Peer Revenue Publish or profit

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Saturday 5 September

Artificial Intelligence

More Than Mimicking Reviewers: Evaluating LLMs for Pre-Submission Peer Review

Peer-review feedback often arrives too late for authors to make meaningful revisions. We study an author-facing LLM system that moves part of this stress test before submission: it generates a broad pool of atomic concerns and compresses them into a short report. We evaluate agreement with historical reviews and, separately, the possible validity of concerns they omit. From 10,000 ICLR 2026 submissions, we use 3,398 manuscripts with accessible versions that predate review. On a ten-paper diagnostic, independent sampling covers 44.9% of historical issues; deduplication and refill reaches 78.7% strict and 84.9% seriousness-weighted coverage, at 3.6$\times$ more requests and 5.2$\times$ more tokens. A hidden Top-32 Oracle preserves the full 79.3% weighted coverage of a 256-candidate pool, but paper-only selectors retain only 40–44%. LLM review therefore provides broad coverage with a large candidate pool but compresses poorly; ablations identify representative selection and matcher sensitivity as the main sources of this gap.
Read the full abstract

Combinatorics

Minor-Order Exponent Profiles of Oscillatory Matrices

For an $n\times n$ oscillatory matrix $A$, let $e_k(A)$ be the least positive integer $m$ for which every minor of order $k$ of $A^m$ is positive. We determine the profile $(e_1(A),\ldots,e_n(A))$. For every nonsingular totally nonnegative matrix, the column index sets of positive entries in each compound row form an interval in the componentwise order. The endpoint maps are order-preserving and compose under multiplication. This yields a two-corner criterion for positivity of all minors of a fixed order. Consequently, $e_k(A)$ is the larger of the first positivity times of two remote corner minors. We prove the sharp inequalities $e_k(A)\leq\max{k,n-k}$ for $1\leq k<n$ and $|e_{k+1}(A)-e_k(A)|\leq1$ for $1\leq k\leq n-2$. For $D=\operatorname{diag}(1,-1,1,-1,\ldots)$, the matrix $DA^{-1}D$ is oscillatory and satisfies $e_k(DA^{-1}D)=e_{n-k}(A)$ for $1\leq k<n$; the determinant exponent remains one. The ordered positions of the positive factors in an adjacent bidiagonal factorization determine the profile independently of their values. Each one-sided factor sequence can be represented by a single permutation, giving a finite characterization of all profiles and an integer unimodular realization of each. We also obtain a compatibility condition across minor orders: if $3\leq k\leq n-3$ and $e_k(A)\leq2$, then $e_j(A)\leq2+\lceil |j-k|/2\rceil$ for $2\leq j\leq n-2$. In particular, $e_k(A)\leq2$ implies $e_{k+2}(A)\leq3$ for $3\leq k\leq n-4$, and both the constant and this range are sharp.
Read the full abstract

Systems and Control

Bidirectional Wireless Communication for Weakly Coupled Implantable Brain-Computer Interfaces

Implantable brain-computer interfaces (BCIs) promise transformative societal impact, from restoring lost motor, sensory, and speech function in patients with paralysis, stroke, sclerosis, and sensory deficits to serving as a high-bandwidth conduit between human cognition and machine intelligence. Realizing these visions requires moving from laboratory prototypes to chronic clinical systems that support thousands to millions of electrodes, several millimeters to centimeters deep beneath the brain surface, under strict heating and size limits. Almost every clinically relevant implant therefore operates in a weakly coupled regime, with coupling coefficients of $10^{-3}$ to $10^{-1}$ across centimetres of lossy tissue. In this regime, the wireless channel sets the limits of power-transfer efficiency, communication bandwidth, and energy per bit. We review the recent progress of bidirectional wireless links across inductive, mid-field, RF, ultrasonic, magnetoelectric, optical, UWB, and electro-quasistatic modalities, and benchmark them against the clinical axes of depth, size, and data rate. Within an approximately 10 mW communication budget set by the approximately $1^\circ$C tissue-heating ceiling, narrowband high-Q links are well suited for power transfer and low-speed data, but at 1-10 nJ/b cannot reach the greater than 10 Mbps to tens of Gbps uplinks that thousand- to million-channel interfaces demand, even with aggressive on-implant compression. This calls for sub-10 pJ/b and ultimately sub-1 pJ/b wireless links, where wideband techniques, such as ultra-wideband (UWB) and brain-channel communication (BCC), are suitable. We close with a quantitative framework for analyzing such wireless links and a co-design roadmap across electromagnetics, circuits, packaging, security, and regulation, toward secure, networked, million-channel brain interfaces.
Read the full abstract

All authors Shreyas Sen, Baibhab Chatterjee, Gourab Barik, Anirudh Roy

Optimization and Control

Flatness of the Generic Under-Actuated 6-Degrees-of-Freedom Unmanned Surface Vessel

