Papers

Early detection of oral cancer via photographic imaging presents a promising avenue for large-scale oral cavity screening. However, the development of robust deep learning models is frequently hampered by the scarcity of high-quality, annotated datasets. To address this limitation, a novel and well-curated resource, the Photographic Multi-purpose Oral Cancer Imaging (PhotoMOCI) dataset, is introduced for developing models across multiple diagnostic tasks in oral oncology. Then, a comprehensive benchmark study was conducted to investigate how various data augmentation strategies influence the performance of image classifiers. Our analysis spans different generative AI frameworks, evaluating the efficacy of traditional methods against advanced generative approaches, including Generative Adversarial Networks (GANs) and Diffusion Models (DMs). Additionally, we propose the Synthetic Image Filter (SIF), a mechanism to select specific samples based on two auxiliary models: Synthetic Proxy Classifier to ensure samples are representative of the target class and Synthetic Image Detector to verify they appear realistic, thereby selecting only the high-utility images that contribute to improving downstream performance. Across the evaluated datasets and classifiers, the best SIF-filtered setup improves accuracy over traditional augmentation in all cases, with gains of +1.73% and +2.35% on PhotoMOCI and +2.38% and +2.08% on KOCD for ResNet50 and ViT, respectively. Our findings reveal that while the direct application of generative data augmentation may yield performance drops, the integration of SIF, considering (i) how synthetic data looks real and (ii) how it reflects the discriminative features of the belonging class, provides a simple yet effective mechanism to filter out synthetic samples that confuse the classifier during training.

Sign language translation and generation share the goal of bidirectional alignment between text and sign representations. However, existing approaches either treat them as isolated tasks or are only verified on limited datasets, limiting effective modeling between modalities. In this paper, we propose SignFLIP, a unified LLM-centered framework for translation and generation. To enable bidirectional mapping between text and sign, SignFLIP adopts a symmetric architecture together with a stage-wise training strategy built on large-scale data. The shared sign–text representation is progressively refined: pre-alignment facilitates subsequent SLT, while the SLT-adapted representation further benefits SLG. Extensive experiments on multiple benchmarks show that SignFLIP shows competitive performance compared with task-specific models on both translation and generation tasks, as well as strong transferability to sign language recognition.

Masked-diffusion language models fill in masked positions in parallel and in no fixed order. Most practical text watermarks assume left-to-right generation. They key each token to the tokens before it, and in a diffusion model those tokens may still be masked. A fixed green list needs no such context, but it favors the same tokens at every position, so these tokens appear more often in watermarked text. An attacker who compares token frequencies in watermarked and unwatermarked text can recover the list and forge text that the provider's own detector accepts. We present TANGO, a watermark for masked-diffusion language models that keys each new token to a nearby token that is already unmasked. A secret key splits the vocabulary into color classes, and TANGO biases the new token toward a color determined by the key and the nearby token's color. The watermark is therefore embedded in pairs of tokens. Because the favored color changes from position to position, token frequencies stay much closer to those of unwatermarked text than under a fixed green list. Detection needs only the text and the key, and it does not assume any unmasking order. On two masked-diffusion models, TANGO detects nearly all unedited watermarked texts and most edited ones, and frequency attacks that forge the fixed green list fail against it.

Electroencephalogram (EEG) artifact removal remains a critical challenge for real-world brain-computer interface deployment due to non-stationary signal statistics and limited calibration availability. Artifact Subspace Reconstruction (ASR) provides an automated framework for variance-based artifact suppression but relies on a static calibration-derived covariance model, limiting stability under intra-subject drift and increasing sensitivity to parameter mis-specification. This work proposes an adaptive ASR framework based on exponential moving average (EMA) subspace tracking. By integrating short-horizon covariance estimation with slow recursive assimilation, the method establishes dual adaptation timescales that enable responsiveness to evolving EEG structure while preserving stability against transient artifacts. This formulation further promotes smoother operating characteristics across rejection thresholds, reducing sensitivity to hyperparameter selection in continuous deployment settings. Evaluated on 24-channel EEG recordings from 10 subjects with prominent blink artifacts across cognitive states, EMA-ASR achieved substantially stronger artifact attenuation than Original and memory-limited ASR variants (53.1% vs. 24.0% and 26.7% blink reduction), at the cost of increased reconstruction and spectral deviation, whereas the memory-limited variant performed comparably to Original ASR. These findings position EMA-based adaptive tracking as a lightweight, deployment-oriented approach to continuous EEG artifact removal under non-stationary conditions, with an explicit trade-off between suppression strength and signal preservation.

