MiCRo is a two‑stage framework that improves personalized preference learning for large language models. It first uses a context‑aware mixture model to capture diverse human preferences from large binary preference datasets, then applies an online routing strategy to dynamically adjust mixture weights based on context, reducing ambiguity. Experiments on multiple datasets show that MiCRo captures diverse preferences and enhances downstream personalization.
By Jingyan Shen, Jiarui Yao, Rui Yang, Yifan Sun, Feng Luo, Rui Pan, Tong Zhang, Han Zhao
The paper introduces ANCHOR, an external supervisory framework driven by large language models (LLMs) that provides evaluative feedback at multiple stages of self‑evolving agents. By integrating ANCHOR into two open‑source self‑evolving agent frameworks, the authors demonstrate that it significantly improves safety performance while preserving core capabilities across coding, mathematical reasoning, and safety tasks. The study also finds that supervision based on execution results is especially effective and that increasing supervision frequency yields diminishing returns, offering practical guidance for future research.
By Dianxing Shi, Bowen Wang, Junqi He, Junhao Chen, Yuta Nakashima
SafeFlow is a real‑time, text‑driven humanoid control framework that blends physics‑guided motion generation with a three‑stage safety gate. It uses Physics‑Guided Rectified Flow Matching in a VAE latent space to produce physically executable trajectories, accelerates sampling with Reflow, and filters unsafe outputs via semantic OOD detection, directional sensitivity checks, and hard kinematic constraints before handing them to a motion‑tracking controller. Experiments on the Unitree G1 show that SafeFlow achieves higher success rates, better physical compliance, and faster inference than diffusion‑ and retargeting‑based baselines while maintaining motion diversity.
By Hanbyel Cho, Sang-Hun Kim, Jeonguk Kang, Donghan Koo
The paper introduces ‘DP-SPIN’, a trusted‑curator framework that generates differentially private semantic plans for aggregate insight generation. ‘DP-SPIN’ maps each record to a bounded sparse nonnegative vector over pre‑defined semantic concepts, sums these vectors into a semantic sketch, and releases a noisy plan containing admitted concepts and their masses. The framework provides user‑level privacy by clipping each user’s contribution and ensures that the final summary is differentially private through post‑processing, with guarantees established under both add/drop and replacement adjacency.
By Behrooz Razeghi
The paper presents a channel-informed neural network for physical-layer key generation (PKG) that extracts binary key features directly from IQ measurements while grounding the representation in the multipath channel. The multi-task recurrent network jointly learns reciprocity-preserving features and an auxiliary channel estimate, using deep metric learning and channel-informed supervision. Experiments on indoor and outdoor software-defined-radio data show lower bit disagreement for legitimate users, improved key diversity with ray-traced augmentation, and successful NIST randomness tests after SHA-3 privacy amplification.
By Jose Angel Sanchez Viloria, George Sklivanitis, Dimitris Pados, Elizabeth Serena Bentley
The paper introduces a multimodal anomaly detection framework that uses cross‑modal reconstruction of heterogeneous time‑series sensor data to detect faults in industrial systems. By learning to reconstruct each modality from the others, the method leverages complementary information across sensing channels without requiring explicit temporal alignment or identical sampling rates. An adaptive test‑time thresholding mechanism further improves robustness to distribution shifts caused by changing operating conditions, as demonstrated by strong fault detection performance in three industrial case studies, especially under out‑of‑distribution regimes.
By Magnus Munk Jensen, Dorte Hammersh{\o}i, Rafa{\l} Wi\'sniewski, Olga Fink
HyCoSeq is a new framework for learning genomic representations in hyperbolic space. It combines weighted Lorentzian residual aggregation with multi‑curvature Lorentz encoding, enabling full Lorentz representations to contribute directly to local aggregation. A bidirectional LSTM further captures contextual relationships across the sequence, extending local hyperbolic convolutions to sequence‑level representations. Experiments show HyCoSeq surpasses existing hyperbolic baselines and competes with much larger pretrained DNA language models without large‑scale pretraining.
By Chenhao Zeng, Zhibin Pu, Shufei Ge
IRENE is a deep learning model that provides probabilistic short‑range precipitation nowcasts over Italy at 1 km spatial and 5‑minute temporal resolution. It uses an encoder–forecaster architecture built on multi‑scale Convolutional Gated Recurrent Units (ConvGRUs) and is trained on national radar composites, with an importance‑sampling scheme and the almost‑fair Continuous Ranked Probability Score as its primary loss. Three training variants—standard, adversarial (IRENE‑GAN), and spectrally constrained (IRENE‑GAN‑RAPSD)—outperform benchmark methods STEPS and DGMR in probabilistic skill, though the advantage in mean absolute error is limited to the first 90 minutes.
By Alessandro Camilletti, Gabriele Franch, Elena Tomasi, Marco Cristoforetti
The paper introduces a look‑ahead splitting rule for Classification and Regression Trees (CART) that evaluates candidate splits by the error reduction achieved after growing a conventional CART subtree beneath each split. To keep the method computationally feasible, a smart look‑ahead algorithm is proposed that learns downstream split values from node‑level features. Experiments on simulated data and two real datasets show that both full and smart look‑ahead methods outperform the standard greedy splitting strategy, especially in hierarchical or interaction‑driven scenarios.
By Andrew Gao, Tianlin Liu, Ruichen Han, Lu Tian
The paper introduces LWVIC4Code, a non‑contrastive, layer‑wise representation learning method for detecting Type‑IV code clones—semantically equivalent fragments that differ syntactically. It builds on the VICReg framework, adding cross‑layer consistency regularization and depth‑dependent weighting to refine semantic information across transformer layers. Experiments on Python and multi‑language datasets show that LWVIC4Code matches or outperforms contrastive baselines and zero‑shot large language models, generalizing well to Java and C# without requiring negative samples.
