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 883 records · Page 49Linked to original sources

arXiv2026

Entertain, then Cancel: How Thinking Makes Reasoning Models Read Some and Or Literally

When a speaker says ``some students passed'', listeners infer that not all of them did. Reasoning models now let users decide how long a model thinks before it answers, but we do not know what extra thinking does to such everyday inferences. We study this directly: we keep a model's weights fixed, change only how much it thinks, and record which reading it chooses. A second tool cuts the model's own reasoning at successive points and asks for an answer, showing where the reading changes. Thinking makes models read logical words as logic: it cancels the implicatures of \emph{some} (\emph{not all}) and \emph{or} (\emph{not both}) in most tested open reasoning models and in GPT-5, while context-driven inferences such as indirect answers improve. The trace shows why: models first entertain the enriched reading and cancel it later. Answering from that early reading, an exploratory inference-time remedy, recovers these implicatures without losing literal accuracy. Evaluations of reasoning models should measure how they interpret, not only how accurate they are.

cs.AI↗

arXiv2026

SpaceCache: Intermediate Feature Reuse for Efficient Semantic Segmentation on Point Cloud Streams

Point-cloud streams are central to autonomous driving and robotics, yet existing segmentation networks process every frame from scratch, although consecutive frames largely overlap. We observe that the intermediate features of point-cloud transformers remain semantically consistent across frames for the same spatial region, especially in deep stages. Building on this, we present SpaceCache, a training-free feature cache that reuses these features across frames. Tokens of previously observed regions skip a sequence of transformer blocks and take their cached features, only the remaining tokens are computed, and the full token layout is restored for dense prediction. Lightweight cache management keeps reused features reliable by limiting each entry's age and evicting low-confidence entries. Unlike token merging, SpaceCache selects tokens at negligible cost and runs its cache operations once per encoder stage. On both outdoor and indoor benchmarks, such as nuScenes, Waymo, and ScanNet, with two transformer backbones of Utonia and LitePT-B, SpaceCache achieves up to 1.65x speedup with less than 1-point mIoU drop and consistently outperforms token-merging methods.

cs.CV↗

arXiv2026

Clean Data Can Still Carry Backdoors: Support-Persistent Backdoors for Model Reuse

Traditional backdoor attacks inject artificial trigger patterns into training data, causing models to misclassify triggered inputs while maintaining normal behavior on clean samples. Since these artificial triggers are absent in clean datasets, standard knowledge distillation (KD) typically eliminates them, leading to the common belief that KD serves as an effective purification defense. In this paper, we propose DAR (Discover-and-Relabel), a support-persistent backdoor method that retains its behavior even after clean-data distillation. DAR identifies naturally abundant patterns in low-dimensional feature subspaces using subspace clustering and applies label-only poisoning to samples that inherently satisfy the discovered rule. At test time, the backdoor is activated by locally modifying the corresponding subspace of the target images. By instantiating spatial and frequency operators, DAR achieves attack accuracy comparable to state-of-the-art backdoor attacks in ImageNet classification and CLIP-based prompt tuning. Notably, DAR keeps attack success above 96% after clean KD and remains defense-resistant in practice. Project page: https://junhoo.me/backdoor

cs.CR↗

arXiv2026

Plasmon Collapse in a Directional Strange Metal

Strange metals are widespread in strongly correlated quantum matter and closely intertwined with high-temperature superconductivity and quantum criticality, yet their microscopic charge dynamics remain fundamentally unresolved. Whether well-defined collective modes survive in an incoherent electronic fluid or dissolve into overdamped continua remains intensely debated. Here, employing high-momentum-resolution inelastic electron scattering complemented by direction-resolved transport, we show that the van der Waals compound $\mathrm{MoOCl}_2$, a "bad" metal known to host unusually "good" plasmons, exhibits a striking dichotomy in electronic coherence and charge dynamics. Transport reveals coherent Fermi liquid behavior along one in-plane crystallographic axis and an incoherent strange metal response along the orthogonal axis. This dichotomy extends directly to the charge response. The Fermi liquid direction supports a well-defined, long-lived plasmon, whereas the strange metal direction exhibits an overdamped excitation that progressively loses its collective character with increasing momentum. Beyond a critical momentum ($q_c$ $\approx$ 0.008 $\mathring{\mathrm{A}}^{-1}$), the mode undergoes a reversal into negative dispersion and ultimately collapses into a broad, nearly featureless continuum with a robust high-energy boundary ($E_c$ $\approx$ 1.75 eV). Remarkably, thickness-dependent measurements show that both $q_c$ and $E_c$ remain nearly unchanged despite substantial modifications of Coulomb screening, revealing them as intrinsic scales of incoherent charge density fluctuations. These results establish a unique single-material platform in which coherent and incoherent electronic states are partitioned into in-plane orthogonal channels, and demonstrate that incoherent electronic fluids retain intrinsically structured charge dynamics beyond the quasiparticle paradigm.

