SAGE (Semantic-Augmented Geometric Experts) is a new point‑cloud registration method that incorporates frozen, pre‑trained semantic representations into the registration pipeline without extra sensors or fine‑tuning. It uses a Semantic‑Guided Mixture of Experts (SG‑MoE) to route data so that each expert specializes in specific semantic‑geometric patterns, thereby reducing mismatches that plague geometry‑only approaches. On the 3DLoMatch benchmark, SAGE achieves the highest registration recall and, with the same number of refinement iterations, also attains the best inlier ratio and recall on both 3DLoMatch and 3DMatch.
By Tomoyasu Shimada, Kazuhiko Murasaki, Shoichiro Takeda, Shogo Sato, Riku Inoue, Toshihiko Nishimura, Ryuichi Tanida
TriDifNet is a diffusion-based network that jointly performs registration, fusion, and detection for misaligned RGB-T pedestrian perception. It includes three modules—RegDM for aligning infrared to visible images, FusDM for generating a detection-oriented fused representation, and DetDM for iterative bounding‑box denoising. The system uses closed‑loop optimization and a Nash‑MTL refinement to balance the three tasks, achieving state‑of‑the‑art detection metrics on the KAIST multispectral benchmark.
By Qianchao He, Jing Liu, Yunhao Li, Xiaoqing Tian
The paper introduces two end‑to‑end multimodal adaptation pipelines for 3D‑aware visual question answering on a custom spatial dataset. One pipeline fine‑tunes HiSpatial‑3B, adding metric‑scale 3D point maps to RGB inputs, while the other adapts Qwen3‑VL using memory‑efficient QLoRA supervised fine‑tuning. The engineering stack includes dataset conversion, image‑annotation alignment, monocular geometry preprocessing, instruction construction, mixed‑domain training, checkpoint management, batch inference, and automated evaluation, achieving significant accuracy gains on both the target spatial benchmark and a styled split.
By Yu Zhang, Xiaogang Wang, Yuhang Cheng, Jingrui Li, Wenbo Chen
The paper introduces XTrack, a unified framework that repurposes a pretrained video Diffusion Transformer (DiT) for cross-view temporal object correspondence. XTrack incorporates focal cross-view attention alignment, hard negative conditioning, and local context cropping to improve grounding across views and time, and it uses a pseudo‑label curation pipeline to generate dense supervision from sparse Ego‑Exo4D data. The authors demonstrate that XTrack achieves robust temporal consistency and identity preservation in cross‑view tracking tasks.
By Siyoon Jin, Dahyun Chung, Honggyu An, Sangbeom Lim, Wonjin Nam, Jaewoo Jung, WonJun Moon, Seungryong Kim
OA‑RAG is an agent that combines formal ontology reasoning with large language model (LLM) reasoning to produce more faithful answers. For each response it lists the ontology observations used as evidence and assigns a grounding level—entailed, supported, related, or unrelated—to indicate the strength of that evidence. On benchmarks from three large ontologies and six real‑world medical QA tasks, OA‑RAG outperforms other ontology‑fine‑tuned and popular RAG methods, especially on reasoning‑intensive questions, and its grounding levels correlate with higher accuracy.
By Hui Yang, Jiaoyan Chen, Yongsheng Gao
The paper introduces LM-DSB, a lightweight multi‑agent system that automates the creation of daily safety bulletins (DSBs) for construction sites using a retrieval‑augmented generation approach with a locally deployable large language model. The system employs specialized planner, researcher, and writer agents coordinated in a hierarchical architecture, fine‑tuned on construction safety documents and grounded in 37 HSE documents. Validation in two phases showed that the multi‑agent RAG system outperformed alternatives in lexical alignment, coherence, and document completeness, and expert reviewers rated it highly for regulatory compliance and factual accuracy.
By Xianxiang Zhao, Advik Mehta, Falak Sethi, Brian Gue, Qipei Mei, Lingzi Wu
The paper introduces Cognition-Aligned Reasoning Distillation (CARD), an RL-based method that converts solver trajectories from large reasoning models into teaching trajectories better suited for smaller student models. CARD uses two reward signals—a Distribution Alignment Reward and a Logical Coherence Reward—to align the teacher’s outputs with the student’s cognitive capabilities and reduce misleading reasoning steps. Experiments show that CARD improves both in-domain and out-of-distribution performance compared to distillation from raw teacher trajectories.
