The paper introduces Murmur2Vec, a lightweight, alignment‑free embedding that uses k‑mer counts hashed with MurmurHash to create a compact representation for biological sequences. It provides a full theoretical analysis, including bias/variance formulas, a Johnson–Lindenstrauss‑style concentration bound, and an excess‑risk bound that clarifies the trade‑off between hash‑table size and classifier performance. Empirically, Murmur2Vec matches or surpasses a fine‑tuned 650M‑parameter ESM‑2 protein language model across several classification tasks, including SARS‑CoV‑2 spike lineage and HIV‑1 Env subtype identification.
By Sarwan Ali, Taslim Murad, Imdadullah Khan, Safi Faizullah
The paper introduces DP-Muon, a differentially private optimization method that incorporates matrix‑orthogonalized momentum. It employs standard global per‑example clipping and releases a single Gaussian‑noised gradient per step, with matrix and auxiliary updates treated as post‑processing. The authors analyze the mean distortion introduced when fresh Gaussian noise passes through a nonlinear matrix map, deriving exact Gaussian heat identities and showing that for a smooth Newton‑Schulz map, the conditional output bias is reduced from second to fourth order in the noise scale. They also establish matrix‑block stationarity bounds, quantify orthogonalization error, and provide criteria for improving the upper bound, while a separate inequality captures the impact of auxiliary Adam updates. Experiments on GPT‑2 at various privacy targets demonstrate that DP‑Muon configurations outperform Adam baselines in test negative log‑likelihood.
By Jihwan Kim, Chenglin Fan
The paper investigates how transformer-based models and traditional feature-based models capture readability signals across five languages using the ReadMe++ dataset. By applying SHAP to identify key features for traditional classifiers and then probing XLM‑R and language‑specific encoders with TCAV, the authors find that transformers recover surface‑length, syntactic, and lexical‑diversity cues and reflect the CEFR ordinal structure, though alignment varies by model family, language, and layer. The study highlights that high linear separability does not guarantee directional influence, cautioning against overreliance on linear probing for readability features.
By Joshua Wong, Chris Tanner
OmniHallu is a unified framework for detecting hallucinations in multimodal large language models across both comprehension and generation tasks involving image, video, and audio modalities. It introduces OmniHallu-Bench, a 10,000-sample benchmark with claim-level human annotations for six cross-modal tasks (I2T, V2T, A2T, T2I, T2V, T2A). The system uses a multi‑agent architecture that decomposes outputs into atomic claims, verifies them with modality‑specific experts, and aggregates evidence through structured reasoning, while a preference‑optimized verifier reduces expert calls by 66% with minimal performance loss.
By Jianjiang Yang, Peihang Li, Shanqing Xu, Mengchen Qian, Lu Zhang, Meng Luo
The paper presents the first controlled study of Whisper adaptation for Greek Automatic Lyric Transcription (ALT), addressing challenges such as melodic variability, rhythmic irregularity, and accompaniment interference. It explores model scaling, multitask training with transcribe-translate ratios, and a two‑stage speech‑to‑singing adaptation, while curating a segment‑level aligned singing dataset from the Greek Audio Dataset. Results show that larger models consistently improve performance, multitask learning benefits smaller models, and the two‑stage adaptation achieves a 27.2% WER, establishing the first Greek ALT benchmark.
By Maria Frangiadaki, Dimitrios Damianos, Kosmas Kritsis, Vassilis Katsouros
The Eloquence team presents three methods for the Interspeech 2026 MLC‑SLM Task 2, a multilingual MCQA challenge covering 21 languages. They fine‑tune Voxtral‑Mini‑3B with LoRA and data augmentation, achieving 0.72 macro‑accuracy; they use multimodal in‑context learning on Voxtral‑24B to correct label bias, reaching 0.81; and they deploy a training‑free retrieval system with a voice‑anchored memory, scoring 0.68. All approaches surpass the official baseline.
