Search arXiv⌕ Search

SEARCH · Search arXiv

Search Search arXiv

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

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,369 records · Page 76Linked to original sources
arXiv2026

Monitorability Is Trainable: Post-Training for Chain-of-Thought Monitorability

Chain-of-thought (CoT) monitoring is one of the few practical tools available for overseeing capable AI systems, but it is fragile: a model's reasoning is only useful to a monitor if it remains coupled to what the model actually does. This coupling can be degraded during training through, for example, CoT compression, optimization against a monitor, and out-of-context generalization from synthetic documents. In this work, we propose an unsupervised post-training objective based on maximizing mutual information between CoT and the output for improving CoT monitorability. Across three model organisms with degraded CoT, it recovers most or all of the lost monitorability. Applied to undegraded policies, it further improves judged monitorability, improves prediction of unsafe behavior by CoT monitors and activation probes, and increases the prevalence of consequential reasoning. These results suggest that monitorability should be treated as a trainable attribute of the policy. A dedicated monitorability post-training stage can extend current practices that protect CoT from direct optimization pressure.

cs.LG↗
arXiv2026

Coronal spectroscopy - Past, Present and Future

After the beginning of photography of solar corona during the eclipses in the year 1860, efforts began to carry out spectroscopic observations of solar corona. During the total solar eclipse of 1868 August 18 at Guntur, Madras state, the observation of emission line 5875Å by Janssen lead to the discovery of Helium. Next year, during the eclipse of 1869 August 7, Charles Young and William Harkness discovered green emission line around 5310Å (later measured as 5303Å). It was interpreted as due to Coronium, an unknown element. The identification of this green line in 1939 due to [Fe xiv] ion showed the corona to be hot at 10^6 K. High resolution spectroscopic observations indicated that width of emission lines to be more than expected. The excess line width was interpreted in terms of existence of non-thermal velocity. Recent spectroscopic observations have indicated a complex nature of variation of emission line parameters with height above the corona. The spectroscopy of solar corona is also performed in other wavelengths, such as UV, EUV, X-rays, visible, infrared, and radio part of the spectrum. Here, we discuss the past spectroscopic observations of solar corona, present observations with various instruments, such as Visible Emission Line Coronagraph onboard Aditya-L1, and need for the spectroscopic observations in future.

astro-ph.SR↗
arXiv2026

Two Sides of the Same Proof: Extraction and Projection in Choreographic Programming

A new approach to the theory of choreographic languages has recently been proposed, based on the interpretation of choreographies with finite behaviors as proofs in an extension of linear logic. This paper further develops this approach to cyclic proofs to be capable of representing infinite behaviors, and prove a completeness result for choreographies with respect to the class of lock-free networks. We show that in a choreographies-as-proofs correspondence, the proof equivalence induced by rule permutations corresponds to bisimilarity, and that, under adequate assumption on the procedure definitions, the proof search procedure has a finite search space, and can be used to extract a choreography from a network.

cs.LO↗
arXiv2026

Spanning bipartite quadrangulations in even triangulations of real projective spaces

We study spanning quadrangulations of even triangulations of real projective spaces, using the higher-dimensional notion of quadrangulation introduced by Kaiser and Stehl\'ık. We prove that every even triangulation of $\RP^d$, where $d\ge 2$, admits a spanning bipartite quadrangulation. Moreover, such a triangulation admits a spanning non-bipartite quadrangulation if and only if it is not balanced; if it is balanced, every spanning quadrangulation is bipartite. Our proofs use coloring monodromy: its orbits yield bipartite quadrangulations, while a covering-space construction gives non-bipartite ones. %In dimension three, a bipartite quadrangulation can always be chosen to induce $K_{2,2}$ on each tetrahedron. More generally, we establish a sufficient condition for the existence of spanning bipartite quadrangulations on closed manifolds in terms of the monodromy action, and derive a consequence involving the fundamental group.

math.CO↗
arXiv2026

Learning to Measure and Diagnose: A Clinical-Dictionary Radiomic Bottleneck Model for Thyroid Nodule Classification

Thyroid ultrasound supports the assessment of nodule malignancy, but classification predictions often lack evidence linked to familiar clinical descriptions. Radiomics provides explicitly defined measurements that can support such explanations through a clinical dictionary. However, using these measurements as interpretable evidence requires reliable estimation from ultrasound images and preservation of measurement fidelity when optimizing classification performance. We propose a clinical-dictionary radiomic bottleneck model for thyroid nodule classification. Within this model, Family-Aware Radiomic Learning (FARL) estimates measurements with balanced supervision across feature families. Fidelity-Guided Bottleneck Modeling (FGBM) constructs a fixed bottleneck, followed by joint optimization of measurement estimation and classification under fidelity guidance. Decision evidence decomposition then traces feature contributions through a linear classifier and organizes them by clinical dictionary categories. Experiments on ThyroidXL and TN3K show competitive classification performance, while case studies illustrate feature contributions for benign and malignant nodules. These findings indicate interpretable classification with inspectable quantitative evidence informed by clinical knowledge. Code is available online.

