arXiv preprints from January 1, 2026 through September 26, 2026 — 02:46:09 EST
Large audio-language models have made rapid progress in recognizing what is present in an audio clip, but spatial audio-language understanding still lacks a clear task interface. A model must also decide where sound events occur, which semantic and spatial attributes belong to the same auditory object, how multiple objects are...
We investigate a unitary matrix model with a complex potential with Fisher-Hartwig singularities. We show that the model exhibits finite-$N$ phase transitions. The order of the phase transition is coupling-dependent. At large-$N$, these transitions are replaced by third-order Gross-Witten-Wadia transitions between multiple ungapped...
We introduce a new basis of the group algebra of the symmetric group, built using the left-to-right minima sets of permutations. We show that on this basis, the descent algebra acts by triangular operators, thus making it an analogue of a cellular basis. The proof involves Dynkin elements (nested commutators) of the free algebra and...
The structure of continuous Hopfield networks is revisited from a system-theoretic point of view. After adopting a novel electrical network interpretation involving nonlinear capacitors, it is shown that Hopfield networks admit a port-Hamiltonian formulation provided an extra passivity condition is satisfied. Subsequently it is shown...
Current Large Language Model reasoning systems process queries independently, discarding valuable cross-instance signals such as shared reasoning patterns and consistency constraints. We introduce Batch-of-Thought (BoT), a training-free method that processes related queries jointly to enable cross-instance learning. By performing...
Trust between entities in any scenario without a trusted third party is very difficult, and trust is exactly what blockchain aims to bring into the digital world with its basic features. Many applications are moving to blockchain adoption, enabling users to work in a trustworthy manner. The early generations of blockchain have a...
Robots deployed in dynamic environments must remain safe even when key physical parameters are uncertain or change over time. We propose Parameter-Robust Model Predictive Path Integral (PRMPPI) control, a framework that integrates online parameter learning with probabilistic safety constraints. PRMPPI maintains a particle-based belief...
Synthetic Data Generation (SDG) can be used to facilitate privacy-preserving data sharing. However, most existing research focuses on privacy attacks where the adversary is the recipient of the released synthetic data and attempts to infer sensitive information from it. This study investigates quality degradation attacks initiated by...
Helium accumulation in structural ceramics used in nuclear, fusion, and aerospace systems causes swelling, cracking, and early failure, yet controlling this damage has remained elusive. Here, we introduce defect landscape engineering, the deliberate creation of vacancy clusters prior to helium exposure, as a general strategy to...
With the rapid development of e-commerce and digital fashion, image-based virtual try-on (VTON) has attracted increasing attention. However, existing VTON models often suffer from artifacts such as garment distortion and body inconsistency, highlighting the need for reliable quality evaluation of VTON-generated images. To this end, we...
Advanced speech synthesis technologies have enabled highly realistic speech generation, posing security risks that motivate research into audio deepfake detection (ADD). While state space models (SSMs) offer linear complexity, pure causal SSMs architectures often struggle with the content-based retrieval required to capture global...
Accurate Travel Time Estimation (TTE) is critical for ride-hailing platforms, where errors directly impact user experience and operational efficiency. While existing production systems excel at holistic route-level dependency modeling, they struggle to capture city-scale traffic dynamics and long-tail scenarios, leading to unreliable...
Identifying the origin of enhanced charging performance in many-body quantum batteries remains a central challenge in quantum thermodynamics. It is unclear whether improvements in stored energy and instantaneous charging power stem from genuinely quantum correlations, such as entanglement, or from coherent collective dynamics, in...
In our work, we present a calculation based on an exact, non-approximated wave equation, as described by the Teukolsky equation, for a Taub-NUT-de Sitter spacetime. We observe the scattering of a monochromatic point source by an observer located at a greater distance from the black hole, and examine the wave-optical images by a single...
Do transformers learn like brains? A key challenge in addressing this question is that transformers and brains are trained on fundamentally different data. Brains are initially "trained" on prenatal sensory experiences (e.g., retinal waves), whereas transformers are typically trained on large datasets that are not biologically...
Wildfires are highly imbalanced natural hazards in both space and severity, making the prediction of extreme events particularly challenging. In this work, we introduce the first ordinal classification framework for forecasting wildfire severity levels directly aligned with operational decision-making in France. Our study investigates...
We have examined the cosmological actions of LRS (Locally Rationally Symmetric) Bianchi type-I universe model in $f(R,T^ψ)$ gravity. For this, we have estimated the Hubble parameter, the effective equation of state parameter ($ω^{eff}$), and the potential of the scalar field as a function of time using equation $H = W(ψ)$. The...
Emotion is a central dimension of spoken communication, yet, we still lack a mechanistic account of how modern large audio-language models (LALMs) encode it internally. We present the first neuron-level interpretability study of emotion-sensitive neurons (ESNs) in LALMs and provide causal evidence that such units exist in...
We present a novel, regression-based method for artistically styling images. Unlike recent neural style transfer or diffusion-based approaches, our method allows for explicit control over the stroke composition and level of detail in the rendered image through the use of an extensible set of stroke patches. The stroke patch sets are...
This paper presents the first probabilistic Digital Twin of operational en route airspace, developed for the London Area Control Centre. The Digital Twin is intended to support the development and rigorous human-in-the-loop evaluation of AI agents for Air Traffic Control (ATC), providing a virtual representation of real-world airspace...
Generative joint source-channel coding (GJSCC) has emerged as a new Deep JSCC paradigm for achieving high-fidelity and robust image transmission under extreme wireless channel conditions, such as ultra-low bandwidth and low signal-to-noise ratio. Recent studies commonly adopt diffusion models as generative decoders, but they...
The reasoning ability of large language models (LLMs) can be unleashed with reinforcement learning (RL) (OpenAI, 2024; DeepSeek-AI et al., 2025a; Zeng et al., 2025). The success of existing RL attempts in LLMs usually rely on high-quality samples of large volumes. In this paper, we challenge conventional assumptions about data...
We present a comprehensive phenomenological analysis of the Radion in the Randall-Sundrum model, focusing on its production via light-by-light scattering in ultra-peripheral proton-proton collisions at the LHC. We provide a consistent derivation of the effective couplings to Standard Model fields, clarifying the normalization of the...
Motivated by a connection to the infinite Ginibre point process, decoupled random walks were introduced in a recent article Alsmeyer, Iksanov and Kabluchko (2025). The decoupled random walk is a sequence of independent random variables, in which the $n$th variable has the same distribution as the position at time $n$ of a standard...
We develop a structure-aware reinforcement learning (RL) approach for delay- and energy-aware flow allocation in 5G User Plane Functions (UPFs). We consider a dynamic system with $K$ heterogeneous UPFs of varying capacities that handle stochastic arrivals of $M$ flow types, each with distinct rate requirements. We model the system as...