Loop Quantum Cosmology (LQC) replaces the big bang singularity of the homogeneous universe by a bounce, usually described by the modified Friedmann equation $H^2=\rho(1-\rho/\rho_c)/(3M_p^2)$. We show that this background dynamics follows from a local cuscuton effective theory, whose scalar equation is a constraint rather than a wave equation. One way to establish this is to write the cuscuton in terms of an angular coordinate $\theta$, identified with the LQC polymerization angle $2\lambda b$. Its constraint gives $H\propto\sin\theta$, while the Einstein constraint gives $\rho=\rho_c\sin^2(\theta/2)$, exactly reproducing the LQC bounce. To our knowledge, this is the first closed-form, local, generally covariant realization of the exact standard flat-FLRW LQC background dynamics for minimally coupled matter satisfying null energy condition, without introducing additional local dynamical degrees of freedom. A branchwise Legendre transformation establishes a canonical equivalence between the clock-gauge-fixed homogeneous mimetic system in the case of vanishing mimetic dust energy density and the cuscuton systems. It selects the constant-tension cuscuton coordinate $\Theta=\sin\theta-\theta$, while $\theta$ retains its interpretation as the LQC polymerization angle. This equivalence does not establish agreement of the inhomogeneous theories or their perturbations. The angular action also admits an explicit branched $f(K)$ representation, where $K$ is the mean extrinsic curvature of spatial hypersurfaces. The construction is therefore an effective covariant representation of the LQC holonomy correction, not a derivation from full loop quantum gravity.

The visibility of Ly$\alpha$ emission during the epoch of reionization depends on both galaxy properties and the transmission of Ly$\alpha$ photons through the intergalactic medium (IGM). Using JWST/NIRCam F470N imaging, we identify a prominent projected overdensity of H$\alpha$ emitters (HAEs) at $z\simeq6.2$ in CEERS field. We use Subaru/HSC NB872 imaging to measure their Ly$\alpha$ fluxes and compare the Ly$\alpha$ visibility of galaxies within $2'$ of the density center with that of galaxies outside this region. The Ly$\alpha$-emitter (LAE) fraction is $0.41^{+0.15}_{-0.14}$ in the inner region and $0.27^{+0.19}_{-0.15}$ in the outer region. Stacking the H$\alpha$ and Ly$\alpha$ images gives median Ly$\alpha$ escape fractions of $0.100^{+0.091}_{-0.053}$ and $0.063^{+0.130}_{-0.063}$ in the inner and outer regions, respectively. Both measures indicate higher Ly$\alpha$ visibility in the inner region, although only at modest significance. HAEs closer to the density center are systematically more massive and have redder UV slopes, which instead favor lower Ly$\alpha$ escape. The environmental difference in Ly$\alpha$ visibility is therefore difficult to explain solely by variations in underlying galaxy population. Our results are consistent with reduced Ly$\alpha$ attenuation through a more highly ionized IGM in the overdense region, as expected if large H\,ii bubbles preferentially form around such galaxy overdensities. The wide-field LAE distribution provides complementary support for this scenario, as the HAE-defined density center coincides with the strongest LAE overdensity across the $1.4\,{\rm deg}^2$ HSC field. These findings highlight the importance of spatial variations in IGM attenuation in shaping Ly$\alpha$ visibility near the end of reionization.