The flatness of the under-actuated unmanned surface vessel with circular hull-shape (“the hovercraft") has long been known. We resolve the long-standing open problem: whether the generic under-actuated 6-degrees-of-freedom (dof) unmanned surface vessel (gUSV) admits a feedback linearization. Lie symmetries of the gUSV are used to construct a dynamic feedback linearization, explicit flat outputs and flat parametrization, for all hydrodynamic parameter values. In this generic case the dynamic compensator has order 6, in contrast to the hovercraft vessel whose dynamic compensator has order 4.
Read the full abstract

Statistics Applications

A Network-Structured Bayesian Hierarchical Model for Sparse Mutation-Drug Response Associations: Application to Cancer Pharmacogenomics

We develop a network-structured Bayesian hierarchical model for sparse association mapping between genomic alterations and quantitative treatment-response phenotypes. The framework combines a Gaussian Markov random field prior that borrows strength across pathway-connected genes, a global-local horseshoe prior inducing sparsity, and a conjugate Gibbs sampler requiring no Metropolis-Hastings steps. Though broadly applicable to high-dimensional settings with known predictor networks, we validate it using cancer cell-line drug-sensitivity data. Applied to GDSC2 ($N=951$ cell lines, $G=219$ driver genes, $D=295$ drugs), the model identifies 126 gene-drug associations (0.195% of 64{,}605 pairs), concentrated in EZH2 (45 drugs, all sensitivity-direction, mean effect $-0.911$ $\ln$IC50) and KMT2D (36 drugs, all sensitivity-direction, mean effect $-0.496$ $\ln$IC50). These markers show external support in an independent PRISM screen (1{,}518 compounds), with KMT2D achieving complete directional replication (36/36) and EZH2 partial replication (8/12). Five-fold cross-validated predictive log-likelihood confirms each prior layer's value: the full model outperforms the no-network ablation by $+3{,}109$ log-units per fold and the no-horseshoe ablation by $+14{,}039$ log-units, consistently across folds. Simulations under three scenarios show the full model achieves the highest precision and lowest false-discovery rate throughout, while the network prior improves sensitivity recovery under network-structured signal. A tissue-stratified extension identifies coherent subgroup refinements, including lung-specific EGFR-inhibitor sensitivity and skin-specific BRAF-Dabrafenib sensitivity. These results show the framework identifies sparse, interpretable, externally supported drug-sensitivity markers while enabling principled investigation of tissue-specific departures from shared effects.
Read the full abstract

Robotics

Closed-Loop Evaluation of Bird's-Eye-View Maps from Cross-View Transformers as Inputs to Behavior-Cloning Policies

In autonomous driving, Bird's-Eye View (BEV) representations provide a structured, top-down abstraction of the vehicle's surroundings and have become a key input modality for Behavioral Cloning (BC) policies. While ground-truth BEV maps are readily available in simulation, real-world deployment requires replacing them with camera-predicted counterparts - a substitution that introduces perceptual errors whose downstream impact on closed-loop driving performance is not well understood. In this work, we investigate the use of Cross-View Transformer (CVT)-predicted BEV maps as direct policy inputs for a BC agent in the CARLA simulator. We propose a six-channel BEV representation covering road surface, planned route, lane boundaries, vehicles, pedestrians, and traffic lights, and introduce a Kernel Density Estimation (KDE) weighting scheme that rebalances the segmentation loss towards underrepresented driving maneuvers such as curves and intersections. Closed-loop evaluation across two CARLA towns shows that the KDE-weighted model is the only predicted-BEV agent to complete a full episode without infractions, despite not achieving the highest aggregate IoU. This discrepancy reveals that global segmentation metrics are poor proxies for driving performance: what determines navigation success is prediction quality at geometrically critical locations, and the route channel emerges as the primary bottleneck for reliable agent navigation under predicted BEV inputs.
Read the full abstract

Artificial Intelligence

Distilling Vision-Language Models for On-Device Fire Understanding

Vision-language models (VLMs) offer a promising alternative to conventional fire detection systems by reasoning about the semantic context of a scene and thus reducing false alarms, yet their large model size makes deployment on embedded fire sensors impractical. In this paper, we study how domain-specialized VLMs can be compressed for fully on-device deployment without losing the safety-critical behavior required for fire detection. We develop a teacher-student knowledge distillation framework in which large VLMs fine-tuned for fire understanding can be distilled into lightweight students. Experiments across multiple VLM families and model scales show that compact students preserve most of their teachers' fire-understanding capability. We further deploy the distilled models on our commercial Detectium fire detection sensor and jointly evaluate reasoning accuracy, latency, and memory usage. The results show that compression and deployment affect not only accuracy but also model failure modes, with Qwen2.5-0.5B providing the strongest overall deployment trade-off. Our findings provide broader guidance for deploying domain-specialized VLMs in resource-constrained, safety-critical settings.
Read the full abstract

