Papers

Invoices are commercial documents that contain various pieces of information, including the purchased items, time, and total money. Making the extraction of important information crucial. The stored information serves different purposes. Korean language is the native language of about 80 million people, playing an important role in not only South and North Korea but also in many other countries such as Vietnam, Philippine where a large number of Korean companies are located. In this context, to automatically extract proper information from the invoices with Korean language, we propose an efficient Optical Character Recognition (OCR) model in which a deep learning model is combined with some image preprocessing techniques. The proposed OCR model is assessed in a rich set of collected invoices showing that 87% F1-score can be achieved with negligible time processing.

Asthma deterioration forecasting must remain reli- able when patient populations, sensor ecosystems, and available modalities change across cohorts. Existing models commonly optimize within-cohort discrimination and may produce poorly calibrated probabilities after transfer. We present CALIBRA, a calibration-first multimodal temporal framework for short- horizon risk prediction with incomplete data. Dedicated recurrent encoders process environmental, pulmonary, symptom, medication, wearable, and context streams; a reliability-conditioned gate suppresses stale or absent modalities, while gradient-reversal training discourages avoidable cohort signatures. A shrinkage- based hierarchical logistic layer calibrates probabilities using a patient-disjoint target subset, and split conformal prediction provides abstention-capable prediction sets. To avoid fabricating clinical evidence, we evaluate the complete implementation on a documented three-cohort semi-synthetic benchmark with controlled distribution shift, informative missingness, and sealed target patients. Across five configured seeds, CALIBRA achieved mean target-test AUPRC 0.224 versus 0.240 for the strongest non-ablation comparator, TemporalTransformer; mean AUROC was 0.717, and Brier score was 0.098. Experiments additionally assess complete-modality failures, calibration, conformal coverage, decision curves, subgroup behavior, ablations, runtime, and parameter count. The results verify the method and reproducible pipeline under controlled shift, but do not establish clinical effectiveness. External validation on harmonized real asthma. Overall this artifact provides evidence for carefully governed real-cohort validation.

Just as Socrates recognized the limits of his own knowledge, Retrieval-Augmented Language Models (RALMs) should learn to abstain when the retrieved evidence cannot support a reliable response. Existing approaches largely rely on monolithic LLMs to handle heterogeneous retrieval failures in a single step, resulting in limited abstention performance and high computational costs. We instead decompose retrieval failures into two distinct states: (i) the unanswerable state, where the required evidence is absent, and (ii) the distracted state, where relevant evidence is mixed with conflicting, negated, or adversarial information. Based on this decomposition, we introduce a lightweight module (Sieve) that screens retrieved document sets for distracting evidence before invoking a costly LLM (Sage) for grounded generation and abstention. Evaluated across both general and high-stakes expert domains, our Sieve and Sage framework preemptively detects distracting noise, improving system accuracy by up to 69.4 percentage points and Macro-F1 by 55.2 percentage points compared to one-stage baselines. Furthermore, it achieves up to a 1.99x speedup, establishing a highly efficient and reliable abstention pipeline for RALM with abstention.

Sampling many responses and keeping one that passes a verifier lets large language models solve problems beyond their single-response ability, but this search must be paid again for every query, while many deployments answer with a single response. Post-training on verified responses can transfer the benefit of search into the model. With a fixed budget, selecting by correctness alone spends slots on problems the model already answers correctly, leaving fewer to correct its failures. To address this imbalance, we propose GapFT, which trains on the gap between Pass@K and Pass@1: problems that the source model fails with one response but solves within K samples. GapFT keeps the objective and training budget fixed and changes only which verified responses enter training; an exact decomposition splits the resulting Pass@1 change into corrected failures and regressions on problems the source model already solved. On LogiQA 2.0 and ReClor with three model families, GapFT is above budget-matched uniform rejection-sampling fine-tuning (RFT) in every setting, with a positive pooled effect, and on Llama-3.1-8B and Mistral-7B it recovers about two thirds to four fifths of the gain of fine-tuning on the entire verified pool with 11-34% of its problems. Further analyses reveal that the gain comes from failures that the first few search samples recover, while failures found only by deeper search displace replay and add no net gain, that filling the same budget with gold-labeled failures search cannot reach lowers accuracy, and that the gain is bounded by how many transferable failures search exposes.

