arXiv subjects

Search papers

Search indexed arXiv papers on artificial intelligence, large language models, computer vision and robotics. Read source abstracts and follow links to arXiv.

At least 1,225 records · Page 68Linked to original sources

arXiv2026

Pos4CHD: A Positive-Free Mask-Guided Counterfactual Synthesis Framework for Fetal Congenital Heart Defect Analysis

The scarcity of positive cases remains a major bottleneck in developing automated diagnostic systems for fetal Congenital Heart Defects~(CHDs), particularly for rare subtypes. Although normal cases are accessible with effort, positive cases are hard to collect due to low incidence rates and privacy concerns, causing severe class imbalance or absence of positive training samples. Existing works resort to synthesizing positives with conditional generative models, but they are task-specific and require positive training samples, hindering application for fetal CHDs. To this end, we propose a positive-free mask-guided counterfactual generation framework for fetal echocardiography, namely Pos4CHD, which enables realistic, high-quality synthesis of pseudo-positive samples from semantic masks, with only normal samples at hand. Technically, we first develop FE-Gen, a mask-guided generative model tailored for fetal heart ultrasound synthesis that mitigates histogram drift and enhances image-mask spatial alignment. Building on FE-Gen, Pos4CHD employs a mask-to-image generation paradigm for CHD synthesis. It first manipulates semantic masks to simulate fetal-heart structural anomalies and then renders anatomically consistent images. Extensive experiments on in-house datasets, together with expert evaluations, demonstrate the visual fidelity~($-12.97$ FID) and anatomical consistency~($+3.96\%$ DSC) of synthesized images. Further validations across diverse real-world CHD diagnosis tasks exhibit promising results and consistent increments, including zero-shot classification~(achieving $93.25\%$ ACC on EbA and $72.92\%$ ACC on TAPVC), lesion detection~($+6.95\%$ mAP50 on VSD), and anatomy segmentation~($+10.49\%$ DSC on fetal heart) compared with baselines. Codes and models are publicly available on \href{https://github.com/DREAMXFAR/Pos4CHD}{GitHub}.

cs.CV↗

arXiv2026

GM-GReFEL: Geometry-Guided Spatiotemporal State-Space Learning for Robust Dynamic Facial Expression Recognition

Recognizing facial expressions in unconstrained video suffers from two coupled problems: self-attention grows quadratically in cost with clip length, and crowdsourced labels are noisy. We present GM-GReFEL, a spatiotemporal state-space network with a geometry-aware decision stage. Its Modulated GroupMamba backbone replaces a spatiotemporal Vision Transformer. Six channel groups are scanned separately, four over the frame plane and two over time, so the cost grows linearly with sequence length ($\mathcal{O}(N)$); after each layer, a Channel Affinity Modulation gate mixes the groups again. On top of the backbone, a GReFEL reliability module turns the normalized entropy of each prediction into a mixing weight between the classifier output and a vote from trainable geometric anchors, so that facial geometry counts for more when the network is unsure. The model is initialized by static-to-dynamic weight transfer and trained with a Triple Loss. In five-fold cross-validation, the best fold scores 66.57\%/76.46\% UAR/WAR on DFEW, 0.3 points or less below the strongest compared method, and 45.66\%/62.27\% on MAFW, ahead of all compared methods; on FERV39k the mean over five splits, 42.30\%/52.66\%, is competitive. The five-fold means are 64.07\%/74.54\% on DFEW, on par with recent large pre-trained models, and 38.90\%/54.37\% on MAFW, below the best of them. A paired $t$-test on DFEW indicates a modest benefit of temporal modeling (+1.61 UAR points, one-sided $p=0.031$). Attention stays on plausible facial regions, and latency is close to a ViT-B/16 baseline (47.4 vs.\ 46.8 ms per clip). Code: https://github.com/ardiytama/GroupMamba-DFER.