Representation Theory

Thick subcategories over weakly symmetric algebras with radical cube zero

In this paper, we study thick subcategories of the (stable) module categories of finite-dimensional weakly symmetric algebras with radical cube zero. When the spectral radius of the Ext matrix is two, thick subcategories containing the algebra, equivalently thick subcategories of the stable category, are classified by Serre subcategories of a certain abelian category; when it is not two, only the trivial thick subcategories occur. As an application, this yields, to the best of our knowledge, the first examples of commutative Noetherian Gorenstein dominant local rings that are not complete intersections.
Read the full abstract

Combinatorics

An infinite walk in $\mathbb{N}^{16}$, using only unit steps, with no three collinear points

Is it possible to walk to infinity, avoiding three collinear points, using as steps only the standard unit basis vectors $(0,0,\ldots, 0, 1, 0, \ldots, 0)$? We prove this is possible in $16$ dimensions.

Software Engineering

Diffs vs. Whole Files: An Empirical Comparison of Iterative Edit-Based and Direct Generation for Flutter/Dart Code Models

Large language models used for code editing can be trained and deployed in at least two output regimes: direct generation, where the model emits the entire modified file in one shot, and iterative diff-based generation ("steps"), where the model emits a sequence of localized search/replace edits applied one at a time until it signals completion or a step budget is exhausted. The diff-based regime is attractive because it mirrors how developers edit code and should require far fewer generated tokens per turn. We train two code models - a 100M-parameter model trained from scratch (Rainbow-Pony-100M) and a fine-tuned Qwen2.5-Coder-0.5B - in both regimes on a shared Flutter/Dart dataset, and evaluate all four resulting models on a held-out set of approx 1,790 tasks per model. Direct generation substantially outperforms diff-based generation on every metric we measure - compilation/static-analysis pass rate, bits-per-byte, character-level similarity to the reference, and blinded LLM-judge ratings of goal fulfillment, correctness, and code quality - and the gap persists after controlling for task difficulty via a matched-ID comparison and when restricting to code that compiles on both sides. We then identify a single, architecture-independent mechanism behind the conditions where diff-based generation does win: it is competitive on short, spatially localized edits, and its category-level wins concentrate in exactly the two task categories - refactoring and error-handling/edge-case fixes - with the lowest mean edit-step count in our dataset. We term this task locality and discuss its implications for when an edit-based training regime is and is not the right choice for a code-editing model.
Read the full abstract

Machine Learning

Nonlinear elliptic homogenization with the parametric Deep Ritz method

Elliptic homogenization is used to determine coarse-grained properties of materials with features on small scales. When these small scale features have rapid, periodic fluctuations, the solution field corresponding to a homogenized constitutive relation closely resembles the true solution based on the heterogeneous material. This homogenized behavior of the material is computed from a cell problem, where a cell is defined to be one period of the fluctuating material. In the context of linear elliptic partial differential equations, the homogenized constitutive relation is defined simply by a constant coefficient tensor, but for nonlinear problems, the homogenized response depends on the macroscopic state and/or its gradient, thus requiring solutions to parametric cell problems. When computing a numerical solution with the homogenized constitutive relation, it is useful to have a differentiable representation of the solution to the cell problem, as derivatives of the homogenized constitutive relation are required in Newton iterations for the macroscopic state field. In this work, we use the Deep Ritz method to solve the parametric cell problems that arise from nonlinear homogenization. First, we exploit the variational structure of the cell problem, then we discretize the dependence of the cell response on both space and the macroscopic state with a neural network. Enforcing boundary conditions on the cell response strongly, we next use the parametric Deep Ritz method to simultaneously solve the cell problem over a range of macroscopic states. We show that this method is accurate, efficient, and offers a continuous and differentiable representation of the cell response over the macroscopic state and gradient. We then show that our parametric representation of the cell response significantly expedites macroscale solutions when compared to a traditional $\text{FE}^2$ scheme.
Read the full abstract

Friday 4 September

Computation and Language

NYUAD at ImageEval Shared Tasks: A Multimodal LLM for Audio-Visual Question Answering and Hallucination Detection in English and Modern Standard Arabic