Industrial predictive maintenance increasingly depends on learning from equipment spread across sites whose sensor data cannot easily be pooled. Federated averaging (FedAvg) solves this with a central aggregation server; gossip learning removes the server, but its behaviour for recurrent failure-detection models has not been measured under controlled conditions. We compare synchronous ring gossip with FedAvg, isolated local training and a centralized reference for a stacked LSTM that detects imminent failure on the NASA C-MAPSS turbofan benchmark. All methods share one open implementation, architecture, initialization, optimizer, data split and training budget, and the primary endpoint uses one terminal window per test engine to avoid the statistical dependence of overlapping windows. On FD001 (five seeds), gossip reached a terminal-window F1 of 89.6 +/- 1.3%, compared with 89.9 +/- 1.1% for FedAvg, 83.6 +/- 6.7% for local training and 93.5 +/- 2.1% for centralized training, while transmitting the same payload as FedAvg without a coordinator. Node models agreed closely but not exactly (1.8% pairwise decision disagreement versus 5.6% without communication). Across FD002-FD004, peer communication improved terminal-window F1 over local training by 13-28 points; gossip matched FedAvg on FD003 and FD004 but was 4.3 points lower on the multi-condition FD002 subset. Simulated message loss, node failure and server outage changed neither method appreciably, whereas larger rings degraded gossip faster. Ring gossip is therefore a practical serverless alternative when data heterogeneity is moderate, and faster-mixing topologies become important as heterogeneity grows.

Let A be a nonempty compact subset of the plane and let A (r) be its parallel set at distance r. We prove that the number of holes of A (r) the number of bounded components of its complement, which is its rst Betti number satises $\infty$ r 0 $\beta$1(A (r) ) dr $\le$ 4050 (diam A) 4 r -3 0 for every r0 > 0, the integrand vanishing for r $\ge$ diam A/ $\sqrt$ 3. The proof rests on Fu's theorem that the critical values of the distance function of a planar compact set form a set of vanishing half-dimensional Hausdor measure, on two lemmas of Rataj, Spodarev and Meschenmoser, and on a square-root summability estimate for the gaps of the critical-value set, of which we give a complete proof. We show by an explicit family of curves two combs facing each other that no analogous bound can hold at a xed radius: a connected curve of bounded length, diameter, oscillation count, parallel-set area and parallel-set perimeter can have arbitrarily many holes at one radius, so the integrated estimate cannot be replaced by a xed-radius bound in terms of these coarse geometric quantities. We also bound the hole count uniformly in the radius by the number of components of local maxima of the distance function, and record a bound on the boundary length of a parallel set by its area. The results supply the deterministic input for limit theorems on the persistent homology of the Wiener sausage.

Explainable Artificial Intelligence (XAI) solutions are essential for building trust in AI technologies and their integration in real manufacturing lines. However, most existing methods are tailored to technical experts, limiting their accessibility to diverse user groups such as blue-collar workers in manufacturing lines who use AI for quality control. In this work, we introduce an XAI interface for object detection in industrial manufacturing based on a fine-tuned vision-language model, designed to generate intuitive explanations for non-expert users. We benchmark existing vision-language models and demonstrate that out-of-the-box models often fall short in delivering clear, context-relevant explanations for non-expert users. To address this, we fine-tune a vision-language model and integrate it into our interface, enabling contextualized, accessible explanations for non-expert users. We demonstrate improvements in explanation clarity, instruction adherence, image groundedness, and contextual awareness over GPT 4o-mini on proprietary and public robotics dataset. This approach advances the accessibility and usability of AI explanations, making them more intuitive and applicable in manufacturing domain.

Video watermarking underpins copyright protection and provenance for generated media, yet almost every video is compressed by a codec before it is stored or shared. A codec discards precisely the perceptually redundant components that most watermarks rely on, so the payload is often lost even when the marked video looked flawless beforehand. Existing methods leave this path open, since they treat compression as one entry in a generic list of distortions, while a real codec is not differentiable and cannot enter gradient-based training. We present COVER, the first learned video watermark built around codec compression as its design target, which survives that compression by embedding the payload in the latent space of a frozen generative video autoencoder and recovering it by re-encoding the received video into that same latent space. To make codec robustness trainable, we build a differentiable codec surrogate bank that simulates the dominant degradation modes of practical compression, and we train the embedder and the latent decoder through three shared recovery paths under a fidelity objective that constrains the residual in the pixel and frequency domains. Across four codecs at 12 settings, COVER attains 93.72% average bit accuracy, ranks first on 11 of the 12, improves the strongest prior method by 2.68 points, and lifts the worst operating point from 68.90% to 73.72% while each marked video stays visually close to the source clip that produced it.