cs.CV↗

arXiv2026

Intraday Forecasting Ability and Liquidity Provision of Institutional Traders

This study uses a high-frequency dataset from a stock exchange to examine how forecasting ability and liquidity provision relate to institutional traders' proprietary intraday profitability. We discover persistent forecasting abilities among traders. Specifically, we demonstrate that hypothetical flow-following strategies based on the execution flows of these skillful traders yield annualized Sharpe ratios ranging from 16 to 63 across various trading frequencies. We match buy and sell executions into positions, linking realized profit and loss (P&L) to horizon-matched forecasting proxies and the liquidity roles of both executions. Our findings indicate a direct correlation between improved forecasting ability and profitability, where a one-standard-deviation increase in the forecasting ability proxy is associated with 1.02 -- 1.19 currency units higher P&L per matched position. Additionally, we measure liquidity-provision intensity within a matched position by its number of passive executions: zero, one, or two. In these regressions, P&L is 1.28 -- 1.31 currency units higher for one passive execution than for none, and a further 3.79 -- 3.82 higher for two than for one, underscoring the significant financial benefits of liquidity provision.

q-fin.TR↗

arXiv2026

Interaction-enhanced antiferromagnetic correlations in a Kondo ladder with a gapped band structure

We investigate magnetic correlations in the two-leg Kondo-Hubbard ladder using the density-matrix renormalization group method. We focus on a parameter regime in which the noninteracting electronic system has a narrow indirect band gap. In this regime, there is no Fermi surface, and the conventional Fermi-surface-based Ruderman-Kittel-Kasuya-Yosida picture is not directly applicable; magnetic correlations between localized spins become weaker as the noninteracting band gap increases. We find that a moderate on-site Coulomb interaction enhances the antiferromagnetic correlations between localized spins. This enhancement, which is also seen in the staggered magnetization induced by a pinning field at the edges, occurs for both ferromagnetic and antiferromagnetic Kondo couplings. Our results show that electron-electron interactions can enhance magnetic correlations between localized spins even when the underlying noninteracting electronic structure is gapped, suggesting an interaction-driven route to magnetic instabilities in narrow-gap Kondo-lattice systems.

cond-mat.str-el↗

arXiv2026

Ultra-hard-edge universality under deterministic preconditioning

Let $A_n$ be an $n\times n$ real matrix with iid entries distributed as a fixed centered, variance-one, symmetric subgaussian random variable. For deterministic invertible $M_n$, set $h_n=\|M_n^{-1}\|_{\mathrm{HS}}$ and $B_n=M_n^{-1}/h_n$. We prove $\mathbb{P}(s_{\min}(M_nA_n)\le\varepsilon/h_n) =(1+o(1))\varepsilonΘ_n(B_n)$ as $n\to\infty$, uniformly over $M_n$ and $e^{-cn}\le\varepsilon\le n^{-C}$. Here $Θ_n(B)=\mathbb{E}\sqrt{nU^\top BB^\top U}$, where $U$ is uniform on $S^{n-1}$. The constants $c>0$ and $C>1$ depend only on the entry law. No condition-number bound or effective-rank assumption is required. The coefficient tends to one for diffuse inverse profiles. For convergent concentrated or spiked profiles, its limit is an explicit Gaussian expectation. The proof combines the submicroscopic comparison of Sah, Sahasrabudhe, and Sawhney with signed-permutation averaging and a weighted comparison for random hyperplane normals. Applications include inverse amplification, square sample covariance matrices with arbitrary deterministic positive-definite population matrices, and independent random environments.

math.PR↗

arXiv2026

Spectroscopy, Regge trajectories and decay properties of the $bc\bar q\bar q$ and $bq\bar c\bar q$ tetraquark systems

