain the boundary Schwarz lemma for pluriharmonic self-mappings of the unit ball $B_{\ell_p^n}$, $p \in [2,\infty]$. Furthermore, we establish the boundary rigidity theorem for holomorphic self-mappings of $B_{\ell_p^n}$, $p \in (1,\infty)$.

Multiqubit positive-partial-transpose (PPT) entangled states play an important role in quantum information theory. We characterize such states of minimum rank in three-qubit system, namely rank four. Depending on whether the Lorentz invariant is zero, we classify such states into two types. The PPT entangled states constructed by unextendible product bases (UPB) have nonzero invariants, which belong to type I. We provide a method to effectively determine whether a state can be constructed from UPB. For states with zero invariant, which belong to type II, we provide an explicit expression up to equivalence of stochastic local operations and classical communications (SLOCC). It turns out that we can represent them with only one complex parameter. We further study SLOCC-equivalence relation within the expression. We also investigate the Lorentz invariants of multiqubit states with rank less than three and analyze their range.