The increasing adoption of multimodal large language models (MLLMs) has enabled more natural interaction with visual and spoken information, creating new opportunities for intelligent audio-visual question answering (AVQA) and hallucination detection systems. AVQA extends traditional visual question answering by requiring models to jointly understand an input image and a spoken question to generate accurate, context-aware responses. Despite recent advances, AVQA and hallucination detection tasks remain challenging for modern standard Arabic (MSA) due to the limited availability of multimodal datasets and the need to reason across Arabic speech and text modalities alongside visual inputs. In this paper, we present an MLLM-based framework for AVQA that takes an image and a spoken question as input and generates answers. In addition, we utilized this framework to detect hallucination in text-image pairs. The proposed approach is evaluated on the ImageEval 2026 Task (Ayn-VQA) benchmark, which focuses on culturally grounded Arabic AVQA. Experimental results demonstrate that the proposed framework effectively understands spoken queries, grounds their responses in the visual content, and produces accurate answers. The results underscore the capacity and limitations of different LLM architectures for both English and MSA and under various conditions.
Read the full abstract

Human-Computer Interaction

"I Wrote, I Paused, I Rewrote": A Mixed-Methods Study of Learner Perceptions of Process-Aware LLM Feedback

Large language models increasingly provide formative writing feedback, but most systems evaluate only the submitted text. Process-aware systems can additionally incorporate intermediate drafts and interaction records, yet access to these traces does not establish whether learners perceive the resulting feedback as useful, fair, or aligned with their own revision experience. We conducted a convergent mixed-method post-task evaluation with 43 undergraduate English-language learners. Participants completed a timed essay, received LLM-generated feedback based on the final essay, intermediate snapshots, and keystroke-derived records, and then answered six five-point Likert items and four open-ended questions. Ratings were concentrated above the neutral midpoint across all six items. Between 74.4% and 95.3% of participants selected ratings of 4 or 5, and all six one-sided exact sign tests remained significant after Holm correction. Perceived correspondence between the revision analysis and the participant's reported thought process was positively associated with overall perceived usefulness, Spearman's $\rho_s=.327$, 95% bootstrap CI $[.009,.609]$, $p=.032$, $n=43$. Qualitative responses were analysed through an LLM-assisted codebook development procedure followed by independent human coding of 14 binary codes. Participants frequently described the product feedback and revision rationale as accurate, while a smaller set reported misinterpretation of surface corrections, time constraints, or rhetorical intention. The findings distinguish perceived detection of revision activity from perceived interpretation of why revision occurred.
Read the full abstract

Statistics Applications

Causal sensitivity analyses for indirect and spillover mediation effects in cluster-randomized trials: Application to a pragmatic trial on chronic pain management

Causal mediation in cluster-randomized trials can clarify whether a cluster-level intervention changes patient outcomes through patient- and peer-level processes, but identification relies on untestable cross-world assumptions. We develop a two-part sensitivity analysis for violations of two distinct cross-world independence assumptions. The first component adapts bridge score sensitivity analysis to assess residual mediator-outcome confounding with clustered mediator exposures; the second uses a Gaussian copula to assess cross-world dependence between an individual's treated mediator and summaries of control-arm peer mediators. The resulting framework parameterizes unidentifiable dependence structures using interpretable correlation parameters and combines them with a flexible Bayesian nonparametric model to characterize the plausible range of the target estimands. A nested common-atoms enriched Dirichlet process models the observed data and supports posterior-predictive standardization. We apply the framework to the Pain Program for Active Coping and Training trial, a primary care-embedded cognitive behavioral therapy intervention for adults with chronic pain receiving long-term opioid therapy. The intervention had a negative total effect on twelve-month pain-related disability. The natural indirect and direct effect decomposition was sensitive to plausible directional residual confounding, whereas the spillover and individual mediation effect decomposition changed little across the copula sensitivity analyses and attributed most of the negative indirect signal to the individual's mediator. Jointly relaxing both assumptions substantially increased uncertainty about the pathways. The analysis supports an overall improvement in disability and a plausible individual pain-impact pathway, while showing that detailed mechanistic claims remain assumption-dependent.
Read the full abstract

General Mathematics

A Radon-Nikodým theorem for Fuzzy Number Multimeasures in Locally Convex Spaces

In this paper, we present a Radon-Nikodym theorem for dominated fuzzy number multimeasures defined on $\Sigma$ and taking values in $\mathcal{F}_{cwk}(X)$, where $(\Omega, \Sigma, \mu)$ is a complete finite measure space and $\mathcal{F}_{cwk}(X)$ is the family of generalized fuzzy numbers on a Hausdorff locally convex topological vector space $X$, i.e., a map $u: X \to [0,1]$ is a member of $\mathcal{F}_{cwk}(X)$, if and only if for each $\alpha \in (0,1]$ the $\alpha$-cut $[u]^{\alpha}$ of $u$ is a member of the family $cwk(X)$ of all nonempty convex and weakly compact subsets of $X$, ($[u]^{\alpha} = \{x \in X: u(x) \geq \alpha \}$).
Read the full abstract