Inspired by the flight and song of birds, we propose an approach that enables cooperating agents to effectively identify obstacles in their environment by having a stationary source agent emit a frequency-modulated chirp while one or more listener agents observe the direct and reflected signals while in motion. We present a novel mapping approach that exploits the “frequency gap” between direct and reflected chirp signals to define candidate reflection ellipses, which are then fed into a 2D accumulation filter to identify locations with the highest density of potential reflections. We demonstrate this work first with simulation results derived from an efficient, bespoke audio simulator, examining the effect of obstacle count, sampling rate, and path curvature on the ability to accurately identify environmental obstacles for a one-listener scenario. We then examine different cooperative motion strategies for two-listener configurations. Finally, we validate the real-world viability of this approach through field experiments in an outdoor park setting to demonstrate the ability to identify frequency gaps using off-the-shelf hardware. Our results provide a foundation for a low-cost approach to environmental mapping in swarm robotic systems.

Bluetooth Low Energy (BLE) employs MAC address randomisation – via Resolvable Private Address (RPA) – to mitigate long-term device tracking on public advertising channels. Existing research has shown that advertising packets contain metadata and structural features that allow re-identifying a target device via manually crafted rules. In this work, we investigate the feasibility of automating the process of tracking BLE devices despite MAC randomisation by leveraging machine learning algorithms for the signature creation. Based on the actual Bluetooth traffic from target devices, we characterise the persistence of advertising-layer features across RPA changes and formulate device linkage as a supervised classification problem. Using simple decision tree classifiers as a proof-of-feasibility approach, we evaluate the distinguishability of target and non-target devices under varying address rotation patterns. Our results reinforce prior work demonstrating that advertising-layer metadata can enable device re-identification under MAC randomisation, to the point where such linkage can be automated using standard supervised learning techniques, without any specific knowledge of the technology.

We demonstrate that forking in monadically NIP theories satisfies a very strong form of triviality over arbitrary sets, generalizing Baldwin-Shelah in monadically stable theories and analogous to Shelah's result for finite satisfiability over models in monadically NIP theories. This implies in particular that strict non-forking independence is particularly well behaved, and that the forking relation induces a canonical semilinear partial order on singletons. We also give an example of an $\omega$-categorical monadically NIP theory so that no finite set is an extension base.

Kaplansky's zero-divisor conjecture asserts that the group ring of a torsion-free group over a field has no zero divisors. It holds for every group with the unique-product property, so a counterexample can only come from a torsion-free group without unique products. In lectures in 2021, Gardam announced that the torsion-free $\tilde{A}_2$ lattice $\Gamma = \langle a, b \mid a b a^2 b^{-1} a^2 b^{-2}, a b^3 a b^4 a^{-1} b \rangle$ does not have unique products and presented it as a new candidate group: it has property (T), and the known methods for proving the conjecture do not apply to it. To our knowledge, no proof of the announcement has been published. We give a proof checked by the Lean 4 kernel and stated against Mathlib's UniqueProds class. The witness is an explicit pair of finite subsets with $|A| = 32$ and $|B| = 28$ in which each of the 896 products coincides with another product. For 658 products the certificate is an identity in the free group; the other 238 certificates are explicit products of conjugated relators, 970 conjugates in all, checked by free reduction. A homomorphism onto $\mathbb{Z}/42$ shows that each pair $(u,v)$ differs from its partner $(u',v')$ as a pair of group elements, which is all the theorem requires. Together with a homomorphism onto the alternating group $A_4$ it also shows that the listed words are pairwise distinct, so the sets have exactly 32 and 28 elements. The witness and certificates come from an untrusted search program and are re-checked by Lean. The development uses only the axioms propext, Classical.choice and Quot.sound, with no sorry and no native_decide. The mathematical statement is Gardam's. To our knowledge this is the first verification in a proof assistant of a unique-product failure in a torsion-free group; torsion-freeness of $\Gamma$ is taken from Gardam and is not formalized here.