By Jordi Luque, Lorenzo Concina, Marco Matassoni, Alessio Brutti, Filippo Vella
The paper surveys how large language models (LLMs) are being applied in legal tasks such as judgement prediction, document analysis, and drafting. It reviews the benefits of automation while highlighting legal challenges like privacy, bias, and explainability. The authors also discuss data resources for legal domain specialization and outline future research directions.
By Zhongxiang Sun
The paper introduces Hikari, a policy‑free, end‑to‑end model that performs simultaneous speech‑to‑text translation and streaming transcription. It employs a Decoder Time Dilation mechanism to mitigate overuse of WAIT tokens during training and a supervised fine‑tuning strategy that helps the model recover from delays, improving the quality‑latency trade‑off. Despite its modest size, Hikari achieves competitive translation quality at consistently low latency, outperforming larger published IWSLT 2026 submissions and proprietary API systems on en‑ja, en‑de, and en‑ru tasks.
By Roman Koshkin, Jeon Haesung, Lianbo Liu, Hao Shi, Mengjie Zhao, Yusuke Fujita, Yui Sudo
The study examined whether brain-language model alignment reflects shared computational mechanisms or merely stable lexical‑semantic correspondences. Using whole‑brain encoding across Mandarin, English, and French, transformer representations predicted activity in a distributed network that overlapped across languages and remained stable across layers. Contextual embeddings and measures of prediction or compression did not outperform static lexical embeddings, suggesting that alignment is robust but not informative about shared computational processes.
By Ni Yang, Rui He, Philipp Homan, Iris Sommer, Davide Staub, Wolfram Hinzen
The paper introduces a new framework for learning continuous-time diffeomorphic image registration by modeling a non-autonomous ODE as a two-parameter flow map. By enforcing cocycle consistency, the method learns flow maps without time discretization or velocity integration during training, enabling efficient inference with few compositions. Experiments on nine datasets show consistent alignment improvements, including a 2.1% Dice gain on brain MRI, 12% TRE reduction on lung CT, and 2.6% Dice improvement on cardiac MRI and ultrasound.
By Mohammadjavad Matinkia, Nilanjan Ray
The paper presents M2Heat, a physics-inspired framework for fusing hyperspectral and LiDAR data in land‑cover classification. It introduces a visual heat conduction module (vHeat) and Frequency Value Embeddings (FVEs) to model anisotropic information flow, capturing global dependencies with sub‑quadratic complexity. Combined with a Cross‑Frequency Fusion (CFF) module, M2Heat delivers discriminative, robust features and achieves competitive performance on Trento, Houston2013, and Augsburg benchmarks while offering an interpretable heat‑conduction perspective.
By Kan Wei, Jiahui Cui, Jing Yao, Xinyu Zhao, Lei Wang, Pedram Ghamisi
Brain-PACE is a deep Siamese MRI framework that directly estimates the pace of structural brain ageing from paired T1‑weighted MRI scans, extending the LILAC model with spatial attention, soft label distribution learning, and a Cramér distance objective. In a study of participants with mild cognitive impairment, 42.6 % showed accelerated ageing, and faster Brain‑PACE scores correlated with greater functional and cognitive impairment as well as higher regional tau burden in key brain regions. The method improves probabilistic performance, reduces prediction bias, and provides predictive uncertainty, offering a complementary longitudinal imaging phenotype sensitive to early neurodegeneration.
By Samuel Maddox (School of Computing Sciences, University of East Anglia), Jacob Newman (School of Computing Sciences, University of East Anglia), Saber Sami (Norwich Medical School, University of East Anglia), Michal Mackiewicz (School of Computing Sciences, University of East Anglia), for the Alzheimer's Disease Neuroimaging Initiative, the Australian Imaging Biomarkers, Lifestyle flagship study of ageing
The paper introduces InterIL, a joint generative model that simultaneously produces a background image and a layout of foreground elements for graphic design templates, addressing the limitations of sequential generation approaches. InterIL connects pretrained image and layout diffusion backbones via a learnable communication module, freezing the backbones to preserve prior knowledge while training only the interaction module. The model also offers a test‑time guidance strategy, enabling users to impose preferences without retraining, and demonstrates superior image, layout, and harmonization quality compared to previous methods.