Taxi-based mobile sensing has emerged as a cost-efficient paradigm for large-scale urban environmental monitoring. In practice, both passive and active sensing strategies are adopted by ride-hailing platforms. Passive sensing, conducted during passenger-serving trips, is constrained by stochastic and spatially imbalanced ride demand, leading to limited and uneven coverage. Active sensing, executed by vacant taxis, provides greater control over sensing operations but incurs additional operational costs, and is thus typically treated as a supplementary strategy. To address the inefficiencies of the conventional "passive-first, active-second" paradigm, which may delay the monitoring of critical Points of Interest (POIs) and increase system-wide costs, we propose a Distributionally Robust Optimization (DRO)-based framework for active mobile sensing. First, we develop an enhanced A*-based drive-by sensing routing policy that integrates vehicle-task matching while capturing both global routing efficiency and local sensing opportunities. Second, we formulate the active sensing problem as a DRO model that explicitly accounts for uncertainty in ride demand through an ambiguity set of probability distributions, enabling robust decision-making for vacant taxi routing and vehicle-task matching. The proposed framework is evaluated on both static and dynamic mobile sensing settings using real-world data. Computational results demonstrate that our approach achieves better performance in terms of sensing coverage, operational cost, and robustness compared to benchmark strategies, highlighting the value of integrating distributional robustness into taxi-based sensing operations.

Relational deep learning models database rows and foreign-key links as a heterogeneous graph for prediction from record attributes and relational context. These graphs contain two distinct temporal signals: record age changes with the prediction cutoff, while intervals between observed records remain fixed. Prior work often treats time as a single signal or studies temporal representation and pretraining separately. We investigate how explicitly encoding both signals affects temporal pretraining for downstream tasks. Our framework combines Multi-scale Time Encoding, which captures record age using learnable time scales and type-specific projections, with Rotary Time Encoding, which represents signed inter-record intervals through rotary transformations during graph propagation. We pair these encodings with three self-supervised objectives: historical relation recovery, horizon-aware future relation activity prediction, and temporal subgraph contrast. All inputs respect their observation cutoffs. Pretraining proceeds in two stages: subgraph contrast first learns neighborhood representations, followed by refinement through either relation recovery or future activity prediction. We evaluate on five RelBench datasets across 11 classification and regression tasks using heterogeneous GNN and graph Transformer backbones. With both encodings, the best evaluated staged schedules improve over supervised training with the same encodings by 3.02% and 1.06% on the two backbones, respectively, and over controls without pretraining or either encoding by 3.24% and 2.37%.

Large-scale antenna arrays are essential to future wireless networks but impose significant hardware and digital-processing burdens. The microwave linear analog computer (MiLAC) offers a promising architecture for addressing these challenges. In a MiLAC, beamforming is realized entirely in the analog domain through a reconfigurable multiport microwave network comprising tunable admittance elements. However, existing studies assume that these elements are continuously tunable, whereas practical implementations can support only finitely many configurable states. This paper investigates finite-resolution MiLAC-aided multiuser beamforming under lossless and reciprocal constraints with prescribed connectivity patterns. Each tunable susceptance is selected from a symmetric uniform finite-resolution codebook. We formulate an online problem that jointly optimizes the discrete susceptances and power allocation for a prescribed codebook, and an offline problem that designs the codebook based on channel statistics. By exploiting the odd-uniform codebook structure, we express each multilevel susceptance as a weighted sum of two-level variables and develop exact continuous penalty models for both the online beamforming and offline codebook design problems. We prove that, for sufficiently large penalty parameters, these penalty models are globally equivalent to their original discrete counterparts. Building on these formulations, we develop efficient ADMM-based algorithms for solving the two problems. Numerical results demonstrate that the performance of an unquantized fully-connected MiLAC can be approached with substantially reduced connectivity and finite resolution. In particular, a 3-bit stem-connected MiLAC achieves more than 95% of the sum-rate performance of the unquantized fully-connected MiLAC while requiring only 22% of its tunable components.

Modern probabilistic machine learning models increasingly produce multivariate outputs with complex dependence structure, from multi-step time-series forecasts to sample path predictions. Understanding which input features drive the predictive uncertainty is important for risk-aware decisions, model diagnostics, and deciding whether the uncertainty should be mitigated or hedged against. This attribution problem requires a choice of how dependencies between output components are treated. Existing approaches reduce the output to a scalar through aggregation or projection before attribution, thereby obscuring whether features affect marginal uncertainty, dependence structure, or both, while component-wise analyses can miss dependence effects entirely. We close this gap by introducing a hierarchy of three entropy-based Shapley games that make this output-side choice explicit for any ordered multivariate outcome, ranging from per-component marginal entropy to fully joint entropy. The hierarchy isolates a cross-component attribution term that captures how each feature shifts the dependence between output components, a quantity invisible to component-wise methods. We establish a chain-rule decomposition of the joint attribution and characterize the cross-component term through conditional total correlation, providing both closed-form and sample-based estimators. Finally, we demonstrate how the framework captures differences in learned joint structure across probabilistic models from distributional regression to a zero-shot time series foundation model.