Functional Analysis

A Counterexample to a Bourgain-Brezis Sum-Space Estimate

A question of Bourgain and Brezis, later restated as Open Problem 7.1 by Brezis, asks whether the $H^{1/2}+W^{1,1}$ norm of a mean-zero scalar function is controlled by the ambient $H^{-1/2}+L^1$ sum norm of its gradient. We show that this estimate fails in every dimension $N\geq 2$. More precisely, for every integer $K\geq 16$, we construct a smooth real-valued mean-zero function $\Phi_K$ on $\mathbb{T}^N$ such that $|\nabla\Phi_K|{H^{-1/2}(\mathbb{T}^N)+L^1(\mathbb{T}^N)}\leq C$ and $|\Phi_K|{H^{1/2}(\mathbb{T}^N)+W^{1,1}(\mathbb{T}^N)}\geq cK^{1/4}-C$, where the absolute constants $c,C>0$ are independent of $N$ and $K$. Equivalently, the ratio of the gradient-constrained decomposition cost to the ambient decomposition cost is unbounded. The construction applies the Hodge projection to a smooth approximation of a normal vector measure on a codimension-one sheet. Its potential records lacunary partial sums at geometrically separated distances from the sheet. A scale-path lower bound, combined with an odd critical Hardy inequality, excludes uniformly bounded $H^{1/2}+W^{1,1}$ decompositions.
Read the full abstract

Metric Geometry

A Power-Diagram Upper Bound for Twenty-Six Variable-Radius Disks in a Square

Let $S_{26}$ denote the supremum of the sum of the radii of 26 pairwise interior-disjoint closed disks of positive radius contained in the unit square. We prove $2.635983084917607783\ldots \le S_{26}\le \sqrt{\frac{2577}{2500}+\frac{11}{\sqrt3}}=2.716919756099889566\ldots$. The lower bound is accompanied by an exact rational feasibility certificate. For the upper bound, we assign the disks their clipped power cells, convert cell area into a reciprocal exterior-angle cost, and combine this local inequality with Euler side counts. The extremal case for the resulting relaxation has four boundary transitions and a four-cell boundary cycle; its continuous angle allocation is reduced to a concave one-variable problem. All finite numerical inequalities are rigorously enclosed by a reproducible certificate. The argument does not determine $S_{26}$ exactly, but gives a nontrivial global upper bound for a benchmark whose best-known packing is asymmetric.
Read the full abstract

Classical Analysis and ODEs

Some New Sharp Boundedness Result Concerning Linear functionals on Banach spaces

In this paper we state, prove, and discuss some new boundedness results concerning sequences of linear functionals of functions from the Lipschitz class \(Lip_1\) and general orthonormal systems. It is also proved that these results are, in a sense, the best possible.
Read the full abstract

All authors L.-E. Persson, V. Tsagareishvili, G. Tutberidze, G. Cagareishvili

Combinatorics

All polyhedral graphs with at most 16 vertices are Schurian: a computational verification

Li, Ponomarenko and Zeman state their belief that every polyhedral graph (i.e. every $3$-connected planar graph) is Schurian: that its coherent closure $WL(X)$ coincides with the orbital configuration of $Aut(X)$. Combined with their Theorem 1.1, this would imply that every polyhedral graph has Weisfeiler-Leman dimension at most $2$. We report a computational verification of the Schurian property for all polyhedral graphs with at most $16$ vertices, a total of 413,024,847,068 graphs, none of which is non-Schurian. The computation combines the generator plantri with a one-sided $1$-WL discreteness filter and, for every graph surviving that filter, a full comparison of the stable $2$-WL partition of $\Omega^2$ with the orbit partition of $Aut(X)$. The computation is guarded by the published anchor counts, a seeded re-enumeration of the two largest stages, independent nauty cross-checks and control graphs with mathematically fixed verdicts; code, checkpoints and a verification script accompany the note as ancillary files. For $n\le 13$ the result reproduces what already follows from the published enumeration of coherent configurations, by an entirely different route; from $n=14$ on, deriving it from that side would require a step we could not find in the literature, and at $n=16$ the all-graphs statement fails outright - the Shrikhande graph is non-Schurian - so there the polyhedral restriction carries the result. The statement is established by exhaustive computation; with respect to the general conjecture it is instance evidence, not a proof. A companion note treats the torus side, where non-Schurian triangulation graphs do occur at $n = 16$.
Read the full abstract