cond-mat.str-el↗

arXiv2026

Local modifications in planar hypohamiltonian graphs

We prove that for $n \geq 36$ there exists a planar hypohamiltonian graph of order $n$. As a corollary, we prove that for $n \geq 33$ there exists an almost hypohamiltonian planar graph of order $n$. Further, we exhibit over 100 planar hypohamiltonian graphs of order less than 40, including new graphs of orders 36, 37, 38, and 39. The proof utilizes three local modifications that alter a graph $G$ by removing a subgraph $H$ and replacing it with $H'$ to form a new graph $G'$: a plus two modification, a plus three modification, and Thomassen's iterable plus four modification. The plus two modification preserves hypohamiltonicity in general, and the plus three modification preserves hypohamiltonicity in our cases of interest. This establishes the main result beginning with Tsai's order 34 examples. Further, we conjecture that among planar hypohamiltonian graphs, Tsai's 34 vertex examples are the smallest order. The conjecture is based on previous results, constraints on the degree modulo 3, and experimentation with planar graphs consisting entirely of 4-gons and 5-gons.

math.CO↗

arXiv2026

An Agentic Framework for Proactive Robots with Procedural Assistance Memory

Proactive robots should anticipate user needs and provide assistance before being explicitly asked. A straightforward approach is to infer human intentions from commonsense knowledge, but the same observed behavior can imply different needs across users, potentially leading to unwanted assistance. We observe that everyday routines in the real world exhibit recurring patterns, so learning these patterns can reduce ambiguity in proactive decisions. Based on this, we propose a memory-based agentic framework in which a VLM task predictor reads a Procedural Assistance Memory (PAM) alongside the current observation to decide when to assist and what task to perform, and a VLA executor performs the predicted task. PAM is structured as a finite-state machine over activity phases, per-state assistance rules, and event-driven transitions; a reflection VLM revises PAM from each day's interaction trace without weight updates or retraining. In our 5-day desk-assistance evaluation, memory-free baselines remain conservative, with limited recall on proactive tasks. Our system improves consistently with accumulated experience, achieving the highest precision, recall, and F1 and reducing first-fire delay from approximately 5 seconds on Day 1 to 1 second by Day 5. Our approach also adapts to changes in the user's routine through continued memory reflection, recovering from memory--routine misalignment after a single memory update. The project page is available at https://davian-robotics.github.io/PAM/

cs.RO↗

arXiv2026

The Birth and Early Feedback of Massive Star Clusters Across the Local Universe

Young massive star clusters (YMCs; ${\gtrsim}10^4\rm\,M_\odot$) represent a distinct, extreme mode of star formation and the closest present-day analogues of the environments in which globular clusters and Cosmic Noon stellar populations assembled. Given the extreme dust extinction in the earliest phase of YMC formation, (sub-)millimeter observations are the only means of probing their assembly and early feedback. Pioneering works with ALMA are limited to individual region studies and to the most massive YMCs in them, leaving many open questions about the initial conditions of YMC formation, the integrated star and cluster formation efficiencies, the dominant pre-supernova feedback channels, and the prevalence and reach of cluster-driven outflows. ALMA2040 will enable mass-complete, resolved surveys in hundreds of host galaxies, transforming the study of massive cluster formation and feedback from individual case analyses into statistical measurements across the full range of galactic environments.