High-resolution orbital imagery offers a rich record of the Martian surface, but sparse geological labels limit supervised representation learning. We present MarsRecon, a geospatially aware pipeline for learning visual and multimodal representations from HiRISE observations of Olympus Mons. The pipeline calibrates NASA Planetary Data System products, extracts valid georeferenced patches, and trains a masked autoencoder on unlabeled imagery. Increasing input resolution and filtering invalid tokens reduced held-out reconstruction loss from 0.1751 to 0.1342 in the principal Stage A model series. We then freeze the visual encoder and align its features with observation text, coordinates, and local–global image context. The strongest current local-primary model achieves image-to-text recall@10 of 0.3787, text-to-image recall@10 of 0.9161, and local-to-global recall@10 of 0.4350 on the held-out test split. These results establish a working Mars-specific pretraining and retrieval pipeline; further crop-overlap controls and downstream geological evaluations are needed to assess the broader utility of its embeddings.

The brain fluid clearance system can be studied with intrathecal contrast-enhanced magnetic resonance imaging (MRI). By injecting a tracer into the spinal canal and performing MRI over 24–48\,h, one can measure how a tracer injected into the cerebrospinal fluid spreads through the brain. As this is an invasive procedure requiring several MRI scans, we ask how much of the later tracer distribution can be predicted from the pre-contrast scan and elapsed time alone. Using 104 training patients (878 MRI volumes), we compared a direct image-to-image regression model with a more complex generative model based on conditional flow matching, and tested both on 23 held-out patients and on 51 patients from a separate cohort with idiopathic normal pressure hydrocephalus (iNPH). Relative to simply copying the pre-contrast scan, the direct model reduced the mean absolute error (intensities normalized to $[0,1]$) from 0.089 to 0.036, about 60%, in held-out patients and from 0.075 to 0.056, about 25%, in the iNPH cohort. The reduction held in every sampled elapsed-time interval and brain compartment, and errors remained largest at 20–30\,h and in the ventricles, where enhancement is strongest. Its predicted change maps agreed with the observed tracer enhancement rather than only reproducing anatomy. The generative model was less accurate in both cohorts. A substantial part of the later tracer distribution is therefore predictable from the patient's brain anatomy as seen on pre-contrast MRI, and the added model complexity gave no benefit at this data scale. Whether this accuracy suffices for a given clinical use remains to be established; with prospective validation, such forecasts could help simplify the MRI protocol used to assess brain clearance.

Quantitative analysis of experimental powder X-ray diffraction data remains challenging when evaluating complex multiphase materials and noisy measurements. We introduce a hybrid quantum neural network framework designed to extract quantitative parameters, such as phase weight fractions and scale factors, directly from one-dimensional powder diffraction patterns without iterative refinement. The model combines noise-aware classical simulator pre-training with fast downstream fine-tuning on quantum processing unit features, ensuring stability against hardware decoherence. We demonstrate the practical utility of this approach by deploying the trained network onto an IBM quantum computer to analyse experimental X-ray diffraction computed tomography datasets from a three-phase solid oxide fuel cell containing ca. 10,000 patterns and a four-phase lithium-ion battery containing ca. 20,000 patterns). The network successfully reconstructs quantitative spatial phase maps in strong agreement with classical Rietveld refinement, paving the way for using quantum computing hardware to analyse real-world materials characterisation data.

Introduction. Open collaborative knowledge graphs such as Wikidata increasingly ground agentic artificial intelligence, information retrieval, and language modeling systems, making systematic auditing of their demographic representation and overall equity a research imperative. Methods. Herein, we present an open-source auditing platform that ingests over 10 million statement bindings representing over 6 million humans on Wikidata via QLever, and evaluates representation of gender, sexual orientation, geography, birthplace urbanicity, ethnicity, multilingual coverage of labels, descriptions, and aliases, occupation, and select intersectional pairs of these entities. It does so by making use of Chi-square goodness-of-fit tests, 95% Wilson-score confidence intervals, and disparity ratios, in light of Rubin's missingness taxonomy. Results. Women accounted for 28.71% (CI +/-0.04) of all humans in Wikidata with a stated gender. 38.26% of humans had a citizenship statement, with Western Europe and North America (WENA) representing approximately 53% of such statements. Among 1.8 million birthplaces that could be classified, rural birthplaces were observed in 2.48% of cases (in comparison to 27.4% global baseline). Fewer than 1.2% of entities carried an ethnicity statement, and non-English Wikidata descriptions covered 18.2% of items. Discussion. Our findings reveal significant missingness across the evaluated axes. Ethnicity and sexual orientation were the most critically under-documented (missing statements) while rural birthplaces and non-WENA citizenship were the most underrepresented.

