The paper introduces a two‑stage speech anonymization framework that preserves both linguistic content and acoustic identity. It replaces personally identifiable information using a generative editing model and applies a flow‑matching anonymization technique (F3‑VA) to create diverse, distinct anonymized speakers. The authors evaluate privacy with speaker verification metrics and utility by training ASR, TTS, and SER models from scratch, showing stronger privacy protection with minimal utility loss compared to existing baselines.
By Yunchong Xiao, Yuxiang Zhao, Ziyang Ma, Shuai Wang, Kai Yu, Jiachun Liao, Xie Chen
The paper presents a spectral hierarchy for classifying the cosmic web by applying scale-weighting kernels to the density field before using eigenvalue-based methods. This framework unifies existing web definitions—potential/tidal, curvature-based, and higher-derivative levels—into a single, interpretable hierarchy that captures structure from large to small scales. The authors quantify the hierarchy’s information content by correlating a web contrast field with halo distributions, showing that it retains significant tracer-relevant information across scales, especially at nonlinear levels.
By Francisco-Shu Kitaura, Francesco Sinigaglia
WAND is an unsupervised tabular anomaly detector that scores each point by how far its projection on unit‑sphere directions deviates from a sub‑Gaussian baseline. The directions that flag a point serve as its explanation, providing per‑feature attribution at no extra cost and recoverable via gradients. On 47 ADBench datasets, WAND matches or exceeds 16 baselines in ROC‑AUC while delivering more accurate, faithful explanations than post‑hoc SHAP, LIME, or ECOD, all with linear scoring time and a probe‑efficiency guarantee.
By Lamine Diop
DIFFINT is a reconstruction‑based anomaly detector that uses a differentiable autoencoder with a latent bottleneck composed of soft, axis‑aligned interval memberships. Each latent unit represents a human‑readable hyper‑rectangle in feature space, allowing the model to encode how strongly an instance falls inside each interval and to compute reconstruction error as the anomaly score. The method provides a certified lower bound on reconstruction error for points outside all active intervals, a suppression mechanism for sparse abnormalities, and a closed‑form, label‑free importance ranking for each (unit, feature) pair, achieving top performance on 48 ADBench benchmarks against 22 baselines.
By Lamine Diop, Marc Plantevit
The paper proposes treating translation as a structured decision space explored by multiple autonomous agents, rather than producing a single output. Using Turkish–Syrian Arabic dialogue, three agents—zero‑shot, dialect‑stabilized, and pivot translation—are compared on 5,000 sentences, with stabilization nearly doubling dialect marker usage and reducing structural instability. The study introduces an interpretability framework that quantifies decision flexibility through dialect marker frequency, lexical proximity, and structural variance.
By Hasan Alkhder, Mohammad Abboush, Igor Tchappi, Ahmet Zengin, Amro Najjar
MIRA is a bilingual benchmark that evaluates whether large language models (LLMs) provide consistent medical information across different user phrasings, languages, and health literacy levels. It contains 4,320 prompts derived from 60 medically reviewed low‑risk health questions and reveals that models tend to omit key information and offer fewer concrete next steps when responding to low health‑literacy signals, a phenomenon termed Differential Information Dilution (DID). A knowledge‑guided mitigation prompt can reduce this dilution for most models, notably improving Claude and Qwen.
By Mengyu Xu, Qiaoxin Yang, Qianqian Wang, Xiwei Dai, Weiyi Wu, Chongyang Gao
FedPS is a federated preprocessing framework that uses aggregated statistics to address missing values, inconsistent formats, and heterogeneous feature scales in structured data. It employs data-sketching techniques to summarize local datasets efficiently, enabling federated algorithms for feature scaling, encoding, discretization, and missing-value imputation. The framework also extends preprocessing-related models, such as Bayesian Linear Regression, to both horizontal and vertical federated learning settings, offering communication‑efficient and consistent pipelines for practical deployments.
By Xuefeng Xu, Graham Cormode
The paper surveys collaborative learning methods that move beyond traditional Euclidean data to graph-structured data. It reviews foundational principles for Euclidean settings—learning effectiveness, efficiency, and privacy—and then extends the discussion to graph data, presenting a taxonomy of distribution scenarios, statistical heterogeneities, and standardized problem formulations. The survey also outlines open challenges and future research directions in this emerging field.
By R\'emi Bourgerie, \v{S}ar\=unas Girdzijauskas, Viktoria Fodor
The paper investigates whether human psychometric questionnaires can reliably characterize large language models (LLMs) in everyday interactions. By comparing eight open‑source LLMs’ value and personality profiles from Likert self‑reports (PVQ‑40/21 and BFI‑44/10) with generation probabilities on value‑laden user queries, the authors find substantial divergence between the two methods. The study shows that questionnaire items contain explicit lexical cues that lead models to respond in socially desirable ways, whereas realistic user queries lack such cues, and demographic persona prompts shift questionnaire responses but not generation outputs, indicating that questionnaire scores overestimate LLMs’ true behavioral tendencies.
By Woojung Song, Dongmin Choi, Yoonah Park, Jongwook Han, Eun-Ju Lee, Yohan Jo
The paper introduces Gradient-Aligned Reward (GAR), a reinforcement learning technique that uses truncated backpropagation to generate a compact gradient vector for each rollout and compares it to an expert-anchor gradient via cosine similarity. This dense, reasoning-aware reward improves large language model chain-of-thought reasoning on math benchmarks and transfers to other tasks without domain‑specific data, while adding less than 9% computational overhead. GAR outperforms existing baselines such as GRPO on Qwen3-4B and Qwen3-8B models.