astro-ph.IM↗

arXiv2026

Mylar: Exploring Self-Curation as a Reflexive Creative Practice

The container holding an artist's work can itself be just as expressive as the work it presents. We investigate self-curation as a creative practice, asking how artists explore and express their identity through the design of their own artwork presentations. Drawing on a study of artists' existing presentation practices, we developed Mylar, a malleable art presentation system that lets artists rapidly generate and compare provisional portfolios, seeing their work anew through different montages. We observed how 12 artists articulated, probed, negotiated, and foraged in Mylar, using rapidly generated portfolios as epistemic artifacts for reflection. We characterize self-curation as a reflexive and plural practice, in which artists use presentation design to shape the legibility of their self and their work and reflexively correspond with provisional presentations to reshape their curatorial intentions. We share design implications for expressivity in online art platforms and for malleable prototyping as a formative design method.

cs.HC↗

arXiv2026

Rubric Judgments Lack Uncertainty: Conformal Weighting for LLM Evaluation

Rubric-based evaluation is widely used to assess large language model (LLM) outputs by decomposing quality into multiple criteria and assigning each a binary judgment from an LLM verifier. However, judgment certainty varies substantially across criteria. Existing methods uniformly aggregate these judgments into a final score, implicitly treating every criterion as equally certain and overlooking this heterogeneity. As a result, uncertain judgments remain incorporated into reward estimation without distinction, potentially yielding misleading training signals for downstream reinforcement learning (RL). To address this limitation, we propose Uncertainty-aware Rubric Weighting (URW), a training-free framework that quantifies criterion-level judgment uncertainty via conformal calibration. Specifically, URW elicits a soft uncertainty signal from a fixed pool of paraphrased judgment templates and converts it into a per-criterion weight using split-conformal $p$-values with provable coverage guarantees. Extensive experiments demonstrate that URW consistently improves pairwise preference accuracy and delivers stronger downstream RL alignment.

cs.AI↗

arXiv2026

DLOSAC: Joint Spatial-Geometric Image Estimation via Delaunay-Guided Local Optimization RANSAC

Reliable image geometric estimation from noisy matching points is typically addressed using RANSAC, often combined with local optimization to refine inliers and model parameters. However, existing paradigms are predominantly residual-centric; even heuristic attempts to enforce spatial uniformity suffer from a sequential, two-stage nature that inherently decouples distribution optimization from geometric refinement, preventing an effective trade-off. In this paper, we propose DLOSAC, a local optimization framework that jointly incorporates both geometric consistency and spatial distribution information. We begin by constructing a Delaunay triangulation over candidate inliers to explicitly encode neighborhood topology. The geometric module introduces an edge-based consistency metric to evaluate geometric errors, while the spatial module derives distribution descriptors from triangle areas and edge attributes to quantify the clustering of inliers. Finally, we combine edge-based geometric constraints, spatial distribution regularization, and traditional point-level geometric constraints to improve the quality of geometric parameters. Extensive experiments on six publicly available real-world datasets, covering three types of geometric estimation tasks, achieve state-of-the-art performance.

cs.CV↗

arXiv2026

On the Complexity of Symmetric Boolean Functions in Multiparty Private Simultaneous Messages

Multiparty Private Simultaneous Messages (PSM), also known as Decomposable Randomized Encoding (DRE) or information-theoretic garbling, is a model for non-interactive secure computation. Each party encodes its input using shared randomness and sends a message to an evaluator who does not know that randomness. The evaluator must recover the function output without learning anything else about the inputs. Perfect privacy requires identical transcript distributions on inputs with the same output. We also study the weaker notion of support privacy, which requires only that these distributions have the same support. We study the communication complexity of symmetric Boolean functions when each of the $n$ parties holds one input bit. Lower bounds. We prove a superlinear lower bound for support-private PSM protocols computing $OR_n$. Our technique gives $Ω(n\log n)$ lower bounds for $MAJ_n$ and $EMAJ_n$ (with even $n$), and an $Ω(n\log q)$ lower bound for $MOD_q$ when $q\leq(1-c)n$ for a fixed constant $c>0$. Upper bounds. We construct perfectly private PSM protocols of total communication $\widetilde{O}(n^{3/2})$ for every symmetric Boolean function, improving on the known quadratic bound. For $MAJ_n$, known branching-program constructions yield an upper bound of $O(n\log^3 n\log\log n)$. For $EMAJ_n$ and $MOD_q$ with prime $q$, the lower bounds are tight up to constant factors in the stated regimes. Under support privacy, we obtain $\widetilde{O}(n)$ communication for every symmetric Boolean function. Earlier lower-bound techniques of Ball et al. (ITCS 2020) yield only trivial linear bounds for symmetric Boolean functions. Concurrent and independent work by Holmgren and Wu proves a $ω(n)$ lower bound for $OR_n$ using the same combinatorial core. Both works used AI assistance.