Large Language Models (LLMs) enable scalable social survey simulation, yet existing pipelines typically use the same strong general-purpose model and a fixed, often large, respondent history for every respondent-question request. This uniform approach overlooks three factors. First, stronger models may rely on their own knowledge rather than respondent-specific evidence while costing more. Second, additional history can help when evidence is limited, but irrelevant responses may add noise and increase input length. Third, the preferred model and history budget can depend on each other. We propose E2Sim, an adaptive resource allocation framework that jointly selects a model and history budget for each request. Given a respondent persona, ranked response history, and target question, a lightweight policy predicts the accuracy and cost of each configuration and selects the most suitable one. We use respondent-history drop-and-swap augmentation to improve robustness to incomplete or variable histories, and a margin-based curriculum that progresses from easier allocation decisions to harder ones. Experiments on four real-world social survey datasets, multiple model pools, and different history-budget spaces show improvements over the oracle best-fixed configurations, with accuracy gains of up to 6.3 percentage points and cost reductions of up to 55.3 percent. Code is available at https://anonymous.4open.science/r/E2Sim-DE66.

Terminal utility evaluates a complete agentic workflow, but learning requires credit for the decisions within it. We introduce Attentive Search over Counterfactual Trees (ASCT), a framework that turns training-time multi-step search into local action credit. At actor-visited states, an auxiliary tree evaluates alternative legal actions from the same recoverable prefix. Its action-value table is centered by the frozen actor's probabilities and supplies credit for PPO on actor-sampled trajectories. This protocol connects counterfactual evaluation to policy learning while deploying the actor alone. Uniform, UCT, and cost-aware AgentUCT instantiate the framework. On HotpotQA agentic retrieval-augmented generation, all three improve mean held-out utility over trajectory-return PPO and workflow-adapted VinePPO. Across three seeds, ASCT-AgentUCT reaches 0.6187 utility versus 0.5939 for VinePPO, with gains in answer F1 and execution cost, and uses 50.3% fewer recorded auxiliary Qwen tokens. Transfer and component-description studies examine the learned policies beyond the training setting.

Cardiac model personalisation requires inferring mechanical parameters that are not directly measurable in vivo. Ultrafast ultrasound shear wave elastography (SWE) enables non-invasive tracking of myocardial stiffness dynamics over the cardiac cycle, providing a target for personalisation. However, mapping these observations to subject specific model parameters remains ill-posed, as multiple parameter sets can reproduce the same stiffness dynamics. We formulate SWE-informed personalisation as a statistical inference problem using simulation-based inference (SBI). Using a subject-adapted 0D cardiovascular model and neural posterior estimation, we estimate model-conditional posterior distributions over active stiffness scale k0, contraction rate kATP, and relaxation rate kSR, conditioned on SWE-derived curve features and subject specific context. Among six healthy volunteers, four passed objective prior-support diagnostics and were retained for quantitative posterior analysis. Curve-level RMSE against the observed SWE target decreased from 12.61 $\pm$ 5.55 kPa for the prior predictive median to 1.14 $\pm$ 0.38 kPa for the posterior predictive median, an 89.7 $\pm$ 4.2% reduction. Posterior analysis revealed parameter-specific uncertainty, k0-kATP compensation, weaker constraint of kSR, and the importance of prior-predictive diagnostics for assessing whether each subject is represented within the modelled SWE feature space. These results support SBI for uncertainty aware SWE-based personalisation, while identifying prior support and forward-model adequacy as key diagnostics.

For vertex-transitive Ramanujan graphs with logarithmic girth, a simple random walk bridge of length of order $\log N$, where $N$ is the size of the graph, has a maximum distance that changes from order $\sqrt{\log N}$ to order $\log N$ in a bounded critical window. We prove this by separating bridges whose lifts to the regular tree close from those whose lifts do not. A uniform two-term return estimate determines the probabilities of these two cases and the real-valued critical center. In the same $O(1)$ window, the normalized range has a two-point limiting law whose mixture weights vary nontrivially across the window.