Methods for water demand forecasting and leak detection are based on public datasets, and for water those are scarce, short, or released only after an undocumented cleaning process, hiding defects of the deployment they came from. We present Tethys: 91 months of hourly water consumption data from 24 buildings of a municipal water network, published with a quantitative account of its quality, rather than in place of one. Raw availability is 59.4%, while loss is not random: 4 fleet-wide outages totalling 595 days interrupt the entire estate at once. We show that the aggregation producing the released files silently introduced 205,200 impossible decreases in a cumulative index, and that correcting it is a one-line change. Because the meters are cumulative, the readings bracketing a short gap fix the volume that passed through it, so 75.9% of hours rest on a measurement, while 24.1% are reported as unknown. We release the dataset, its per-hour provenance, and the pipeline that produces it.

Every smooth closed Riemannian surface of genus zero is isometric to the unit sphere carrying a conformal metric, the round metric multiplied by a positive factor. This paper solves the Poisson equation and the eigenvalue problem of the Laplace–Beltrami operator of such a metric with smooth spherical splines, taking the conformal factor as the only description of the geometry. Because the Dirichlet energy is conformally invariant in two dimensions, the stiffness matrix is the round-sphere matrix for every conformal metric, and the factor enters only the load vector, the mean-value constraint and the weighted mass matrix. A surface embedded in space with a conformal parametrization, a surface of revolution, and a surface given only as a triangle mesh are instances of the same problem that differ in how the factor is obtained; the first two are treated here, and the factor is checked against an area identity before any equation is discretized. The unknown is a spline on a fixed spherical triangulation in broken Bernstein–Bézier form, with $C^r$ smoothness imposed algebraically through edge functionals and the conforming space realized by a null-space matrix computed once and reused for every metric. Each step of the computation is stated with the formulas it needs. Experiments on the round metric, a prolate spheroid, a dumbbell of revolution, an Evans–Fung red-blood-cell profile and a prescribed factor with no embedding show the rates $d+1$ in $L^2$ and $d$ in energy for spline degree $d$, the rate $2d$ for eigenvalues with reproduction of the even-order eigenvalues of the round metric up to degree $d$ to roundoff, and, against parametric surface finite elements on the same triangulations, smaller errors at comparable numbers of unknowns for degrees four and six.

We introduce Bernstein constraint complexes, a representation of finite element differential complexes in which every triangle of a triangulation keeps its own Bernstein–Bézier coefficients and global continuity is imposed only through smoothness functionals attached to edges and vertices. Continuity of scalar splines, tangential continuity of spline vector fields, normal continuity of spline vector fields, and higher smoothness differ only in the edge functionals. No global basis is constructed. The structural result is a coefficient-level commuting identity: on every edge the smoothness functionals of a Bernstein derivative are explicit combinations of the smoothness functionals of its argument, through univariate Bernstein difference matrices. This gives a local proof that the kernels of the smoothness matrices form a subcomplex, for the de Rham traces and for componentwise $C^r$ spline profiles, and a computable row-space test for candidate smooth profiles. We realize the lowest-order Powell–Sabin exact sequence in this form and confirm its exactness by ranks. The constrained Galerkin problem is solved in the broken coefficients with exact smoothness equations; we prove equivalence with the conforming method, treat semidefinite element operators through a pseudoinverse interface system, and give a singular pencil whose finite spectrum is the constrained Maxwell spectrum. Under the covariant and contravariant surface Piola maps the reference functionals of the de Rham sequence transfer unchanged to conforming curved triangulations. Experiments compare the null-space method, the augmented Lagrangian iteration of Awanou, Lai and Wenston, and direct solution of the saddle system on the same problems, and cover planar source and eigenvalue problems, a $C^1$ biharmonic solve, the exact sphere, and an embedded hyperboloid.