cs.CR↗

arXiv2026

Strong Lefschetz property of the Kirchhoff algebra of the complete tripartite graph $K_{1,1,n}$

By Whitney's 2-isomorphism theorem, the cycle matroids of two graphs are isomorphic if and only if one graph can be transformed into the other by finitely many applications of two operations: separation or identification at a cut-vertex, and twisting. The complete tripartite graph whose vertex set is the disjoint union of two singletons and an $n$-element set is invariant, up to isomorphism, under these operations. In this paper we prove that the Macaulay inverse system (the Kirchhoff algebra) determined by the basis generating polynomial of the cycle matroid of such a graph has the strong Lefschetz property over a field of characteristic $0$. The proof reduces the non-degeneracy of the higher Hessians to the ranks of disjointness matrices of families of subsets (the Gottlieb--Wilson theorem) and to the representation theory of the symmetric group in relation to the Johnson scheme.

math.CO↗

arXiv2026

AutoOnboarder: Autonomously Onboarding LLM Agents

Agentic large language models have been documented to lose reasoning capability when placed inside a harness, the software that mediates their observations and actions. To reconcile these losses, we introduce AutoOnboarder, an autonomous system that adapts an agent to a task the way a trainer onboards a new employee. It places the agent under the supervision of an intelligent ``Onboarder'' coding agent, which iteratively tailors its foundation model, prompt, and harness source code to simultaneously improve performance and decrease cost. Like a real onboarding process, the Onboarder is only given tens of examples, and all evaluation data is isolated on a separate virtual machine to prevent leakage. We report the lessons learned from building and testing AutoOnboarder on genomic-sequencing prior authorization form submission. Autonomously onboarded agents are able to match or exceed the performance of a Gemini 3.5 Flash baseline at up to 29$\times$ lower costs. However, these agents remain imperfect, and the Onboarding process varies greatly in terms of final performance. We conclude that future Onboarding systems require a robust environment for the Onboarder, including a structured thought interface, (controlled) internet access, and a wide model search space.

cs.AI↗

arXiv2026

Using Heteroscedasticity for Causal Discovery under Hidden Confounding

The presence of hidden confounding in observational studies poses a fundamental challenge to the identification and estimation of causal structures encoded by directed acyclic graphs (DAGs). While instrumental variables (IVs) have been widely used to address hidden confounding, most existing methods require valid instruments, which may be unavailable in practice. In this paper, we develop a novel framework for causal discovery and inference in DAGs that accommodates both hidden confounding and invalid IVs by exploiting heteroscedasticity. We establish a novel linear characterization of leaf nodes through heteroscedasticity-induced relationships between cross-covariance and self-covariance vectors, enabling recursive recovery of the causal order without requiring any candidate IV to be valid. Given the identified order, we then sequentially construct surrogate IVs to identify the causal effects and directions, thereby recovering the entire DAG. We also propose a data-adaptive procedure for estimating the DAG and causal effects. We show that the proposed method consistently recovers the underlying DAG and yields asymptotically normal estimators of causal effects. Extensive simulation studies demonstrate substantial improvements over existing methods. We further illustrate the practical utility of the proposed method through an application to gene regulatory network analysis.

stat.ME↗

arXiv2026

Regions of Attraction of CBF-Based Safety Filters for Stable Planar Linear Systems

In this paper we study the dynamical properties induced by safety filters based on control barrier functions (CBFs). These are obtained by minimally modifying a nominal controller with the goal of rendering it safe. We consider the simple setting of a planar linear plant with a stable linear nominal controller and an affine constraint. We show that even in this simple setting, the filtered system can oscillate or diverge while remaining safe. We fully characterize the parameters for which the origin is globally (exponentially or asymptotically) stable. For all other parameters, the filter creates a saddle on the constraint boundary, and we characterize the region of attraction of the origin. For underdamped nominal dynamics, its boundary is an unstable periodic orbit, a homoclinic loop, or the saddle's stable manifold, depending on whether the CBF gain is below, at, or above a homoclinic transition value. A vehicle sideslip example and two three-dimensional examples illustrate the results and what changes beyond planar dynamics.