Bottom-charm tetraquarks probe the interplay of heavy-quark dynamics, light-quark correlations, color structure and nearby hadronic thresholds. We investigate the spectroscopy and strong-rearrangement properties of the $bc\bar q\bar q$ and $bq\bar c\bar q$ ($q=u,d$) systems in a Cornell-type diquark--antidiquark framework with nonrelativistic (NR) and semirelativistic (SR) kinetics. Scalar and axial-vector cluster configurations are treated explicitly, and model-parameter uncertainties are propagated into state-dependent mass errors. For the $\bar{3}_c\otimes 3_c$ axial-axial ground states, the $0^+$ masses are $7530.1\pm75.4$ MeV (NR) and $7720.7\pm78.8$ MeV (SR) for $bc\bar q\bar q$, and $7307.6\pm90.5$ MeV (NR) and $7523.5\pm93.6$ MeV (SR) for $bq\bar c\bar q$. The multiplets follow $0^+<1^+<2^+$, and radial spacings decrease with excitation. In the $bq\bar c\bar q$ sector several low-lying configurations approach the $B^*D^*$ threshold: the axial-axial $0^+$ state lies $25.7$ MeV below it in NR but $190.2$ MeV above it in SR, so its threshold assignment is model sensitive and not evidence of robust strong-interaction stability. Spin and color wavefunctions are recoupled independently, giving configuration-dependent $PP$, $PV$, $VP$ and $VV$ rearrangement strengths and distinct color-singlet projections for the $\bar{3}_c\otimes 3_c$ and $6_c\otimes\bar{6}_c$ sectors. The strong short-distance attraction of the sextet coefficient $k_s=-10/3$ makes the finite SR expansion unreliable in that branch, so quantitative masses are restricted to the antitriplet sector. The results provide experimentally useful signatures and a baseline for coupled-channel and color-mixing studies.

hep-ph↗

arXiv2026

RankAlign: Vision-Language Representation Alignment via Rank Transformation

While the Platonic Representation Hypothesis posits that vision and language models converge toward compatible latent geometries, recovering class-level correspondences between unpaired modalities without cross-modal supervision remains challenging. Existing relational matching methods compare within-modality similarity kernels directly, although the same semantic structure may be represented through different numerical parameterizations across modalities. We propose RankAlign, a rank-based normalization strategy for class-level unpaired vision-language matching. RankAlign transforms each within-modality similarity kernel into an ordinal relational representation before permutation matching, while leaving the correspondence space and matching solver unchanged. The resulting representation is invariant to strictly increasing reparameterizations of similarity values while preserving the ordinal structure used for matching. We provide exact-recovery and stability guarantees that characterize correspondence recovery under both exact and partial ordinal agreement. Experiments across diverse frozen vision and language encoders and multiple datasets demonstrate improved and more stable class-level correspondence recovery, including under label corruption and domain shift.

cs.CV↗

arXiv2026

Rethinking Fisher Vectors for Second-order Multi-scale Deep Feature Aggregation in Scalable Scene Retrieval

Aggregating CNN responses into a compact discriminative vector, given their lack of geometric invariance, is a challenge. To address this challenge, we propose a codebook-free method to encode second-order information of multi-scale CNN features within the Fisher Kernel framework. We use a scalable retrieval pipeline to show the efficacy of the proposed vector. The increased discriminative power is confirmed by experimental evaluation of our method on two difficult benchmarks (INRIA Holidays, UKBench). The significant improvement in retrieval accuracy over state-of-the-art approaches comes with only a slight dimension increase.