Multi-view clustering aims to capture cross-view consistency while exploiting view-specific information. However, shared representations learned to capture cross-view consistency may still retain view-identifying information, potentially compromising the consistency of cross-view clustering structures. To address this issue, we propose ACGRL, an adversarial consistency-guided representation learning framework for multi-view clustering. ACGRL employs a gradient-reversal view discriminator to reduce view identifiability and obtain invariant reference representations. These representations are then frozen to provide fixed references for disentangling view-specific information from cross-view common information in the subsequent learning stage. The fixed reference representations are concatenated with the learned view-specific representations for reconstruction and clustering, with cross-view cluster alignment encouraging consistent clustering assignments. Experiments on four benchmark datasets demonstrate the superior clustering performance of ACGRL compared with representative multi-view clustering methods.

We prove that the weak Haagerup constant is multiplicative with respect to direct products, namely $\Lambda_{\mathrm{WH}}(G \times H)=\Lambda_{\mathrm{WH}}(G) \Lambda_{\mathrm{WH}}(H)$ for any locally compact groups $G$ and $H$. Knudby previously proved the submultiplicative inequality and asked whether equality always holds.

On-policy distillation (OPD) is a widely adopted post-training technique for LLM reasoning. It is commonly believed to transfer knowledge from a stronger teacher, yet what OPD actually distills into the student's internal representations remains unclear. We study this question with sparse crosscoders, which learn one feature dictionary shared by the student before and after OPD and the teacher. Standard crosscoder analyses, however, identify model-specific features but cannot tell how a model's use of its features changes, since all models are encoded into one set of feature activations. We therefore propose the swap readout, which reads each student checkpoint's feature activations on its own, measuring how training changes the student's use of each feature, even for checkpoints unseen by the crosscoder. Across three OPD settings, we find that OPD neither creates features nor passes on the teacher's own, and leaves the firing rates of over 98% of the student's frequently used features within 20%. We further examine the SFT warm-up on the teacher's rollouts that commonly precedes OPD and makes it more effective. Rather than adding features, the warm-up reweights the shared ones in two ways. First, it already raises and lowers many of the features that OPD later raises and lowers, doing part of OPD's work in advance. Second, it changes features that OPD alone would not, notably those for conversation format, reasoning style, and mathematical notation, and these changes persist through OPD. Imposing this reweighting on a directly distilled student's features, without changing its weights, brings its accuracy close to that of the warmed-up student, whereas the same change on shuffled features does not. Together, these findings suggest that OPD reweights existing features rather than acquiring new ones: the student learns from the teacher how to use the features they already share.

The Katok examples on $S^2$ are induced by Randers metrics obtained by perturbing the round metric by the standard rotational Killing field: allowing a scaling factor of the round metric gives a two-parameter family of Randers metrics $F_{\alpha,\beta}$, parametrised by positive real numbers $(\alpha,\beta)$. We compute the set of all $(\alpha,\beta)\in (0,2+\sqrt{3})^2$ for which $D^*(S^2,F^*_{\alpha,\beta})$, the unit codisc bundle with respect to $F^*_{\alpha,\beta}$, symplectically embeds into the round codisc bundle $D^*S^2$. This set has the structure of an infinite staircase. We also establish a dictionary between embeddings of these codisc bundles, singular $A_1$-ellipsoid embeddings, and $\mathbb{Z} _2$-equivariant ellipsoid embeddings, showing that these embedding problems are equivalent. Furthermore, we prove the analogous results for the $\mathbb{R} P^2$ case and discuss a broader family of examples for which this dictionary applies.

Vector-vortex photons, characterized by a nonseparable coupling between polarization and orbital angular momentum (OAM), offer opportunities for optical manipulation, quantum communication, nuclear photonics, etc. However, their generation in the $\gamma$-ray regime remains challenging. Here, we put forward a novel method to generate vector-vortex $\gamma$ photons via nonlinear Compton scattering in elliptically polarized laser pulses. We reveal that tailoring laser ellipticity directs the multiphoton absorption to coherently populate OAM modes with opposite winding numbers, $\pm \ell$, tied to orthogonal circular polarizations, producing nonseparable spin-OAM photon states. For a linearly polarized laser of moderate intensity (dimensionless amplitude $a_0 \sim 1$), the mode-pair concurrence–a 0-to-1 measure of spin-OAM entanglement–reaches unity for MeV $\gamma$ photons, realizing maximally nonseparable radial- or azimuthal-type vector-vortex states. The laser amplitude further controls the accessible OAM spectrum. Our method offers a route to MeV vector-vortex photons, opening a new avenue for nuclear-scale structured photonics and high-energy quantum information.