eess.SY↗

arXiv2026

Vision-Guided Closed-Loop Autonomous Navigation of Light-Driven Janus Microbots

Artificial microbots capable of precise and autonomous navigation could provide versatile platforms for microscale transport, manipulation, and task execution in complex environments. However, reliable navigation becomes increasingly challenging at the micrometre scale, particularly when the surrounding environment changes dynamically. Here, we develop a vision-guided closed-loop navigation framework for light-driven cube-shaped hematite/TiO2 Janus microbots with a characteristic size of approximately 3 μm. Real-time visual localization using a custom-trained YOLOv11 object-detection model is integrated with discretized waypoint-based magnetic steering to enable precise navigation along complex prescribed trajectories, with more than 90% of the measured trajectory deviations remaining below 6 μm across the tested path geometries. The framework further enables arbitrary user-defined free-hand trajectories and autonomous navigation through a physically confined maze. In dynamically changing environments, A* path planning enables real-time path replanning while avoiding multiple independently swimming 5 μm spherical Janus micromotors. Finally, the microbot selectively captures multiple cargo particles and autonomously transports and delivers them to a predefined location while simultaneously avoiding dynamic obstacles. These results establish an integrated platform combining programmable trajectory control, autonomous path planning, real-time environmental adaptation, and task execution within a single light-driven Janus microbot system.

cs.RO↗

arXiv2026

Learning and Forecasting Multiscale Spatiotemporal Epidemic Dynamics with Lagrangian-Eulerian Graph Neural ODEs

Accurate medium- to long-range epidemic forecasting requires extrapolating high-dimensional, heterogeneous and nonstationary dynamics from partial observations. Yet hybrid models often leave implicit which observations inform which epidemic processes and how learned graph interactions modify system state. We introduce LE-EpiGNODE, which replaces generic message passing with process-specific, mechanism-guided neural operators: age-contact information drives Lagrangian infection generation, whereas mobility drives conservative Eulerian population transport within a structure-preserving universal differential equation. We evaluated LE-EpiGNODE across 14 controlled synthetic mechanisms spanning time-modulated, state-dependent, nonlinear and coupled dynamics, three complementary COVID-19 settings, 21 comparator models and five targeted ablations. Forecasting gains tended to become more pronounced at longer horizons; at the longest horizon, LE-EpiGNODE achieved the lowest WMAPE in all three real-world datasets, reducing error by 17.7-40.0% relative to the best baseline. Controlled ablations linked these gains to the Lagrangian-Eulerian decomposition and process-aligned priors, while synthetic ground truth showed that learned operators and induced state fluxes track known generating dynamics. These results support mechanism-organized graph neural dynamics as a route to reliable long-horizon extrapolation with directly testable, process-level mechanistic interpretation beyond trajectory accuracy.

cs.LG↗

arXiv2026

Bayesian Nonparametric Longitudinal Factor Model for Microbiome Counts

Longitudinal microbiome studies provide important insights into the temporal evolution of microbial abundances and dynamic cross-feature dependence. Modeling longitudinal microbiome data is challenging because microbiome counts are highly heterogeneous and zero-inflated, and, moreover, repeated measurements induce within-subject temporal dependence and possibly time-varying cross-feature covariance. To address these challenges, we develop a flexible Bayesian longitudinal factor model using a rounded log-normal latent representation. The mean abundance process begins with a Dirichlet process mixture baseline, which accommodates sparse and multimodal abundance distributions, and subsequently evolves through an autoregressive process. Autoregressive coefficients differ across prespecified biological groups, which are defined by genotypes in our application. A separate autoregressive factor process captures shared cross-feature dependence. The model therefore distinguishes subject--feature mean evolution from shared dependence and yields a closed-form, time-indexed covariance. Simulation studies evaluate dynamic covariance recovery under matched and misspecified data-generating mechanisms. In the longitudinal mouse-gut microbiome application, LODI identifies highly persistent feature-specific abundance trajectories and shared latent factors that distinguish the three genotype groups over time. At week 15, the dominant factor follows the observed WT < ACK/G < HACK/G tumor-severity ordering, with its strongest contributions arising from OTUs classified within Clostridiaceae, Clostridium, and RF39.

stat.ME↗