cs.AI↗

arXiv2026

PriMa-Causa: Causal Prescriptive Maintenance

Automated manufacturing requires data-driven decisions under uncertainty and sparse observational records, and unplanned downtime is a dominant cost of getting them wrong. Prescriptive maintenance recommends a maintenance action against an impending failure rather than merely predicting it. Established reference architectures couple predictive models with rule- or optimisation-based recommendation logic. Selecting a corrective action, however, requires an interventional quantity: how the line behaviour would change if the action were set. Manufacturing data is observational, so a model fitted to predict the outcome estimates an associational quantity and may misdirect the choice of action without the prediction error revealing it. Causal estimators target the interventional quantity but require an expert-led causal study per case, for which the bounded decision window of operation leaves no room. This dissertation analyses how causal reasoning can be embedded into prescriptive maintenance so that actions are ranked by their estimated interventional effect. The result is PriMa-Causa, built from three components: a causal reasoning layer that integrates the estimate into the PriMa reference model, a parametric simulator that supplies the interventional ground truth absent from manufacturing data, and a graph-conditioned tabular causal foundation model that, pre-trained once on this synthetic production-line prior, answers interventional queries in a single forward pass. PriMa-Causa attains a lower prescription error than associational baselines, with the largest margin under weak overlap and sparse context, and a partially disclosed causal graph reduces this error further. The causal inductive bias of the synthetic prior and the graph conditioning can thus replace per-case estimator fitting, a step toward automating causal inference for autonomous and self-healing maintenance.

cs.LG↗

arXiv2026

Prompt Sensitivity in Tool-Augmented Model Comparisons

Language models are compared on question-answering benchmarks that pose each question in one wording, and a reworded question can change which of two models scores higher. We ask whether giving the models tools makes such comparisons more stable. An accuracy gain cannot answer this, because tools change several steps between the question and the scored answer at once. We propose a matched protocol that separates those steps: the answer readout, reasoning, the decision to call, the tool's answer, and the report supplied without a call. We apply it to six open models in five domains under five wordings. We find that tools raise accuracy where a program answers the question and widen the wording sensitivity of many pairwise comparisons. The widening follows how often a model calls, and call frequency follows the wording for some models and not others. It also depends on the interface and the instructions: a native tool-calling interface and a second tool description leave no detected widening in molecules, and sensitivity persists where tools return intermediate results. These findings show that deterministic reports do not guarantee stable comparisons. We recommend reporting call frequency and comparison stability under several wordings beside accuracy.

cs.CL↗

arXiv2026

Gamma conjectures for Milnor hypersurfaces

We prove Gamma conjectures I and II of Galkin, Golyshev and Iritani for Milnor hypersurfaces, the smooth hypersurfaces of bidegree $(1,1)$ in the product of two projective spaces. We provide a criterion for Gamma conjecture II for general Fano manifolds, generalizing the criterion by Hu and Ke. Using the mirror symmetry technique of Galkin and Iritani, we establish global Gamma-I for Milnor hypersurfaces, thereby proving Gamma conjecture I as a special case. Using it, together with the Galois action and our criterion, we verify Gamma conjecture II for Milnor hypersurfaces.

math.AG↗

arXiv2026

WildFakeToolBench: Benchmarking AI-Generated Image Detectors with Web-Sourced Image Processing

As AI generated images are getting popular these days, discussions on such generations are widely existed in current public forums, software repositories, and photography tutorials. Moreover, with the help of LLM-assisted code translation, malicious actors can adopt such information to evade detectors without specialized knowledge or local GPU execution of the resulting tools. To evaluate such risks, we introduce \textsc{WildFakeToolBench}, a benchmark suite of $156$ executable image-processing tools derived from $220$ public sources across $142$ domains. Each tool retains its source evidence and passes syntax, execution, and output-distortion checks without using feedback from the evaluated detectors. On $105$ generated images from five generators, we evaluate six state-of-the-art detectors with images processed by the tools in our collected bench to assess their effectiveness against such in-the-wild risks. The strongest individual tools achieve attack success rates of $25.25\%$--$100\%$ across detectors, exceeding the strongest of six selected image-processing baselines for each detector. Evaluating subsets of an eight-tool pool yields worst-case coverage above $80\%$ for every detector. Effectiveness is sparse and detector-specific: high coverage across candidates does not imply reliable one-shot evasion. These findings not only highlight the potential of such web-sourced risks, but also show the potential on establishing AI-image detector robustness evaluation framework based on those in-the-wild information.

