arXiv preprints from January 1, 2026 through July 21, 2026 — 19:16:39 EST
An accurate assessment of L2 English pronunciation is crucial for language learning, as it provides personalized feedback and ensures a fair evaluation of individual progress. However, automated scoring remains challenging due to the complexity of sentence-level fluency, prosody, and completeness. This paper evaluates the zero-shot...
Motivated by DNA storage systems and 3D fingerprinting, this work studies the adversarial torn paper channel with edit errors. This channel first applies at most $t_e$ edit errors (i.e., insertions, deletions, and substitutions) to the transmitted word and then breaks it into $t+1$ fragments at arbitrary positions. In this paper, we...
Interconnect power consumption remains a bottleneck in Deep Neural Network (DNN) accelerators. While ordering data based on '1'-bit counts can mitigate this via reduced switching activity, practical hardware sorting implementations remain underexplored. This work proposes the hardware implementation of a comparison-free sorting unit...
Motion representation plays an important role in video understanding and has many applications including action recognition, robot and autonomous guidance or others. Lately, transformer networks, through their self-attention mechanism capabilities, have proved their efficiency in many applications. In this study, we introduce a new...
We study the free boundary problem for a finite-depth layer of viscous incompressible fluid in arbitrary dimension, modeled by the Stokes or Navier-Stokes equations. In addition to the gravitational field acting in the bulk, the free boundary is acted upon by surface tension and an external stress tensor posited to be in traveling...
Vision-language models (VLMs) are increasingly important in medical applications; however, their evaluation in dermatology remains limited by datasets that focus primarily on image-level classification tasks such as lesion recognition. While valuable for recognition, such datasets cannot assess the full visual understanding, language...
Cosmic voids, the large underdense regions of our Universe, have emerged over the past decade as powerful cosmological laboratories: their simple dynamics, sensitivity to local gravitational effects and cosmic expansion, and ability to span large volumes, make them uniquely suited to test fundamental physics. Fueled by advances in...
We derive the QCD$\times$QED factorization theorem governing the leptonic decay $B^-\toμ^-\barν_μ(γ)$ at all orders in $α_s$ and $α$. Electromagnetic corrections to this decay probe multiple scales, which we disentangle through a sequence of effective field theories (EFTs). The resulting state-of-the-art prediction for the...
GRB~221009A has been the brightest gamma-ray burst (GRB) observed to date, and its afterglow has been characterized with unprecedented detail at TeV energies by LHAASO. Quite puzzlingly, it is also the most energetic GRB known. Among the riddles posed by this mysterious source, however, the sheer energetics are hardly the most...
Models for image representation learning are typically designed for either recognition or generation. Various forms of contrastive learning help models learn to convert images to embeddings that are useful for classification, detection, and segmentation. On the other hand, models can be trained to reconstruct images with pixel-wise,...
Generalizing video matting models to real-world videos remains a significant challenge due to the scarcity of labeled data. To address this, we present Video Mask-to-Matte Model (VideoMaMa) that converts coarse segmentation masks into pixel accurate alpha mattes, by leveraging pretrained video diffusion models. VideoMaMa demonstrates...
The internal composition of sub-Neptunes remains a prominent unresolved question in exoplanetary science. We present a technique to place constraints on envelope mean molecular weight that utilises observations of escaping hydrogen or helium exospheres. This method is based on a simple timescale argument, which states that...
We present Motion 3-to-4, a feed-forward framework for synthesising high-quality 4D dynamic objects from a single monocular video and an optional 3D reference mesh. While recent advances have significantly improved 2D, video, and 3D content generation, 4D synthesis remains difficult due to limited training data and the inherent...
Precision psychiatry aspires to elucidate brain-based biomarkers of psychopathology to bolster disease risk assessment and treatment development. To this end, functional magnetic resonance imaging (fMRI) has helped triangulate brain circuits whose functional features are correlated with or even predictive of forms of psychopathology....
Symbolic systems operate over precise identities: variables denote specific objects, pointers target precise memory locations, and database keys refer to singular records. Neural embeddings generalize by compressing away semantic detail, but this compression creates collision ambiguity: multiple distinct entities can share the same...
We present LightOnOCR-2-1B, a 1B-parameter end-to-end multilingual vision--language model that converts document images (e.g., PDFs) into clean, naturally ordered text without brittle OCR pipelines. Trained on a large-scale, high-quality distillation mix with strong coverage of scans, French documents, and scientific PDFs,...
Videos convey richer information than images or text, capturing both spatial and temporal dynamics. However, most existing video customization methods rely on reference images or task-specific temporal priors, failing to fully exploit the rich spatio-temporal information inherent in videos, thereby limiting flexibility and...
Long chain-of-thought (CoT) trajectories provide rich supervision signals for distilling reasoning from teacher to student LLMs. However, both prior work and our experiments show that trajectories from stronger teachers do not necessarily yield better students, highlighting the importance of data-student suitability in distillation....
An increasing number of discoveries of isolated and quenched dwarf galaxies are challenging the idea that the present-day local environment of low-mass systems is the main determinant of their quenching. We present new Hubble Space Telescope (HST) data of one such system, the dwarf galaxy Canes Venatici C (CVn C). CVn C is a low-mass...
This work investigates a class of non-autonomous $T$-periodic piecewise smooth differential systems and their associated time-$T$ maps. Our main result provides an analytical approach for detecting, within this class of piecewise differential systems, isolated invariant tori associated with normally hyperbolic invariant closed curves...
We present Soft Tail-dropping Adaptive Tokenizer (STAT), a 1D discrete visual tokenizer that adaptively chooses the number of output tokens per image according to its structural complexity and level of detail. STAT encodes an image into a sequence of discrete codes together with per-token keep probabilities. Beyond standard...
Retrieval is being redefined by agentic AI, demanding multimodal reasoning beyond conventional similarity-based paradigms. Composed Image Retrieval (CIR) exemplifies this shift as each query combines a reference image with textual modifications, requiring compositional understanding across modalities. While embedding-based CIR methods...
The rapid expansion of the Fast Radio Burst (FRB) field has been accompanied by a simultaneous growth of FRB conferences. While these meetings are essential for interacting with other researchers and establishing collaborations, many remain only accessible to those with substantial travel funding, flexible schedules, or geographical...
Accurate localization in Orthogonal Frequency Division Multiplexing (OFDM)-based massive Multiple-Input Multiple-Output (MIMO) systems depends critically on phase coherence across subcarriers and antennas. However, practical systems suffer from frequency-dependent and (spatial) antenna-dependent phase offsets, degrading localization...
Reinforcement learning (RL) is essential for enhancing the complex reasoning capabilities of large language models (LLMs). However, existing RL training pipelines are computationally inefficient and resource-intensive, with the rollout phase accounting for over 70% of total training time. Quantized RL training, particularly using FP8...