cs.CV↗
arXiv2026

APED-CXR: Adaptive Privileged Evidence Distillation for Chest X-Ray Grounding with Image-Only Inference

Compact chest X-ray (CXR) models that localize findings suit deployment, but box annotations are limited and report-aware multimodal models are expensive at inference. We introduce APED-CXR, which distills report-informed spatial evidence from a frozen multimodal teacher into a student with 11.1M parameters that predicts diseases and boxes from shared evidence maps and requires the image alone at test time. During training, the teacher uses the radiograph, report, and queried class to evaluate regions proposed from the student's current evidence maps, and its support scores serve as training targets for the student's pooled evidence scores within these regions. Trained on PadChest and evaluated on held-out PadChest-GR, refreshing the student's proposals for teacher evaluation improves box-level localization F1 from 0.302 with fixed warm-up proposals to 0.338 (IoU >= 0.5; gain 0.036, 95% CI [0.022, 0.049]), while maintaining diagnostic AUC. This gain depends on the teacher checking current proposals: with equally many random boxes, a fixed dense box set, teacher-generated boxes, or report rules instead of the teacher, F1 stays near the fixed-proposal level, and proposals two refreshes old recover only part of the gain. On NIH ChestX-ray14, a secondary test set, box recall also rises from 0.423 to 0.447.

cs.CV↗
arXiv2026

Engineering Requirements for Generative AI Agents: The BARD Process Theory

How teams engineer requirements for systems built around Generative AI (GenAI) Agents remains empirically uncharted. Existing Requirements Engineering (RE) research mainly addresses AI and ML-based systems broadly but provides no process account for agentic systems, whose non-deterministic behavior cannot be anticipated at design time. To understand how teams actually proceed, we applied Socio-Technical Grounded Theory across 36 semi-structured interviews with practitioners, interleaving data collection and analysis so that emerging findings drove our theoretical sampling and shaped successive rounds of coding. The result is Behavioral Agent Requirement Discovery (BARD), the process theory of RE for GenAI Agents. BARD holds that requirements are not fixed up front but discovered through a recurring cycle of four activities: teams scope the need, build the agent, observe its behavior, and revise what they ask for. Through this four-phase loop, the requirement evolves from a fixed specification towards behavioral and evaluation criteria. Four tensions drive transitions between phases, and two conditions frame the loop. BARD provides researchers with a framework for iteratively refining RE workflows, while giving practitioners a structured basis to embed into their development processes.

cs.SE↗
arXiv2026

Delay-Doppler Domain Aided OFDM (DD-a-OFDM): Pulsone Pilot and Cross-Domain Iterative Receiver Design

In high-mobility channels, the widely adopted orthogonal frequency division multiplexing (OFDM) modulation waveform suffers from significant performance degradation due to channel variation and severe inter-carrier interference. Although delay-Doppler (DD) domain multicarrier (DDMC) modulation enjoys a higher channel estimation (CE) accuracy via exploiting the inherent DD-domain channel stability, sparsity, and separability, its high detection complexity limits its practical deployment, making OFDM still a promising waveform candidate for the sixth-generation (6G) cellular networks. To simultaneously harness the complementary strengths of DDMC and OFDM and mitigate their respective limitations, this paper proposes a DD-domain aided OFDM (DD-a-OFDM) framework, including a pulsone pilot (PP) design and a cross-domain iterative receiver design, to enhance the CE and detection performance of OFDM in high-mobility channels. Specifically, a pulsone pilot sequence is periodically inserted before the cyclic prefix (CP) for each OFDM symbol in the time-domain, which is equivalent to a single DD-domain pilot impulse and enables accurate DD-domain CE. Then, we propose a cross-domain iterative receiver in which the time-frequency (TF) domain detection results are transformed into the DD-domain for interference cancellation and CE; in turn, the DD-domain estimated channel is transformed into the TF-domain for channel reconstruction and symbol detection. The proposed cross-domain iterative receiver achieves a higher CE accuracy and detection performance with a much lower pilot overhead, compared to the conventional OFDM system. Simulation results validate the fast convergence and superior performance of the proposed DD-a-OFDM scheme and demonstrate that it maintains a robust CE and detection performance even in high-mobility channels.

eess.SP↗
arXiv2026

Weighted Risk Aggregation under Long-Memory Dependence: Pathwise Fluctuations of Exposure-Weighted Losses