cs.CV↗

arXiv2026

Balanced Posted Prices in Stationary Two-Sided Markets

We study an intermediary that buys nonperishing items from arriving sellers and sells them to arriving buyers, subject to limited storage. Sellers and buyers arrive according to independent Poisson processes. The intermediary pays and charges the same fixed item prices, aiming to maximize long-run gains from trade: buyer value minus seller cost. Our benchmark may choose allocations directly, has unlimited storage, and must purchase each unit before allocating it. Its optimal long-run gain rate equals the value of a linear program of average purchase and allocation rates, with or without knowledge of future arrivals. For buyers whose bundle values are maxima of nonnegative additive valuations (XOS), we compute static common prices that guarantee half of this benchmark for arbitrary finite item capacities, even one storage slot per item, regardless of how ties are resolved without using future arrivals. The prices are computable in polynomial time using demand and supporting-value queries. The construction equalizes itemwise lower bounds on buyer and seller contributions to gains from trade, accounting for procurement costs. We call this contribution balance. As a single-item foundation, we also equalize the rates of purchases and sales selected by the price, before checking inventory availability. We call this rate balance. For capacity $C$, it yields the tight guarantee $C/(C+1)$. The two principles parallel proofs of the classical prophet inequality based on half the expected benchmark and on a median threshold, adapting these balancing ideas to recurring two-sided markets with limited inventory.

cs.GT↗

arXiv2026

Generalized modeling reveals stabilizing mechanisms in cross-feeding microbial communities

Understanding what makes microbial communities stable and able to resist perturbation is a fundamental question in microbial ecology. Theoretical approaches to this problem typically require making strict assumptions about the structure, parameters and functional forms used in models, hindering general insight. Here, we address this problem by developing and analyzing a generalized model of microbial community dynamics, capturing resource-mediated competitive and cross-feeding interactions between populations. The generalized modeling approach allows us to assess the stability of microbial systems across large ensembles of models with variable size, structure and non-linear dynamics. Our analysis shows how community stability is constrained by system size and identifies its key metabolic and dynamical drivers, highlighting the central roles of self-limitation, metabolic efficiency and plasticity. Furthermore, we show how the structure of communities mediates their stability, allowing communities to counteract the destabilizing cross-feeding loops through organization into hierarchical trophic structures. Together, these results provide new insight into the stability of microbial communities and offer a flexible analytical framework for future work to explore the responses of communities to perturbation.

q-bio.PE↗

arXiv2026

Self-averaging enables reconstruction of higher-order chaotic networks

Higher-order interactions create a combinatorial space of candidate dynamical terms, making network reconstruction difficult when node dynamics, coupling functions and interaction topology are unknown. Here we identify a regime in which weakly coupled higher-order chaotic networks admit a higher-order interaction reduction. Weak coupling largely preserves the statistical structure of intrinsic chaotic motion, while dense higher-order neighborhoods provide repeated sampling of weakly correlated multi-node inputs received by hub nodes. Consequently, the aggregate interaction field can be represented by a compact function of the hub state, together with a node-dependent offset and finite-neighborhood fluctuations. We exploit this reduction to learn dynamically relevant coupling forms and reconstruct interactions of different orders by sparse regression. Numerical experiments validate the effectiveness of the framework within an effective region of the tested coupling-parameter space, including representative weak-coupling settings. Reconstruction remains stable at the larger network sizes and higher hub-neighborhood densities examined. Additional tests on networks constructed from empirical topologies show that structural priors reduce data requirements and that the framework remains applicable to heterogeneous chaotic dynamics. These results identify reduction of aggregate higher-order inputs as an interpretable route to dimensional compression and joint dynamical and structural reconstruction.

nlin.CD↗

arXiv2026

Calibration-free IMU-based Relative Orientation Estimation under Dynamic Human Motion