Mobile river monitoring robots must interpret obstacles and water boundaries that geographic waypoints alone cannot describe. On resource constrained platforms, converting imperfect visual predictions into timely and inspectable guidance is a distinct challenge. An object label or steering command does not explain which evidence supports a decision or when that evidence is unreliable. We present PAANI, an on-device perception to guidance architecture that combines a project trained YOLO11n detector and a custom MobileNetV3 Small semantic segmenter with timestamp aligned evidence fusion on Arduino UNO Q. Bounded tracking supplies object persistence, while an explicit corridor policy combines surface labels, accepted detections, urgency and mask uncertainty. Each final advisory exposes its contributing evidence and policy reasons. ROS 2 interfaces connect the local AI pipeline to a separate Gazebo vessel, localization and control testbed. Training uses 10,000 WaterScenes images for four-class detection and 1,127 MaSTr1325 images for segmentation, including 198 segmentation validation images. The selected FP32 ONNX models occupy 14.817 MB. Detector checkpoint test mAP at 0.5 IoU is 0.7388, while the separately evaluated rectangular ONNX export achieves validation mAP at 0.5 IoU of 0.7367. Segmentation ONNX validation mIoU is 0.9750. A five-minute UNO Q recording produced median and 95th percentile pipeline latencies of 467.8 ms and 580.3 ms at a configured 0.5 Hz cadence. The evaluation also identifies black input misclassification and a sampling rate mismatch that prevents the diagnostic apparent motion estimator from collecting sufficient evidence. These results support an inspectable and reusable edge robotics foundation while clearly distinguishing model accuracy and on-board execution from validated on-water collision avoidance.

Open-vocabulary 3D Scene Graphs (3DSGs) ground each object node in a vision-language embedding, yet they record every entry as equally certain, so a robot querying the map cannot tell which of its entries are unreliable. Estimators of semantic uncertainty could supply that distinction, but they require repeated sampling of a model, training, or held-out labels, none of which are available to a deployed system at query time. We present a framework that exploits the detector confidence and the embeddings a 3DSG already stores, converts them into a probability that an entry is correct, and propagates that probability through the containment hierarchy into a belief that a room contains a queried class. Four signals, each paired with the object-level error it indicates, are converted to probabilities at the logit scale learned by the vision-language model and combined in closed form with no additional perception or training. Objects sharing a detector and a vocabulary fail together, so the framework aggregates them in the fully correlated limit, where an aggregation under independence would treat one repeated error as repeated evidence. Evaluated on HM3DSem against a state-of-the-art 3DSG system, the framework improves object retrieval and lowers the error of the room-level assertions of the graph it reads.

Transient pressure diffusion in heterogeneous porous media becomes difficult to resolve efficiently when permeability is discontinuous and spans several orders of magnitude. In this paper, we develop a domain-decomposed random-feature evolutional deep neural network (DD RF-EDNN) that separates spatial approximation from temporal evolution. Permeability-informed random features are compressed into an orthonormal pressure space, and a conservative finite-volume operator is projected onto this space so that only the reduced coordinates are advanced in time. This formulation retains the dissipative structure of the discrete flow problem without repeated neural-network optimization. The analysis quantifies dictionary and singular-value truncation errors, proves contractivity under an energy-consistent dissipativity correction, and derives a conditional grid-level estimate for homogeneous dynamics and time-independent data admitting a steady lifting. Numerical experiments on low-permeability inclusions, high-conductivity blocks and channels, and three-dimensional grids assess the accuracy of the method across distinct coefficient configurations. The method reaches a final-time relative $L^2$ error of $9.54\times10^{-4}$ in the principal benchmark and maintains errors on the order of $10^{-2}$ at a permeability contrast of $10^3$. Numerical diagnostics further show that the observed error balance and temporal convergence are consistent with the analysis. These results support DD RF-EDNN as a structure-preserving and interpretable reduced formulation for transient simulations on prescribed heterogeneous media.