We model insurance loss in each period as deterministic exposure multiplied by a nonnegative function of a long-memory Gaussian sequence. A weighted law of the iterated logarithm yields the exact almost-sure scale and sharp constant of centered cumulative loss. Loss per unit of cumulative exposure converges almost surely to its mean, with an exact error scale common to all admissible exposure profiles; the sharp constant depends on the exposure exponent. Lognormal and quadratic examples, of Hermite ranks one and two, and a simulation illustrate the results. A known spectral formula expresses the quadratic coefficient through a leading eigenvalue.

math.PR↗
arXiv2026

On Almost MDS codes holding $(n-k-1)$-designs and an open problem on Witt design

Let $\mathcal{C}$ be an almost maximum distance separable (AMDS) code over the finite fields with parameters $[n,k,n-k]$ whose minimum-weight codewords support an $(n-k-1)$-$(n,n-k,λ)$ design. We first determine the parameters of the dual code $\mathcal{C}^{\bot}$. We also characterize the design and its complementary design supported by the minimum weight codewords of $\mathcal{C}^{\bot}$. In particular, we present explicit formulas for the weight enumerators of all subset codes of $\mathcal{C}^{\bot}$ with length at least $k+1$. These formulas are completely determined by the cardinality of the subset, the dimension, together with the parameter $λ$, and are independent of the specific choice of subsets. As an application, we answer an open problem posed by Wang, Tang and Ding in 2023 by proving that the $3$-design supported by the minimum-weight codewords in their proposed cyclic dual code is isomorphic to the complementary design of the Witt spherical geometry design.

cs.IT↗
arXiv2026

Quantum criticality in a two-component two-dimensional nematic metal

Nematic fluctuations alter the low-energy electronic behavior and reshape the mean-field phase diagram. We study a two-dimensional nematic metal with a two-component internal degree of freedom. Near the quantum critical point (QCP), longitudinal nematic fluctuations generate the non-Fermi-liquid self-energy $\propto|ω|^{2/3}$, whereas transverse fluctuations give a subleading logarithmic, marginal-Fermi-liquid correction. Nevertheless, the transverse sector dominates the correction to the phase diagram through a quantum order-by-disorder mechanism: the emergent nematic background reshapes the electronic spectrum, alters the decay of nematic bosons, and produces a negative singular free-energy correction. This correction reconstructs the phase diagram, drives the transition first order, and is described by a scaling function whose two asymptotic limits define two critical regimes with free-energy corrections scaling as $R^{8/3}$ and $R^3$, respectively, where $R$ is the order-parameter magnitude; their powers follow from the competition between the characteristic momentum of normal-state nematic fluctuations and that of the ordered-state background. These two regimes are experimentally distinguishable through nematic-susceptibility hysteresis measurements. The corrected phase diagram shows that the enlarged nodal $α$ phase masks the otherwise bare QCP, making it inaccessible; consequently, the logarithmic self-energy correction is by no means negligible at low energies.

cond-mat.str-el↗
arXiv2026

LEAF-Mamba: Feature Adaptation and Fusion for Whole Slide Image Analysis

Whole slide image (WSI) analysis enables computational diagnosis and prognosis, but gigapixel images and limited annotations make slide-level learning challenging. Multiple instance learning addresses this task by aggregating patch representations from frozen pretrained encoders. Recently, Mamba has enabled efficient modeling of long patch sequences. However, task supervision alone does not explicitly constrain information recovery during feature adaptation, while Mamba couples local processing with long-range propagation rather than retaining separate representations for fusion. Motivated by these challenges, we introduce LEAF-Mamba, a framework that jointly addresses latent evidence adaptation and fusion for frozen-feature WSI analysis. A reconstruction-regularized adapter learns compact, task-specific representations while reconstructing the input features. Parallel Mamba and convolutional streams then capture long-range dependencies and sequence-local interactions, with an input-dependent gate modulating their joint representation before slide-level pooling. Experiments across feature extractors demonstrate improved average performance over strong MIL baselines in both diagnostic classification and survival prediction. Component analyses further support the contributions of adaptation and fusion. These findings highlight the value of jointly designing feature transformation and contextual aggregation to translate pretrained representations into effective slide-level predictions.

cs.CV↗
arXiv2026

Prompt Injection Through Forensic Evidence: A Taxonomy and Proof of Concept of Anti-Forensic Attacks on LLM-Assisted Digital Investigations

Large language models (LLMs) are entering digital forensics as examination assistants and autonomous agents. Digital forensics inverts the usual setting for indirect prompt injection: the evidence under analysis is attacker-controlled by definition, and the examiner is obliged to read it closely. This makes the reasoning layer a new anti-forensic target, next to the classical ones of data and tools. Classical anti-forensic taxonomies do not cover it, and prompt injection defences do not anticipate it. We extend the anti-forensic model to reflect the manipulation of reasoning, and present a taxonomy of attacks carried through adversarial evidence: an attacker selects an objective, a technique, a payload location, and a concealment method. We observe a perception gap, i.e., the fact that deterministic tools, the human examiner, and the model perceive the same evidence differently, which is what lets an injection reach the model unseen by the examiner. We demonstrate three attacks against an LLM-assisted analysis pipeline, changing a reported conclusion, writing attacker-chosen content to the analyst's workstation, and causing outbound network traffic, each driven only by text planted on the seized device. Mitigations to these attacks are also proposed.