Recent magnetometer-free inertial measurement unit (IMU) methods can reduce drift in joint motion estimation using rigid-body constraints, but they typically require pre-calibrated sensor locations, assumptions of specific joint models (e.g., hinge joints) or limited motion types, or an external reference for global orientation. Furthermore, challenges related to high-frequency impacts (e.g. at foot strikes during gait) and numerical differentiation of noisy measurements have not been explicitly addressed. We propose an extended Kalman filter that estimates relative IMU orientation and sensor location simultaneously, requiring neither offline sensor location calibration nor a global orientation anchor. A systematic simulation study shows that the quality of the relative orientation and the sensor location estimates depends on different motion-intensity sources: the joint center acceleration and the relative angular velocity, respectively. A low-pass filter (LPF) is used to mitigate real-signal outliers before applying the rigid-body constraint. When applied to the OpenSense dataset with real IMU measurements across four subjects with three lower-body joint types during walking and complex tasks including various motion types, the proposed method achieves an overall relative orientation error of 2.61°, compared with 10.40° for a baseline using offline-estimated sensor locations and 7.43° for a baseline using marker-derived sensor locations. These results show that the proposed calibration-free estimation framework handles real IMU data effectively and achieves accurate relative orientation without offline sensor location calibration.

cs.RO↗

arXiv2026

Functional understanding of structural components in polypseudorotaxane gels

Biomedical devices, wearable interfaces, and soft robots require compliant soft materials with reliable mechanical performance, yet the molecular principles needed to design such materials are not always fully understood. Thus, there is an urgent need to understand how to design novel stretchable and soft hydrogels with enhanced toughness, i.e. resistance to breakage. Slide-ring gels (SRGs) are a class of unconventional soft materials formed by polyrotaxanes (PR) building blocks whereby sliding cyclodextrins (CD) replace traditional chemical and physical crosslinks and in turn enhance toughness and stretchability by several orders of magnitude. The rational design of SRGs re- mains challenging due to a lack of a mechanistic understanding of the role played by individual components. In this work, we implement simple synthetic perturbations that individually target polymer-CD and CD-CD interactions in unstoppered pseudopolyrotaxanes (pPR) solutions and measure the consequent changes in viscoelastic properties. Weakening the intermolecular interaction between CDs abolishes crystallisation and significantly weakens the gel. At the same time, thermodynamically favouring threading by using azobenzene-modified PEG chains favours the formation of stable and long-lived networks even at remarkably low CD stoichiometries. We rationalise these observations with coarse-grained Molecular Dynamics simulations which clarify the structure-function relationship of the components forming these materials. Our work contributes to guiding the rational design and development of enhanced SRG hydrogels and elastomers.

cond-mat.soft↗

arXiv2026

Beyond Watching and Searching: Benchmarking Agents for Video-centered Deep Research Reports

Deep research agents synthesize open-web evidence into long reports, but current benchmarks do not evaluate research centered on a source video. We introduce Video-Centered Deep Research Report Generation. Given a video and user request, an agent identifies research questions, retrieves external multimodal evidence, and produces an interactive report linked to relevant video moments. This task has direct practical value because it helps users locate relevant moments, verify claims, reproduce demonstrated procedures, and make informed decisions from video content. We present a skill-based workflow and OmniDR-Bench for this task. For evaluation, we propose Omni Atomic Rubric Trees (Omni-ARTs). Each tree contains three dimensions, 23 shared directions, and task-specific atomic binary criteria. The directions test audiovisual grounding, multimodal evidence use, and interactive presentation. Audited Rubric Construction (ARC) builds these trees through independent evidence collection, automated checks, behavioral calibration, and iterative human review. OmniDR-Bench contains 50 Chinese and English tasks across 19 realworld subcategories. It includes 6,414 criteria. Across 18 agent configurations, the best system scores 63.3 out of 100. Current agents still need broader content coverage, stronger causal and logical analysis across modalities, and more accurate citation support. We further analyze models, harnesses, skill configurations, video categories, and research trajectories. The code and data are released at https://github.com/marinero4972/OmniDR-Bench.

cs.CV↗