A major challenge in neuroscience and machine learning is to connect single-neuron influence, population dynamics, and circuit connectivity in a causal account of computation. Much previous work has shown that low-rank connectivity can generate low-dimensional dynamics in trained artificial neural networks, but this leaves unclear the functional relevance of the higher-rank structure of biological neural circuits and many artificial neuronal networks. Here we analyse recurrent neural networks trained to locate rewards by integrating continuously varying speed inputs in one- and two-dimensional spatial tasks. We find that dominant low-dimensional dynamics encode task locations, and can be causally manipulated to instruct behavioural outcomes. However, after decomposing the underlying circuitry we found that while low-rank connectivity accounts for the low-dimensional dynamics, accurate performance and generalisation to novel speed distributions requires high-rank connectivity. We demonstrate that this is achieved through distributed signalling that corrects errors in low-dimensional location representations. Thus, combined perturbation-, representation-, and circuit-level analyses demonstrate a novel mechanism for robust spatial computation and show how high-rank connectivity in neural circuits can provide a scaffold that supports precision and generalisation.

We study the stochastic first-order oracle complexity of finding $\epsilon$-stationary points of the primal function in smooth nonconvex-strongly-concave minimax optimization. For sufficiently small $\epsilon$, we establish lower bounds of $\Omega(\kappa L\Delta\sigma^2\epsilon^{-4})$ under the bounded-variance assumption and $\Omega(\kappa^{3/2}\bar L\Delta\sigma\epsilon^{-3})$ under the additional assumption of averaged smoothness. Here, $L$ and $\bar L$ denote the smoothness and averaged-smoothness constants, respectively, $\Delta$ is the initial primal gap, $\sigma^2$ bounds the oracle variance, and $\kappa=L/\mu$ or $\bar L/\mu$ in the respective settings, where $\mu$ is the strong-concavity parameter. Our bounded-variance lower bound improves the dependence on the condition number from $\kappa^{1/3}$ in previous lower bounds to $\kappa$, while our averaged-smoothness lower bound is the first of its kind. In both settings, the resulting lower bounds match existing upper bounds in their dependence on $\kappa$ and $\epsilon$. Our proofs are based on a unified quadratic lifting construction that transfers a hardness instance for stochastic nonconvex minimization to unconstrained minimax optimization while preserving the required variance and smoothness properties.

In this paper, we introduce OT-PCA, a novel approach for conducting Plaintext-Checking (PC) oracle based side-channel attacks, specifically designed for Hamming Quasi-Cyclic (HQC). By calling the publicly accessible HQC decoder, we build offline templates that enable efficient extraction of soft information for hundreds of secret positions with just a single PC oracle call. Our method addresses critical challenges in optimizing key-related information extraction, including maximizing decryption output entropy and ensuring error pattern independence, through the use of genetic-style algorithms. Extensive simulations demonstrate that our new attack method significantly reduces the required number of oracle calls, achieving a 2.4-fold decrease for hqc-128 and even greater reductions for hqc-192 and hqc-256 compared to current state-of-the-art methods. Notably, the attack shows strong resilience against inaccuracy in the PC oracle-when the oracle accuracy decreases to 95%, the reduction factor in oracle call requirements increases to 7.6 for hqc-128. Lastly, a real-world evaluation conducted using power analysis on a platform with an ARM Cortex-M4 microcontroller validates the practical applicability and effectiveness of our approach.