Computational Engineering

MiDShip: Multimodal Dataset of Ship Cargo Hold Structures for Engineering Design

Ship structures govern vessel strength, safety, and manufacturability, but their design must satisfy hundreds of classification society requirements, making the process complex and iterative. Data-driven approaches are limited by the lack of structured datasets linking design geometry, structural performance, and rule-based constraints. This paper presents MiDShip, a multimodal dataset of 12,753 synthetic cargo-hold structural designs: 6,020 random, 496 generated by an SGLD-inspired procedure, and 6,237 generated by an equation-informed repair procedure. Each design includes parametric data, full and mesh-ready 3D geometry, engineering drawings and annotations, a bill of materials, and preliminary structural evaluations. Twenty-five constraints derived from a subset of ABS MVR are also evaluated. None of the random designs satisfies all constraints. Among the SGLD-inspired designs, 322 (64.9%) were fully compliant, with an average of 0.409 violations, 82.7% below the seed mean and 96.9% below the random-design mean. The repair procedure, developed through LLM-assisted code analysis, produced 4,952 fully compliant designs (79.4%), averaging 0.296 violations, 97.1% below the paired-source mean. In equal-size comparisons, mean nearest-neighbor distances in the scaled 120-parameter space were 3.495 for repaired designs, 1.144 for SGLD batches, and 3.729 for random designs. The primary contribution is the synchronized dataset and its generation and evaluation infrastructure; the generation studies demonstrate its utility rather than proposing new optimization algorithms. MiDShip supports machine learning, generative design, and automated rule-based evaluation for ship structures.
Read the full abstract

All authors Noah J. Bagazinski, Md. Ferdous Alam, Jaya Manideep Rebbagondla, Faez Ahmed

Artificial Intelligence

K-Dense BYOK: An Open-Source AI Research Assistant That Runs Locally and Keeps a Hash-Chained Lab Notebook

K-Dense BYOK (bring your own keys) is a free, open-source AI research assistant for scientists in any field that runs on the researcher's own computer. The researcher supplies access to a model of their choice, hosted or running locally, and the application supplies everything else: a place for the work to run, a layer of scientific scaffolding, and a complete record. Each project is an ordinary folder, so the data, the code, the results, and the record stay on a machine the researcher administers and can be read years later without the application. Three things separate it from a chat assistant or a general-purpose coding agent. It ships a library of written scientific procedures, guided workflow templates, catalogs of where research data can be found, and reviewer and writer roles the agent can hand work to. It keeps a Living Lab Notebook whose entries link into an argument and are added to but never erased. And it records what happened by watching what the agent does rather than by taking the agent's word for it, in a log the agent has no tool that can write to. That choice targets the most common failure, model overclaiming, in our earlier benchmark of nine frontier models, by making claims checkable rather than preventing them. On twenty interdisciplinary research prompts, scored under a rubric fixed in advance, K-Dense BYOK led two managed platforms on both scientific quality and research execution. Its deliverables were the only ones that recorded the software they ran in, and the only ones that usually arrived with a command that regenerates the results. One of the managed platforms ran the same frontier model and supplied neither. Those environment records were files the agent wrote, not part of the observed log, which does not yet capture the software environment itself. The code is available under the MIT license at https://github.com/K-Dense-AI/k-dense-byok.
Read the full abstract

All authors Aubrey M. Brueckner, Darshil Patel, Yuhuan He, Timothy Kassis

Human-Computer Interaction

A Comprehensive Evaluation Framework for Conversational Home Energy Management Systems

The growing complexity in home energy management (HEM) demands advanced systems that guide occupants toward informed energy decisions reflecting their background, preferences, and context. Large language model (LLM)-integrated HEM systems (HEMS) have demonstrated promise, but previous studies relied on single-turn or single-task evaluations with response accuracy as the primary metric. Whether such systems deliver effective interactions across the extended multi-turn dialogues typical of real-world use remains an open question. This study introduces a comprehensive evaluation framework of LLM-integrated HEMS derived from the Goal-Question-Metric methodology, organized across five categories: task performance, factual accuracy, interaction quality, control capability, and system efficiency. A total of 23 metrics across multi-turn conversations are proposed and an LLM-as-judge pipeline is employed to enable scalable automated scoring. Its reliability is validated against three trained human coders: after iterative rubric calibration, twelve of the fifteen LLM-scored metrics reached strong agreement (ICC >= 0.73), three of them perfect, while the remaining three exhibited near-zero variance in human scores and are instead reported via mean absolute error (0.04-0.28). To demonstrate the framework's effectiveness, 970 dialogues – 16 scenarios and five personas – were generated and evaluated across four conversational HEMS configurations spanning a sophistication gradient, from a vanilla LLM with raw energy data to a multi-agent HEMS. The framework distinguished the four configurations across multiple evaluation dimensions, revealing their respective strengths and weaknesses. This study contributes to conversational HEMS by providing a reproducible, multi-dimensional evaluation methodology that comprehensively assesses sustained, context-aware system performance.
Read the full abstract

Statistical Methodology

Dynamic Spatial Bayesian Machine Learning Model: Applications to Intergenerational Economic Mobility and Geographic Income Inequality in the United States