By Leqi Zheng, Jinbo Su, Fang Niu, Chaokun Wang, Weiping Wang, Jiajun Zhang, Shannan Yan, Jie Wu, Zhaolu Kang, Rong Fu, Hang Zhang
A two-stage forecasting system is introduced for predicting CPU workload in private clouds. The model first forecasts customer service requests in Transactions Per Second (TPS) and then estimates future CPU usage from the TPS forecast, both stages using XGBoost within a cascaded architecture. Experiments on real private‑cloud traces show SMAPE below 7% for most applications, with the best case achieving an MAE of 0.7372 and an R² of 0.9185, and stable error accumulation over a 60‑step horizon.
By Ashir Javeed, Anton Borg, H{\aa}kan Grahn, Lars Lundberg, Dhyey Patel, Sogand Shirinbab
RobustSeiz is an open‑source, model‑agnostic framework designed to benchmark the robustness of EEG seizure detection models under realistic clinical stressors. It standardizes four public scalp‑EEG corpora into BIDS‑EEG trees, applies controlled distribution shifts—including environmental, noise, and adversarial transforms—across predefined hyperparameter grids, and reports comprehensive performance metrics such as sensitivity, precision, F1, false positives per 24 h, onset timing, and predictive agreement. The framework offers a Dockerized GPU pipeline, experiment registry, and both full‑evaluation and research‑subset modes, and demonstrates its utility by evaluating a contemporary detector on TUSZ across the full shift grid.
By Mohammad Mohammadi, Alireza Zarei
The study investigates how users perceive the helpfulness and privacy-preservation of large language model (LLM) responses in privacy-sensitive scenarios. Using 94 participants and 90 PrivacyLens scenarios, researchers found that users’ evaluations of identical LLM outputs varied widely, whereas five proxy LLM judges showed high agreement but low correlation with user judgments. The results suggest that proxy LLMs cannot reliably estimate users’ diverse perceptions of utility and privacy, highlighting the need for more user-centered evaluation methods.
By Xiaoyuan Wu, Roshni Kaushik, Wenkai Li, Lujo Bauer, Koichi Onoue
The paper reports that counterfactual fairness audits of clinical language‑model agents are unreliable without accounting for a per‑action instability floor. By repeatedly running identical vignettes, the authors found that actions changed 8.7% of the time, with instability varying eightfold across actions. A second model confirmed a pooled floor of 6.7%, showing that any reported fairness estimate lacking this floor cannot be interpreted as evidence of disparity.
By Rohith Reddy Bellibaltu, Manpreet Singh, Deepak Parashar, Rahul Joshi
The paper introduces a deep variational framework (DVF) for computing stable configurations of confined smectic liquid crystals using a modified Landau–de Gennes model. By representing orientational and positional order parameters on a regular reference domain and incorporating physical confinement through coordinate mappings, the DVF overcomes spectral bias with a warmup penalty, enabling robust recovery of oscillatory smectic states. The method reproduces known smectic‑A defect structures, predicts new layer morphologies in various confinement geometries, and even forecasts a chevron‑like smectic‑C state on a tangent‑anchored sphere.
By Yuchen Xie, Baoming Shi, Yucen Han, Lei Zhang
The paper introduces Test Cases Scaling (TCS), a two‑stage reinforcement learning framework that automatically generates sound and discriminative test cases for code large language models (LLMs). Stage 1 trains a test generator to produce tests aligned with a reference solution, while Stage 2 refines the generator to produce counterexample tests that target the solver’s current failure modes. Experiments on TACO and LiveCodeBench show that TCS improves pass@1 and inference‑time answer selection, and the learned test generator can also effectively rank outputs from other LLMs.
By Jiacheng Xu, Wentao Zhang, Zhiyi Lyu, Fuxiang Zhang, Chaojie Wang, Yang Liu, Bo An
The paper presents a differentiable hybrid modelling framework that combines a JAX finite volume population balance solver with learnable neural network components. This approach learns constitutive laws and initial conditions directly from experimental data, improving the fidelity of transport models in chemical engineering. The framework’s differentiability also enables optimisation of experimental settings for desired process outcomes.
By Arthur Jessop, Mohammed Alsubeihi, Ben Moseley, Ashwin Kumar Rajagopalan
The review examines how Uncertainty Quantification (UQ) can enhance machine learning applied to biosignals such as EEG, ECG, EOG, and EMG. It surveys 53 papers, outlining current methods, shortcomings, and theoretical frameworks, while highlighting misconceptions and gaps in diagnostic and prosthetic control contexts. The authors recommend further research on human-system interaction with UQ models in clinical settings.
By Ivo Pascal de Jong, Andreea Ioana Sburlea, Matias Valdenegro-Toro
The paper introduces CHARM, a lightweight fine‑tuned language model framework for detecting moral foundations in text. CHARM combines MAC cross‑attention, rationale alignment, and hate‑speech modulation to operationalize distinct psychological constructs, achieving up to 15.3% higher AUC in‑domain and outperforming supervised baselines on all out‑of‑domain datasets. The authors demonstrate CHARM’s scalability by applying it to large‑scale COVID‑19 Twitter data, revealing a strong link between moral value alignment and online endorsement behavior.
By Huixiang Fu, Marian-Andrei Rizoiu
The paper introduces DirBucket, a provider-side semantic watermarking system designed to audit document reuse in third‑party retrieval‑augmented generation (RAG) marketplaces. By embedding watermarks as meaning‑preserving paraphrases biased toward provider‑specific directions, DirBucket enables black‑box detection of unauthorized document reuse while maintaining retrieval quality. Experiments on mixed‑provider benchmarks show DirBucket consistently detects non‑compliance within 23 answers, resists adversarial laundering, and transfers effectively to real‑world corpora.
By Alexandr Goultiaev Tolstokorov, Kyriakos Mouratidis, Javad Dogani, Nikolaos Laoutaris