In 2005, Tuval et al. introduced a continuum model for the interaction of aerobic bacteria, oxygen, and an incompressible fluid. For the associated three-dimensional chemotaxis-consumption systems, global classical regularity for large data has remained a longstanding question, even after omitting the nonlinear convection term from the fluid equation. We answer this question and prove global classical solvability of the chemotaxis–Stokes Cauchy problem in \(\mathbb{R}^3\), with linear cell diffusion and bilinear signal consumption, under the following data assumptions. The initial density and signal are smooth, bounded, and nonnegative, with finite density mass and second moment and finite signal Fisher energy; the divergence-free initial velocity belongs to \(H^4\). No smallness assumption is imposed, and a positive constant signal background is allowed. The solution remains bounded on every finite time interval. The main new ingredient is a localized entropy estimate uniform under density normalization, even when the resulting consumption coefficient becomes arbitrarily large. A terminal representation of the signal separates two complementary mechanisms: consumption excludes normalized density concentration at positive-signal points, while a localized weighted estimate improves density integrability at zero-signal points. Together, these mechanisms yield decay of a scale-invariant density quantity and allow a local $\varepsilon$-regularity criterion to be applied without a separate smallness assumption on the signal gradient.

Nuclei instance segmentation is fundamental to quantitative pathology. Yet high-accuracy methods mostly build on large vision transformers (ViTs) or heavy convolutional backbones, whose parameter count and compute cost hinder deployment in clinical and edge settings. We present NucShare, a lightweight model whose shared decoder handles nuclear pixel segmentation, horizontal-vertical distance, and classification through learnable transposed convolutions for upsampling and three lightweight heads, using 37% of independent decoders' parameters with no loss in accuracy. Output-level distillation then transfers knowledge from NucShare-T (UNI2-h) to NucShare-S (ConvNeXt V2-Tiny), reducing 690.5M parameters to 34.7M. On PanNuke they reach mPQ 0.524 and 0.519, above the prior best CFR-SAM-H at 0.511, and NucShare-S surpasses HoVer-Net and Mask R-CNN by 5.6 and 15.0 points; on zero-shot MoNuSeg it attains bPQ 0.668 against CFR-SAM-H's 0.662. Code: https://github.com/L-iMeng-L/Newidea

We numerically study laser deceleration, cooling and trapping of host fast ortho-positronium (Ps) bunches as produced in silica microchannel targets using positron bunches from a Surko trap. While Doppler laser cooling routinely allows to cool neutral atom species below milli-Kelvin kinetic temperatures, the fast ground state annihilation and the high recoil shift to line-width ratio in Ps pose new challenging limitations on timing and the required laser power. Earlier theoretical work and first experiments already exhibit viable prospects for radiation pressure cooling to create unprecedented cold and dense Ps ensembles. Our extended numerical studies confirm very good prospects to simultaneously decelerate, cool and optically trap Ps atoms in a standard Doppler cooling geometry on the $1^3\mathrm{S}$–$2^3\mathrm{P}$ transition. Including the full Zeeman state manifolds and adding additional transverse lasers on the $2^3\mathrm{P}$–$3^3\mathrm{D}$ transition improves timing and final temperature to allow for optical trapping at the end. The required laser powers and geometry parameters to implement effective slowing and trapping within a time frame, where the majority of atoms is not annihilated, is in reach of current technology of $\mathrm{mJ}$ energy pulses of approximately $100\ \mathrm{ns}$ duration. Future extensions to collectively enhance cooling and trapping in optical cavities or hollow-core fibres should finally allow fast preparation of ultra-cold Ps systems as future basis of superradiant lasing and Bose–Einstein condensates.

Aligning language models with a specified normative framework requires translating abstract principles into concrete examples and preference signals from which models can learn. We present an expert-driven methodology for constructing such alignment data and apply it to a normative framework grounded in Islamic ethical, theological, and jurisprudential traditions. Over approximately one year, seven domain experts systematically probed language models to identify alignment deficiencies, curated desired responses, and constructed preference pairs from model outputs and expert judgments. The resulting Arabic-English datasets contain approximately 2.8K supervised fine-tuning (SFT) examples and 5.4K preference pairs spanning a broad range of normative domains. We evaluate the datasets through controlled post-training experiments comparing a Baseline model with models incorporating the curated SFT data alone and both the SFT and preference data. In blind expert evaluation on 150 separately constructed prompts, the model trained with the curated SFT data was preferred over the Baseline in 51.3% of assessor judgments, compared with 14.4% in the opposite direction (p < .001 at the prompt level). Adding the preference data resulted in a smaller difference, with the model trained with both datasets preferred over the SFT model in 28.0% of judgments versus 20.9% in the opposite direction; this difference was not statistically significant at the prompt level (p = .166). Standard Arabic and English benchmarks show no broad degradation in general-purpose capabilities. These results demonstrate how expert-defined normative principles can be systematically operationalized into alignment data and evaluated through controlled model training.