cs.CR↗
arXiv2026

Metamodel-Guided LLM Decoding for Consistent DSL Generation

Large language models (LLMs) are increasingly used to generate software engineering artifacts written in domain-specific languages (DSLs). Correct DSL instances usually must comply with not only the concrete syntax but also the abstract syntax defined by a metamodel augmented with semantic well-formedness constraints. Existing constrained decoding approaches typically guarantee compliance only with concrete syntax or with a limited set of semantic constraints expressed over the generated text, making metamodel-level constraint consistency difficult to enforce during generation. We present MeMoDecode: a decoding framework that enforces a DSL's concrete syntax, abstract syntax, and well-formedness constraints jointly. Specifically, whenever the LLM generates a complete metamodel concept, MeMoDecode incrementally maps the current output to a partial model and verifies whether it can still be completed into a consistent instance. We combine MeMoDecode with a simple retry mechanism that resumes decoding from the most recent completable snapshot whenever an inconsistency is detected. Under a sound backend verifier, MeMoDecode never performs worse than grammar-constrained decoding on alignment metrics satisfying natural assumptions. Across four DSLs and six open-weight LLMs ranging from 0.6B to 70B parameters, MeMoDecode achieves the highest consistency, improving consistency by 23% on average over unconstrained decoding and by 8% on average over grammar-constrained decoding. It also improves benchmark performance while incurring only modest runtime overhead. These results show that metamodel-guided decoding is an effective approach for enforcing DSL consistency during decoding across diverse DSLs and LLM scales.

cs.SE↗
arXiv2026

On Lusztig Sheaves for Tame Quivers

Let $Q$ be a connected quiver of tame type, and let $U_v^-$ be the positive part of the corresponding quantized affine algebra. In Lusztig's construction of the canonical basis of $U_v^-$, the simple perverse sheaves occurring, up to shift, as direct summands of the flag sheaf complexes are given by intersection cohomology complexes $\operatorname{IC}(X(c,m),\mathcal{L}_χ)$. We study the restrictions of these intersection cohomology complexes to the locally closed subsets $Y(c,m)$ obtained by fixing the preprojective, non-homogeneous regular, and preinjective summands in the category $\operatorname{rep} Q$. We prove pointwise purity on the loci where the homogeneous regular part is a direct sum of pairwise non-isomorphic quasi-simple modules, and establish a semisimple decomposition of the restrictions to the corresponding larger strata. The graded multiplicities of the local systems associated with irreducible characters of symmetric groups determine the transition coefficients from the canonical basis to the PBW basis constructed via quiver representations. This gives a geometric explanation of the positivity of these coefficients and the unitriangularity of the transition matrix.

math.RT↗
arXiv2026

Bridging Foundational and Agentic Capabilities of LLMs

Large language models (LLMs) are increasingly deployed as agents, yet it remains unclear how their foundational capabilities translate into agentic capabilities. End-to-end scores leave unresolved whether low benchmark capability or weak deployment drives a failure, limiting actionable diagnosis. We bridge foundational and agentic capabilities through an operation-level measurement framework. We represent each model with a foundational profile derived from capability benchmarks and an operation profile derived from interaction trajectories, and define capability realization gaps as deviations of observed operation performance from predictions based on foundational capabilities. Experiments on fifteen model services in $τ^2$-Bench show that foundational profiles only partially predict agentic operations. Models with comparable profiles can differ substantially in execution, and these gaps remain consistent across task subsets. Operation profiles add information and substantially improve task-score prediction. The framework locates execution departures at the operation level to inform model development and agent evaluation. We will release our code and analysis artifacts.

cs.CL↗
arXiv2026

Choice principles in countable partition models

We study the countable partition model over a set of atoms of arbitrary uncountable cardinality $κ$. Its supports are partitions of the atoms into countably many blocks, and a permutation fixes a support when it preserves each block setwise. This model was proposed by Bruce and studied by Banerjee. We prove that dependent choice holds for every $κ$, although the filter of subgroups is not closed under countable intersections. Choice for well-orderable families holds exactly when $κ<\aleph_{ω_1}$; when $κ\ge\aleph_{ω_1}$, an $ω_1$-indexed family witnesses its failure. The Hartogs and Lindenbaum numbers of the set of atoms are both $ω_1$. The proofs use permutations matching blocks of equal cardinality and establish shift completeness of the filter. Together with Banerjee's failure of $\mathsf{DC}_{ω_1}$, these results determine which dependent choice principles hold in the model.

math.LO↗