We develop a Dynamic Spatial Panel Bayesian Additive Regression Trees model with Horseshoe shrinkage (DSP-BART-HS) for high-dimensional spatio-temporal panel data. We jointly evaluate the model against a comprehensive suite of structural spatial econometrics, non-parametric machine learning methods, and small-area estimators across nine data-generating scenarios spanning 200 replicates each, including irregular spatial topologies, dense policy effects, and non-linear individual-level interactions. DSP-BART-HS is the best or statistically indistinguishable from the best estimator in every scenario. Conventional region-time-aggregate comparators suffer severe performance degradation – trailing by a factor of three or more – whenever individual-level non-linearity drives outcome variance, a limitation this framework's tree-ensemble design directly addresses. The model also maintains strong predictive accuracy under a zero-training-region spatial holdout via its spatial diffusion mechanism. We demonstrate practical utility across two U.S. county-level panel applications – intergenerational economic mobility and geographic income inequality – under genuine unseen-region, random, and temporal holdouts, with results validated through repeated-split uncertainty quantification and paired significance testing against every comparator. One clear limitation emerges: under temporal extrapolation specifically, a lighter per-region autoregressive specification (MTS-CAR-X) achieves a robust, and consistent advantage. DSP-BART-HS nonetheless establishes strong predictive performance, automated variable selection, and flexible inference for hierarchical spatio-temporal panel data.
Read the full abstract

Computers and Society

Who Judges the Frame? Auditing Multimodal LLM Judges for News Framing Across Event-Level Perspectives

News coverage of major world events is shaped not only by what is reported, but also by how events are framed through text, images and their combination. At the same time, Large Language Models (LLMs), including multimodal LLMs, are increasingly used as scalable instruments for analysing framing, sentiment, ideological slant and perspective differences in multimodal media datasets. This creates a methodological challenge. When used as measurement instruments, LLM outputs may reflect not only content properties, but also model-specific tendencies, prompt design choices, and social, political, cultural, linguistic or modality-specific assumptions. This work audits LLMs as instruments for large-scale framing and perspective analysis in multimodal news coverage. Using an event-centered dataset of 2025–2026 news coverage, where each event includes left-, center- and right-oriented articles about the same headline, we combine embedding-based measures of within-event viewpoint similarity with model-based assessments of framing constructs across modality-specific and metadata-visible input conditions. Rather than treating either dataset labels or model outputs as ground truth, our goal is to examine the usefulness and limitations of LLM-based media analysis. The study contributes an audit protocol that highlights the need to report modality effects, metadata sensitivity, and prompt-induced artifacts alongside substantive claims about news framing and ideological viewpoint differences.
Read the full abstract

Computation and Language

Measuring Human-Like Bias in LLMs? A Critique of Human-Derived Bias Constructs in LLM Evaluation

Researchers increasingly use human-derived bias constructs to study Large Language Models (LLMs), including social-cognitive constructs such as implicit bias and stereotype activation, and cognitive biases such as anchoring, framing effects, and confirmation bias. Such approaches offer alternatives to overt bias probes, particularly when direct questioning may obscure bias or when model behaviour appears normatively acceptable. However, adapting human bias constructs to LLMs introduces an inferential gap. Psychological instruments were developed to study human cognition and social behaviour, whereas LLM evaluations rely on probabilities, text completions, rankings, or simulated decisions. This paper critiques human-centered bias evaluation in LLMs. We show how this gap arises from mismatches pertaining to human-derived constructs, human-model differences, and evaluation contexts, which can blur distinct interpretations of model bias. We then introduce a framework providing an analytical lens for relating these elements to warranted interpretations, with attention to target constructs, operationalizations, scope of inference, and limits of human analogy.
Read the full abstract

Computer Vision

A Framework for Egocentric and Exocentric Procedural Understanding via Temporal Segmentation and Semantic Abstraction

Long-horizon ego/exo data contains rich procedural evidence, but are redundant, noisy, and costly to process or retain. We propose a compact framework that converts continuous multimodal workplace video into a structured Procedural State Memory, implemented as a Work Environment Model (WEM). Inspired by event segmentation theory, we detect boundaries using changes in visual context, location, motion, narration, gaze/object interaction, and optional exocentric workspace evidence, rather than fixed windows or visual novelty alone. Each segment is abstracted into an evidence-linked event card containing actor, interval, location, action, objects/tools, pre/post state, confidence, and provenance. These event cards incrementally update the WEM, enabling compact, auditable documentation and retrieval under on-premise privacy constraints. We instantiate the design with frozen DINOv2 and VJEPA-2 encoders and a local language model, and outline evaluation criteria for segmentation quality, memory compression, retrieval fidelity, and long-horizon QA.
Read the full abstract

Signal Processing

Sparse-LDV Neural Reconstruction with Vibro-Acoustic Evidence Fusion for Early Bolt Preload Loss