Memory-dependent manipulation requires robots to make decisions using information that is no longer available to their current sensors, such as recalling an earlier visual cue, tracking task progress, counting repeated events, or estimating elapsed time. We present ReCAT, a language-conditioned policy with structured recurrent memory. An instruction-conditioned encoder forms features from the current observation. A recurrent memory integrates the observation stream through Mamba-2 layers and one causal attention layer. A flow-matching Transformer decoder reads the current and the historical representation through separate cross-attention in every block. ReCAT reaches 95.3% average success on LIBERO and 62.4% on RMBench, with the best or tied-best result on six of nine tasks. On three real-robot tasks probing spatial recall, event counting, and interval timing, the best ReCAT variant reaches 66.7% average success, against 8.3% for the strongest short-history baseline. Controlled comparisons within ReCAT show that the observation encoder and every-block memory conditioning are needed for this performance. They also show that update rules developed for efficient sequence modeling behave differently as robot memory: additive updates have the highest observed success on counting and timing, and delta-rule updates on spatial recall. Project website is at https://intuitive-robots.github.io/ReCAT

LLM coding agents can traverse hundreds of intermediate code states before submitting a solution. Evaluating only the final artifact leaves the evolution of security findings unmeasured. We introduce the Security Debt Line Integral (SDLI), which accumulates static-analysis risk when an agent reaches a new best test pass ratio. We instantiate it with four static application security testing (SAST) tools and study artifacts from 830 passing SWE-bench runs, 712 ProgramBench final workspaces, and 13 public MirrorCode trajectories. The two large populations use the final-state special case of SDLI. Two-tool Common Weakness Enumeration (CWE) class agreement occurs in 3.9% of SWE-bench runs and 26.2% of the 80 ProgramBench runs passing at least 90% of official tests. These are scanner findings, not validated vulnerability rates. Excluding three advisory-heavy classes reduces the latter rate to 6.2%. Same-task runs differ in their measured scores, while one reconstructed ProgramBench run exposes persistent findings from its first implementation write. A repair case study reduces the scanner signal while preserving tested behavior, but also reveals sensitivity to equivalent API rewrites. SDLI offers a way to study progress and security findings together. Its value for steering agents and confirming exploitable vulnerabilities remains to be established.

The Prandtl-Glauert-Lorentz transformation establishes a deep mathematical correspondence between compressible potential flow theory and special relativity. Despite being implicit in the aerodynamic literature since the early twentieth century, this analogy remains largely unrecognised across the two communities and has never been surveyed in a unified manner. This review presents, in a self-contained and pedagogical framework accessible to both communities, the historical, mathematical, and applied dimensions of this correspondence. After tracing the parallel developments that gave rise to compressible aerodynamic theory and special relativity, we establish the formal equivalence between the Lorentz transformation and the Prandtl-Glauert coordinate change, and reinterpret relativistic concepts such as time dilation, the Doppler effect, and the Lorentz factor within the acoustic framework. Thanks to the added mass concept, we finally extend this analogy and show that the whole mathematical framework of special relativity is exactly mirrored by the linearised compressible potential flow theory.

The meaning of a concept in use is shaped by the situation, task, goals, and prior knowledge. For example, a request to make a poster "visually appealing for a five-year-old" might evoke bright colors and cartoon imagery for one collaborator, but less text, bold shapes, and visual simplicity for another. We call such task-relevant differences conceptual misalignment. We introduce Alignment Games, a framework for making these differences visible and repairable during human-AI interaction. Drawing on theories of situated conceptualization, we characterize task-specific conceptual frames in terms of relevant attributes, values, relations, constraints, and priorities. We then define alignment moves that intervene on the situation, the reasoning used to interpret it, or the resulting frame. Through examples from educational content generation, creative coding, and argumentative writing, we show how these moves can be composed into repair sequences and derive design principles for supporting task-sufficient conceptual alignment at runtime.