By Luciano Marchezan, Kevin Delcourt, Eugene Syriani, Houari Sahraoui
arXiv:2609. 17477v1 Announce Type: cross Abstract: The absolute capacity of dense associative memory has mainly been analyzed for unbiased patterns.
By Yuto Sakurai, Takeaki Shimokawa, Kazunori Iwata, Kazushi Mimura
The paper introduces a collaborative optimization Boosting model for multiclass imbalanced learning that integrates density and confidence factors to create a noise‑resistant weight update mechanism and a dynamic sampling strategy. The modules are tightly coupled to coordinate weight updates, sample region partitioning, and region‑guided sampling. Experiments on 40 public imbalanced datasets show the model significantly outperforms seven state‑of‑the‑art baselines.
By Chuantao Li, Zhi Li, Jiahao Xu, Jie Li, Sheng Li
The paper introduces CoER, a framework that defends language‑model agents against adaptive indirect prompt injection (IPI) by employing attacker‑defender co‑evolution and refinement. CoER models IPI as a general‑sum Markov game, uses Co‑PPO to maintain historical opponent populations, and fine‑tunes defenders only on verified safe demonstrations. In experiments across seven domains, CoER cuts attack success from 38.5% to 0.2% while boosting task utility from 63.2% to 76.3%.
By Boyang Zhang, Qingxin Xiao, Lingwei Dang, Qingyao Wu
The study evaluates ten bias audit instruments across ten advanced language models on occupational gender, age, and socioeconomic status. While each tool reliably detects bias, their rankings of model performance are essentially random, indicating that different audits measure distinct constructs. The findings show that a single audit can identify bias direction within its own framework, but no audit can consistently rank models against one another.
By William Guey, Pierrick Bougault, Wei Zhang, Vitor D. de Moura, Jos\'e O. Gomes
The paper introduces the SAST-IR framework to evaluate large language models’ robustness against persuasion attacks in a memory‑less setting, revealing a flaw called "Refusal Inertia" that masks true vulnerability. Using the CP‑Agent and a custom CounterFact‑Strict dataset, the authors demonstrate that simple, diverse attack strategies achieve a 96% success rate, while complex attacks often trigger defensive compliance. The study highlights severe brittleness in current state‑of‑the‑art models when deprived of conversation history.
By Zhuoang Cai
The paper challenges the notion that refusal in large language models is governed by a single direction in activation space. It demonstrates that different refusal and non‑compliance categories map to distinct geometric directions, yet steering along any of these directions yields similar refusal–over‑refusal trade‑offs, acting as a shared one‑dimensional control knob. Using sparse autoencoders, the authors reveal a structured internal representation of refusal, comprising a reusable core of shared latents and style‑ or domain‑specific latents, and show that linear interventions collapse this structure into uniform behavioral control.
By Faaiz Joad, Majd Hawasly, Sabri Boughorbel, Nadir Durrani, Husrev Taha Sencar
The paper introduces VPRef, the first cross‑domain benchmark for Referring Remote Sensing Image Segmentation, containing 46,972 language‑image‑annotation triplets with a three‑tier linguistic hierarchy. It proposes a parameter‑efficient adaptation method based on the Segment Anything Model and Low‑Rank Adaptation, using pseudo‑label self‑training for visual drift and random multi‑granularity prompt mixing for textual drift. Experiments show the approach improves cross‑domain segmentation while altering only 1.08 % of the base model’s parameters, offering a strong baseline for future research.
By Quanwei Liu, Tao Huang, Jiaqi Yang, Wei Xiang
SAVOR is a training framework for multimodal large language models that adds token and answer confidence to the output schema, optimises a Group Relative Policy Optimisation objective to penalise calibration error and poor abstention, and uses the learned confidence at inference to revisit visual evidence only when uncertain. Experiments on POPE, HallusionBench, AMBER, and MMHal-Bench with InternVL3-8B and Qwen3-VL-8B backbones show that SAVOR reduces hallucination while maintaining general capability on MME and MMBench, achieving lower Expected Calibration Error than DPO and decoding baselines.
By Zixiu Ding, Zilin Zhao, Yingjie He, Xinlang Kang, Guansu Wang, Wei Zhang
The paper introduces a one‑pass conditional 3D rectified flow (3D Flow) framework for denoising whole‑body PET images, employing an optimized non‑uniform sampling strategy and a linear‑interpolant velocity‑matching objective. It reconstructs a full 3D volume in about 30 seconds, dramatically faster than multi‑hour 3D diffusion models, while maintaining high global image quality and lesion conspicuity even at ultra‑low doses (down to 1/100 of standard). Zero‑shot transfer tests on independent clinical data demonstrate robust performance across datasets and unseen dose levels.
By Jiale Shen, Guolin Wang, Chenhao Wang, Xinhui Su, Wei Luo, Feng Yu
PriorPose introduces a reference-guided correspondence framework for category-level object pose estimation that jointly performs canonicalization and alignment in a shared feature space. By embedding partial observations and a category prior as token sets in a seeded transformer, the network predicts per-point NOCS fields and a canonical deformation, while a deep pose head regresses the similarity transform. A two-part shape consistency objective couples correspondence, deformation, and pose, reducing reliance on memorized canonical orientations and avoiding error cascades, leading to state-of-the-art results on standard and larger-category benchmarks, especially under strict pose thresholds.
By Yihan Chen, Huan Ren, Wenfei Yang, Hang Du, Tianzhu Zhang, Feng Wu