arXiv preprints from January 1, 2026 through September 23, 2026 — 17:52:34 EST
The rapid advancement of large language models (LLMs) has led to growing interest in using synthetic data to train future models. However, this creates a self-consuming retraining loop, where models are trained on their own outputs and may cause performance drops and induce emerging biases. In real-world applications, previously...
For a compact and connected Lie group $G$, we present an explicit construction of an $\mathbb{S}^1$-gerbe over the differentiable stack $[G/G]$ in the framework of $\mathbb{S}^1$-central extensions of Lie groupoids. This gives a complete proof of the construction outlined earlier by Behrend--Xu--Zhang, together with an explicit proof...
A colloidal monolayer embedded in the bulk of a fluid experiences a "compressible", long-range hydrodynamic interaction which, far from boundaries, leads to a breakdown of Fick's law above a well defined length scale, showing up as anomalous collective diffusion. We here extend the model to study the effect of the hydrodynamic...
Hyperspectral cameras provide numerous advantages in terms of the utility of the data captured. They capture hundreds of data points per sample (pixel) instead of only the few of RGB or multispectral camera systems. Aerial systems sense such data remotely, but the data must be georectified to produce consistent images before analysis....
Sampling is renowned for its privacy amplification in differential privacy (DP), and is often assumed to improve the utility of a DP mechanism by allowing a noise reduction. In this paper, we further show that this last assumption is flawed: When measuring utility at equal privacy levels, sampling as preprocessing consistently yields...
Mixed ionic-electronic conductors host tightly coupled interactions among mobile ions, electronic charges, and the polymer matrix, giving rise to complex multimodal responses spanning electrical, mechanical, and morphological transformations. These materials underpin organic electrochemical transistors (OECTs), which translate such...
In addition to satellite systems, carrier phase positioning (CPP) is gaining attraction also in terrestrial mobile networks, particularly in 5G New Radio evolution toward 6G. One key challenge is to resolve the integer ambiguity problem, as the carrier phase provides only relative position information. This work introduces and studies...
We explore the limitations of using imbalance dynamics as a diagnostic tool for many-body localization (MBL) and show that spatial averaging can mask important microscopic features. Focusing on the strongly disordered regime of the random-field XXZ chain, we use state-of-the-art numerical techniques (Krylov time evolution and full...
We present new initial data codes for constructing stationary, axisymmetric equilibrium models of differentially rotating neutron stars in full general relativity within the Frankfurt University/KADATH (FUKA) suite of initial data codes. FUKA leverages the KADATH spectral library to solve the Einstein equations under the assumption of...
Neuro-symbolic reasoning systems face fundamental challenges in maintaining semantic coherence while satisfying physical and logical constraints. Building upon our previous work on Ontology Neural Networks, we present an enhanced framework that integrates topological conditioning with gradient stabilization mechanisms. The approach...
Chain-of-thought (CoT) reasoning has emerged as a powerful tool for multimodal large language models on video understanding tasks. However, its necessity and advantages over direct answering remain underexplored. In this paper, we first demonstrate that for RL-trained video models, direct answering often matches or even surpasses CoT...
Spatial-Temporal Graph (STG) forecasting on large-scale networks has garnered significant attention. However, existing models predominantly focus on short-horizon predictions and suffer from notorious computational costs and memory consumption when scaling to long-horizon predictions and large graphs. Targeting the above challenges,...
The graph alignment problem aims to identify the vertex correspondence between two correlated graphs. Most existing studies focus on the scenario in which the two graphs share the same vertex set. However, in many real-world applications, such as computer vision, social network analysis, and bioinformatics, the task often involves...
Embodied question answering (EQA) in 3D environments often requires collecting context that is distributed across multiple viewpoints and partially occluded. However, most recent vision--language models (VLMs) are constrained to a fixed and finite set of input views, which limits their ability to acquire question-relevant context at...
Existing long-term personalized dialogue systems struggle to reconcile unbounded interaction streams with finite context constraints, often succumbing to memory noise accumulation, reasoning degradation, and persona inconsistency. To address these challenges, this paper proposes Inside Out, a framework that utilizes a globally...
Natural Language Inference (NLI) has been an important task for evaluating language models for Natural Language Understanding, but the logical properties of the task are poorly understood and often mischaracterized. Understanding the notion of inference captured by NLI is key to interpreting model performance on the task. In this...
We show that a general class of social impact models with higher-order interactions on hypergraphs can be exactly reduced to an equivalent model with pairwise interactions on a weighted projected network. This reduction is made by a mapping that preserves the microscopic probabilities of changing the state of the nodes. As a...
The Higgs boson (H) decay width is determined from the ratio of off- and on-shell production of H $\to$ WW $\to$ e$νμν$ using proton-proton collision data corresponding to an integrated luminosity of 138 fb$^{-1}$ collected at $\sqrt{s}$ = 13 TeV by the CMS experiment at the Large Hadron Collider. The off-shell signal strength is...
Large Language Models (LLMs) for complex reasoning is often hindered by high computational costs and latency, while resource-efficient Small Language Models (SLMs) typically lack the necessary reasoning capacity. Existing collaborative approaches, such as cascading or routing, operate at a coarse granularity by offloading entire...
Detecting and counting copies of permutation patterns are fundamental algorithmic problems, with applications in the analysis of rankings, nonparametric statistics, and property testing tasks such as independence and quasirandomness testing. From an algorithmic perspective, there is a sharp difference in complexity between detecting...
This paper investigates the fundamental communication--sensing tradeoffs of uplink dual-functional integrated sensing and communication (ISAC) multiple access under finite blocklength (FBL) constraints. Unlike conventional asymptotic analyses, we explicitly account for the limitations under FBL constraints imposed by short packets and...
We consider random integer partitions $λ$ that follow the Poissonized Plancherel measure of parameter $t^2$. Using Riemann$-$Hilbert techniques, we establish the asymptotics of the multiplicative averages $$Q(t,s)=\mathbb{E} \left[ \prod_{i\geq 1} \left(1+\mathrm{e}^{η(λ_i-i+\frac{1}{2}-s)}\right)^{-1} \right] $$ for fixed $η>0$ in...
Sabam, a Belgian collective rights management organisation, wanted an internet access provider and a social network site to install a filter system to enforce copyrights. In two recent judgments, the Court of Justice of the European Union decided that the social network site and the internet access provider cannot be required to...
Document Question Answering (DocQA) focuses on answering questions grounded in given documents, yet existing DocQA agents lack effective tool utilization and largely rely on closed-source models. In this work, we introduce DocDancer, an end-to-end trained open-source Doc agent. We formulate DocQA as an information-seeking problem and...
Diagrams are crucial for communicating complex information, yet creating and modifying them remains a labor-intensive task. We present GenAI-DrawIO-Creator, a novel framework that leverages Large Language Models (LLMs) to automate diagram generation and manipulation in the structured XML format used by draw.io. Our system integrates...