We propose a Verification Reward Model (VRM) framework for training language models to create and repair chip verification artifacts. The central object is a versioned verification contract that binds requirements, permissible stimulus, observation boundaries, reference behavior, evaluation budgets, and acceptance criteria. Compiler, simulator, formal, mutation, coverage, and expert-review evidence are converted into auditable records. Deterministic acceptance checks remain outside the learned model. A learned outcome model predicts expensive future evidence from the specification, the generated artifact, and explicitly masked partial evidence; a semantic critic identifies evidence-supported weaknesses; a deterministic reward composer translates the resulting quality vector into task-conditioned training rewards. Extensions include paired clean/fault interventions, marginal fault-discovery rewards, uncertainty-aware evaluation scheduling, and a quarantined cross-domain learning loop. Evaluation emphasizes independently validated fault detection, false alarms, generalization across design families, and total cost to reach a specified quality level. We describe a first experiment on a small, open-tool-compatible benchmark with lightweight models, with full UVM capability admitted through feature-specific qualification. This is a position paper: we specify the framework and the experiments that would test it, and we report no training, EDA, or silicon results.

We explore the use of hybrid quantum-classical generative adversarial networks for synthesising CAD-ready vector geometries. Unlike prior work that operates in rasterised or single-Bézier domains, we introduce HMB-GAN (Hybrid Multi-Bézier GAN), an end-to-end differentiable generative framework that constructs closed shapes through stitched multi-segment Bézier representations with geometric continuity enforced by construction. We compare a quantum-enhanced generator with a classical generator within this architecture and evaluate them across point cloud distribution metrics and geometric shape statistics. Results show that despite faster convergence, a reduction in model parameter count, and slightly improved performance on point cloud metrics, the quantum generator suffers from excessive simulator overhead and thus classically-simulated evaluation suffers from hardware constraints. These results demonstrate the feasibility of modelling structured geometries through hybrid quantum architectures whilst highlighting contemporary hardware limitations.

Large Language Model (LLM) agents are increasingly being explored for chip design, but most existing approaches operate directly at RTL. We ask whether agents can design better chips by leveraging higher-level abstractions. We compare Direct RTL Design, Agent-based HLS Design, Post-Compiler HLS Refinement, and Post-HLS RTL Refinement, and combine Agent-based HLS Design with Post-HLS RTL Refinement as Agent-based HLS with RTL Refinement (AHRR). We use FPGAs as a practical, easy-to-deploy platform for end-to-end evaluation, but note that the design-flow tradeoffs we study are largely independent of the target technology. Across a diverse 11-tasks benchmark suite, AHRR achieves a 2.6$\times$ geometric-mean speedup over Direct RTL Design across our benchmark suite. Case studies show that HLS distills design knowledge into abstractions that agents can leverage, while RTL refinement recovers lower-level optimization opportunities. Together, these results make AHRR a promising workflow for agentic chip design. The code and evaluation artifacts are available at https://github.com/ZijD/AHRR.

A repair favored under an isolated input fault can be inferior when deployed modules share the faulty information. We introduce an information-interface audit for world models, distinguishing fidelity gaps, where exact inputs become estimates, from availability gaps, where inputs are missing. Fixed-weight interventions measure prediction error, input dependence, and paired closed-loop benefit, including dependencies introduced by reconstruction. In simulated quadrotor model predictive control, coupled, opposite-sign 10% mass/thrust calibration errors reduce a physics-anchored model's success from 69% to 8%; uncertainty training restores 65%. Wind reconstruction recovers control benefit but inherits calibration dependence. For a positive calibration offset, reconstruction-only corruption favors uncertainty-trained reconstruction, whereas shared corruption favors the baseline. Acceleration diagnostics reveal compensation between reconstruction bias and nominal-model error, also observed with a disturbance observer. Repair selection therefore depends on the information paths used in deployment.

Good tutoring adapts to the individual: it tracks what a learner knows, notices why they go wrong, and asks the next question that will help most. Most deployed tutoring tools instead serve fixed item banks and treat a wrong answer as a single bit of signal. We present CoLearn, an interactive, agentic tutor that supports an iterative tutoring loop: the learner practises, and the system builds an evidence-grounded memory of the learner's mastery and misconceptions. This memory is updated as evidence accumulates and is used to generate the next personalised question. CoLearn has three components: (i) a persistent learner-state memory that updates per-topic mastery with a soft-evidence variant of Bayesian Knowledge Tracing, where a large language model acts as a continuous observation function; (ii) adaptive question generation that targets the learner's weakest topic and recurring misconceptions; and (iii) an evidence view that makes personalisation visible and testable through live progress visualisation and blind A/B comparison. In blind A/B evaluation, questions conditioned on this memory are preferred over non-personalised ones 68-69% of the time, and in persona simulations with hidden ground-truth mastery the agent's belief converges toward the learner's true mastery.