Laser Doppler vibrometry (LDV) is costly, while early bolt-preload loss may produce weak and resonance-dependent changes. We combine sparse-LDV reconstruction referenced to a dense commissioning baseline with force-normalized microphone evidence. At m = 24 measurement nodes, each single-bolt 5-Nm state is reconstructed by a neural-network fold that excludes that state from training. A late-fusion rule calibrated on the healthy baseline combines the 90th percentile of local-FRAC deficits at unmeasured nodes with microphone H1 changes normalized by impact-level repeatability. Sparse LDV alone flags 24/28 mild condition-resonance pairs (50% preload loss) and acoustic evidence flags 25/28, fusion flags all 28/28. All four LDV misses occur at RG3 (resonance group) and are recovered acoustically. Fusion remains 28/28 for repeatability-normalized acoustic thresholds from 3 to 10. The result supports complementary vibro-acoustic evidence for early preload-loss screening while avoiding pointwise fusion of separately acquired campaigns.
Read the full abstract

Computer Vision

Robust Online Aero-Engine Blade Defect Detection via Dual-Alignment Test-Time Adaptation

Reliable visual inspection is essential for quality assurance in aero-engine blade manufacturing, where defect appearance may vary across production lines, imaging conditions, blade poses, and surface backgrounds. Such domain shifts cause a mismatch between training and deployment data and degrade the reliability of deep defect detectors in online inspection. This problem is particularly challenging because aero-engine blade images usually contain sparse defects, making pseudolabel-based adaptation vulnerable to noisy or missing predictions. To address this issue, we propose Aero-engine Blade Defect Detector (ABDD), an online adaptive detection framework based on test-time adaptation. ABDD introduces a Dual-Alignment Strategy to jointly adapt global visual style and local defect morphology by combining feature-statistics alignment with pseudo-box alignment. To reduce error accumulation from unreliable pseudo labels, an Uncertainty-aware Box Filtering mechanism evaluates pseudo boxes using classification confidence, classification entropy, and localization entropy. In addition, a lightweight Sparse Dilated Mona module enables parameter-efficient delta tuning while limiting source-domain forgetting. ABDD is evaluated on CD-AeBD and HD-AeBD under multiple domain-shift scenarios, with TTA strategies compared under a unified RT-DETR + Swin-T architecture. Experiments show that ABDD consistently improves detection robustness under domain shifts, and its practicality is further validated on an industrial inspection platform.
Read the full abstract

All authors Zhaoyang Wang, Haiyong Chen, Dongying Li, Yining Wang, Huapeng Wu, Xinwei Lv, Atik Shahariar

Differential Geometry

Detecting Einstein metrics on products of spheres

On \[ S^5\times S^2,\qquad S^5\times S^3,\qquad S^5\times S^4,\qquad S^7\times S^2, \] we construct infinitely many pairwise non-isometric real-analytic Einstein metrics. After discarding finitely many members, none is isometric or homothetic to a Riemannian product or to a classical two-summand Böhm metric. The metrics have the form \[ dt^2+a(t)^2g_{S^p}+b(t)^2g_{S^p}+c(t)^2g_{S^q}, \qquad p,q\ge2,\qquad 2p+q\le8. \] The construction starts with solutions for which the two non-collapsing radii at one singular orbit are very small. After magnifying this region, the Einstein equations approach a Ricci-flat problem whose solutions tend to a cone by a decaying oscillation. The main point is to determine exactly which part of this oscillation can still be seen by the closing conditions at the other end of the manifold. Rather than following the full, increasingly oscillatory shooting map, we extract two response quantities from the Ricci-flat limit. One determines the common oscillatory phase; after that phase has been removed, the second closing condition is reduced to a scalar sign change. A uniform inner–outer matching argument transfers this limiting sign change to the exact Einstein equations. As the initial scale tends to zero, the phase makes arbitrarily many turns, and the closing conditions are therefore met infinitely many times.
Read the full abstract

Robotics

Probabilistic Plan Legibility with Off-the-shelf Planners

Legible planning is the creation of plans that best disambiguate their goals from a set of other candidates from an observer's perspective. In this paper we propose a method for legible planning for arbitrary PDDL domains, by extending previous research on legibility to classical planning without requiring to construct ad-hoc planners. We also discuss how the observer perspective may be estimated through a second order theory of mind that connects the planner's and the observer's task spaces. Our solution can for example be deployed in human-robot teaming scenarios, where an autonomous robot in a team can implicitly communicate its goal by producing legible plans. We present benchmark results on several PDDL planning domains. Our results generally show that plan legibility is a trade-off with plan efficiency, however, not all planning domains allows to increase legibility in the same way and a regularizing factor to balance legibility and efficiency was proved necessary.
Read the full abstract