arXiv preprints from January 1, 2026 through July 20, 2026 — 17:29:15 EST
Large Audio-Language Models (LALMs) have demonstrated strong performance in spoken question answering (QA), with existing evaluations primarily focusing on answer accuracy and robustness to acoustic perturbations. However, such evaluations implicitly assume that spoken inputs remain semantically answerable, an assumption that often...
The long-term dynamical evolution of asteroid families is governed by the interplay between orbital and rotational evolution driven by thermal forces and collision. We aim to observationally trace the rotational evolution of main-belt asteroid families over Gyr timescales. We analyzed rotational properties of 8739 asteroids with spin...
Physics-Informed Neural Networks (PINNs) provide a learning-based framework for solving partial differential equations (PDEs) by embedding governing physical laws into neural network training. In practice, however, their performance is often hindered by limited representational capacity and optimization difficulties caused by...
High-mobility communications, which are crucial for next-generation wireless systems, cause the orthogonal frequency division multiplexing (OFDM) waveform to suffer from strong intercarrier interference (ICI) due to the Doppler effect. In this work, we propose a novel receiver architecture for OFDM that leverages the angular domain to...
Flat-band superconductors provide a regime in which kinetic energy is quenched, so that pairing is governed primarily by interactions and quantum geometry. We investigate characteristic superconducting length scales in all-flat-band systems under the assumptions of time-reversal symmetry and spatially-uniform pairing, focusing on the...
L. Bary-Soroker and R. Shmueli (2026) have given an asymptotic formula for the number of irreducible polynomials over the finite fields $\mathbb F_q$ of $q$ elements, such that their coefficients are perfect squares in $\mathbb F_q$ and also extended this to classes of polynomials with coefficients described by finitely many unions of...
Agentic applications are LLMs that iteratively invoke external tools to accomplish complex tasks. Such tool-based agents are rapidly becoming the dominant paradigm for deploying language models in production. Unlike traditional single-turn inference, agentic workloads chain together multiple LLM calls and tool executions before...
The Lombard effect plays a key role in natural communication, particularly in noisy environments or when addressing hearing-impaired listeners. We present a controllable text-to-speech (TTS) system capable of synthesizing Lombard speech for any speaker without requiring explicit Lombard data during training. Our approach leverages...
Score-based diffusion models generate new samples by learning the score function associated with a diffusion process. While the effectiveness of these models can be theoretically explained using differential equations related to the sampling process, previous work by Song and Ermon (2020) demonstrated that neural networks using...
Self-supervised pretraining in remote sensing is mostly done using mid-spatial resolution (MR) image datasets due to their high availability. Given the release of high-resolution (HR) datasets, we ask how HR datasets can be included in self-supervised pretraining to enhance MR image representation learning and downstream segmentation...
This work investigates the spatial trade-offs arising from the design of the transmit beamformer in a monostatic integrated sensing and communication (ISAC) base station (BS) under bursty traffic, a crucial aspect necessitated by the integration of communication and sensing functionalities in next-generation wireless systems. In this...
Ensuring that large language models (LLMs) respect diverse cultural values is crucial for social equity. However, existing approaches often treat cultural groups as homogeneous and overlook within-group heterogeneity induced by intersecting demographic attributes, leading to unstable behavior under varying persona granularity. We...
At present, neural network-based models, including transformers, struggle to generate memorable and readily comprehensible music from unified and repetitive musical material due to a lack of understanding of musical structure. Consequently, these models are rarely employed by the games industry. It is hypothesised by many scholars...
Integrating deep learning into healthcare enables personalized care but raises trust issues due to model opacity. To improve transparency, we propose a system for dental age estimation from panoramic images that combines an opaque and a transparent method within a natural language generation (NLG) module. This module produces...
We consider linear codes over a field in which the error values are restricted to a subgroup of its unit group. This scenario captures Lee distance codes as well as codes over the Gaussian or Eisenstein integers. Codes correcting restricted errors gained increased attention recently in the context of code-based cryptography. In this...
In this paper, we introduce Mackey functors for a t.d.l.c. group and define the cohomological dimension of this group over the Mackey category. We then compare this dimension to the rational discrete cohomological dimension defined by Castellano and Weigel, as well as to the Bredon cohomological dimension of that t.d.l.c. group with...
We develop a data-driven algorithm for automatically selecting the regularisation parameter in Bayesian inversion under random tree Besov priors. One of the key challenges in Bayesian inversion is the construction of priors that are both expressive and computationally feasible. Random tree Besov priors, introduced in Kekkonen et al....
Laser hardening of metals occurs under the influence of a shock wave, which changes the distribution and density of one-dimensional defects - dislocations. There is a relationship between the density of dislocations, the grain size and the resistance of a single crystal to shear loading. The mechanism of hardening processes continues...
We investigate a nonideal, thermodynamically consistent Brusselator reaction-diffusion (RD) system that explicitly incorporates molecular interactions among species in both the diffusion process and the underlying chemical reaction network. Within this framework, we systematically revisit the Cross-Hohenberg classification of...
Image style transfer aims to integrate the visual patterns of a specific artistic style into a content image while preserving its content structure. Existing methods mainly rely on the generative adversarial network (GAN) or stable diffusion (SD). GAN-based approaches using CNNs or Transformers struggle to jointly capture local and...
Giant Shubnikov-de Haas oscillations (SdHO) with V-shaped minima are experimentally studied in a high-mobility two-dimensional electron gas based on GaAs/AlGaAs heterostructures. A phenomenological model with two parameters (transport momentum relaxation time $τ_{\text{tr}}$ and quantum scattering time $τ_q$) is developed, accurately...
Existing spacecraft rendezvous and docking control methods largely rely on predefined dynamic models and often exhibit limited robustness in realistic on-orbit environments. To address this issue, this paper proposes an Imitation Learning-based spacecraft rendezvous and docking control framework (IL-SRD) that directly learns control...
Large Language Models (LLMs) are increasingly integrated into software engineering workflows, yet current benchmarks provide only coarse performance summaries that obscure the diverse capabilities and limitations of these models. This paper investigates whether LLMs' code-comprehension performance aligns with traditional human-centric...
Immersive spatial audio has become increasingly critical for applications ranging from AR/VR to home entertainment and automotive sound systems. However, existing generative methods remain constrained to low-dimensional formats such as binaural audio and First-Order Ambisonics (FOA). Binaural rendering is inherently limited to...
Electrovac pp--waves in Brinkmann form provide exact Einstein--Maxwell solutions for co--propagating null radiation. Motivated by lensing or scattering, one often ``modulates'' a plane electromagnetic wave by a weak transverse envelope $1+γf(x,y)$. We show that, within the aligned null pp--wave ansatz ($A_v=0$, no $v$--dependence,...