In quantum networks, nodes are connected via sharing of Einstein-Podolsky-Rosen (EPR) pairs, ideally with high fidelity and high rate. However, the fidelity of EPR pairs degrades due to imperfect generation and decoherence errors. Entanglement Distillation is a method that increases the fidelity but operates probabilistically and may destroy all involved EPR pairs upon failure. This failure reduces available EPR pairs for application use, thereby decreasing the service rate. Quantum Error Correction (QEC) is another mechanism to protect EPR pairs against error by forming what we term as Coding-Enhanced Memory (CEM). While effective, CEM requires extra time and resources to form the code, which also reduces the service rate. Existing methods often use static combinations of distillation and CEM, ignoring demand variations. This results in a low service rate without significant fidelity gain. Limited resources together with this rate-fidelity tradeoff make it essential to schedule when to run distillation, form CEM, or serve requests. We propose ReFINE, a demand-aware preemptive scheduler that based on application requirements either serves an available EPR pair immediately or preserves it in CEM. This selective use of CEM, only when needed, enables a better balance for rate-fidelity tradeoff than always using CEM. Between request arrivals, ReFINE either schedules distilling EPR pairs or forming CEM to protect distilled pairs, following one of the three priority policies: ReFINE-D (Distillation-First) first generates EPR pairs for distillation and then forms the CEM, prioritizing service rate. ReFINE-M (Memory-First) first forms the CEM, then generates the EPR pairs for distillation, prioritizing fidelity. ReFINE-C (Concurrent) performs both distillation and CEM formation concurrently, balancing between fidelity and service rate.

Molecular solubility directly affects key aspects of molecular development such as reaction feasibility, formulation performance, separation efficiency, and solvent selection. However, experimental measurement across solutes, solvents, and temperatures remains costly and sparsely sampled. Existing computational models often rely on fixed-solvent assumptions, deterministic formulations, or simplified representations of solute-solvent interactions, limiting their ability to capture complex molecular interactions, continuous temperature effects, and experimental uncertainty. Here, we introduce EnSol, an environment-aware probabilistic framework for molecular solubility prediction. EnSol represents the solute and solvent as molecular graphs and learns separate representations for each before bringing them together through cross-attention to capture solute-solvent interactions. Temperature is incorporated directly into the solvent environment through feature-wise modulation, and a mixture density network predicts full solubility distributions to capture both temperature-dependent behavior and experimental uncertainty. On the independent SolProp and Leeds benchmark datasets, EnSol achieved Spearman correlations of 0.876 and 0.601, respectively, outperforming state-of-the-art solubility prediction models across both benchmarks. Beyond computational benchmarking, experimental validation across chemically diverse solute-solvent pairs showed that EnSol maintained strong predictive performance and supported reliable solvent ranking, achieving a Spearman correlation of 0.715. These results show that EnSol can support reliable solubility prediction and solvent selection across diverse chemical systems while accounting for predictive uncertainty.

Before patients can use AI-assisted psychiatric intake systems, health systems need practical ways to routinely evaluate these tools against their clinical standards for quality assurance. Because clinicians may use different intake styles, evaluation for this task must (1) support comparison across interviewing approaches, (2) minimize clinician burden, and (3) measure clinically relevant performance for health systems deploying these technologies. We present a clinician-grounded evaluation platform built around a memory-augmented patient simulator for open-ended AI interviewing, InterviewPlayground. We created interactive patients using InterviewPlayground with our expert-authored vignettes, constructed a simulated intake platform for the interviews, and designed evaluation modalities relevant to intake. In a pilot of 6 clinicians in a 25-minute assessment compared to a GPT-based LLM intake interviewer, the LLM recovered more of the clinically relevant items embedded in the patient vignettes (88.0% vs. 38.9%), but made more clinical inferences not based on the interview (56.8% vs. 27.8%), and characterized identified safety concerns less often (33.3% vs. 66.7%), setting the stage for deployed quality assurance for this task.