By Shirong Yang, Bo Yang, Ying Cao
The paper introduces the task alignment proxy, a method that accelerates hyperparameter selection for merging models fine‑tuned on diverse vision tasks. It addresses the challenge of training heterogeneous decoders, which makes traditional downstream performance evaluation costly. By using the proxy, the authors demonstrate that model merging can be applied efficiently to multi‑task vision models beyond CLIP‑based classification.
By Pau de Jorge, C\'esar Roberto de Souza, Bj\"orn Michele, Mert B\"ulent Sar{\i}y{\i}ld{\i}z, Philippe Weinzaepfel, Florent Perronnin, Diane Larlus, Yannis Kalantidis
arXiv:2609. 11073v1 Announce Type: cross Abstract: Data-driven distributionally robust optimization (DRO) typically treats the conditional outcome law as fixed and uses ambiguity sets to capture estimation error.
By Luhao Zhang, Shixiang Zhu
The paper proposes a flexible four‑coefficient parameterization for per‑token gating in on‑policy knowledge distillation, unifying existing methods such as EOPD and ToDi as special cases. Experiments on TweetEval with a Qwen3 teacher‑student pair show that the full family of gating configurations outperforms single‑channel baselines in most cells, and dynamic gating beats static baselines in a majority of isolated comparisons. The authors present the framework mainly as a shared coordinate system for comparing gating designs rather than as definitive statistical evidence.
By Suwan Wu, Yumeng Lin, Pengcheng Yuan, Xiaolong Jiang
The paper investigates how graph neural networks can quickly estimate the loss of algebraic connectivity after multiple road-link disruptions, comparing learned corrections to first‑order Fiedler sensitivity with analytical baselines. Experiments on synthetic failures and real OpenStreetMap data across six countries show that residual GCN and GraphSAGE models reduce mean absolute error for spatial and targeted failures, while second‑order perturbation offers minimal improvement. The study also demonstrates that sparse scaling allows the approach to scale to 20,000 nodes and that the spectral residual acts as a domain‑sensitive inductive bias for connectivity screening.
By Van-Truong Le
The paper introduces Narrative Consolidation, a new NLP task that aims to merge overlapping narrative documents—such as legal testimonies or historical accounts—into a single, chronologically coherent text, rather than merely compressing them. It defines the task, proposes an evaluation framework, and presents the Gospel Consolidation Language Resource, a benchmark built from the four Biblical Gospels with 169 canonical events and cross‑document alignments. Experiments show that providing an explicit temporal backbone dramatically improves performance, a simple length heuristic outperforms graph‑based methods, and temporal edges are the key discriminative signal.
By Roger A. Finger, Eduardo G. Cortes, Sandro J. Rigo, Gabriel de O. Ramos
The study audited ten different classifiers—including linear, tree‑ensemble, neural, glass‑box, and tabular foundation models—on national health survey data to predict myocardial infarction. By systematically removing features that could cause target leakage, the authors found that all models’ AUROC scores collapsed into a narrow band, indicating that reported high accuracy in prior work was largely due to leakage rather than model sophistication. The glass‑box explainable boosting machine performed comparably to other models while being much faster, and the authors demonstrated that fairness, calibration, and uncertainty can be audited and repaired without sacrificing performance.
By Raad Bin Tareaf, Murad Al-Rajab, Samia Loucif, Samer Ellaham, Cedric Schmitz
The paper introduces MUtE, a dual framework that simultaneously erases concept-specific information from representations and generates counterfactual mappings. By deriving new erasure functions based on optimal bounds, MUtE imposes a translational bias on counterfactual trajectories, aligning with geometric properties of concepts in language models. The authors demonstrate that this approach improves downstream algorithmic fairness and enables the generation of counterfactual texts.
By Antoine Saillenfest