By Yashuo Luo, Qianren Mao, Ziqi Qin, Junnan Liu, Haitao Li, Qili Zhang, Yuan Gao, Zhijun Chen, Xu Wang
The paper shows that sovereign ESG metrics are mispriced by linear macro‑financial models during systemic shocks, revealing a fundamentally non‑linear risk transmission. Using a gradient‑boosting model on 147 sovereigns from 2010‑2024, forecast errors are reduced by over 39% during extreme macro events. The study identifies a strict, non‑linear "Governance Premium"—environmental and social gains provide no capital advantage until a regulatory threshold is crossed—and offers regime‑conditional uncertainty bounds that remain valid even with severe ESG reporting failures.
By Sourav Bose, Taoufik Bouraoui
HIDRA is a hierarchical dual-routing attention framework designed for replay‑free lifelong imitation learning in robotics. It separates instruction‑level expert retrieval from context‑dependent refinement and adds key‑level geometric regularization to keep tasks aligned yet distinct, thereby reducing routing interference. Experiments on LIBERO benchmarks show that HIDRA outperforms both replay‑free and replay‑based baselines, improving AUC, forward transfer, and maintaining low forgetting, especially under high interference and language variation.
By Fanqi Yu, Matteo Tiezzi, Cigdem Beyan, Tommaso Apicella, Vittorio Murino
The paper introduces conservative contextual dueling bandits, which enforce that each displayed item’s cumulative score remains above a specified fraction of a trusted baseline’s score. It presents algorithms—MaxInP-BC for linear utilities and BTIGW-BC for general utilities—that guarantee safety with high probability while achieving sublinear regret. Experiments show these methods satisfy the safety constraint more reliably than unconstrained counterparts, with only a moderate increase in regret.
By Rohan Deb, Mohammad Ghavamzadeh, Arindam Banerjee
The paper introduces Content-Invariant Gradient Attribution (CIGA), a method for selecting training data for large language models that mitigates surface-form bias in gradient-based attribution. CIGA identifies layers whose gradients are both influential and stable under semantics-preserving perturbations, then fuses these signals with semantic representations to produce robust data rankings. Experiments show that CIGA yields more stable rankings, maintains downstream performance, and reduces attribution cost.
By Jia Gu, Liang Pang, Huawei Shen
Designing power electronics converters traditionally requires intensive human expertise across topology sizing, commercial component selection, circuit simulation verification, and schematic drafting....
Multi-State Linear Attention (MSLA) improves long‑sequence modeling by compressing context into separate memory states, but its aggregation only uses mean alignment, ignoring distribution. MERLIN (MomEnt Routing for multi‑state LINear attention) enhances this by tracking first and second moments of each segment’s keys, enabling queries to recover both mean and variance of alignment for more content‑aware aggregation. Experiments on Gated DeltaNet‑1.3B and Mamba‑2‑780M show MERLIN consistently outperforms existing aggregation methods on long‑context tasks, achieving significant gains on RULER and LongBench while using less memory.
The paper introduces Faithful Personalized Fact-Checking Explanation Rewriting, a method that tailors fact‑checking explanations to individual readers while keeping factual accuracy intact. It uses LLM‑simulated readers to generate relative preference data, then applies a two‑stage framework called PERPL: first constructing persona‑specific supervision via cross‑persona pairing, and second refining a lightweight rewriter through exploration‑guided preference optimization. Experiments show PERPL outperforms larger general‑purpose LLMs in aligning with personalized preferences while preserving factual fidelity.
The paper introduces a lightweight neural network inspired by insect olfaction that classifies raw gas sensor readings using a sparse, high‑dimensional binary representation. It operates online, can be recalibrated from limited data, and outperforms linear SVMs on gas mixtures while matching optimized nonlinear SVMs. The architecture uses only binary weights and Hebbian learning, enabling deployment on ultra‑low‑power neuromorphic hardware, as demonstrated on the TEXEL chip with 74.6% accuracy at 27 µW static power.
The paper investigates how nineteenth‑century scholarship—specifically Pressutti’s regesta and partial editions—can aid handwritten text recognition (HTR) for the registers of Pope Honorius III, which lack modern expert transcriptions beyond regnal years 1‑2. By comparing legacy layers to audited transcriptions, the authors identify a three‑tier division of labor: at the orthographic tier, projecting legacy editions onto curial spelling with replay reduces word‑error rate by 1.5 points; at the lexical tier, a legacy lexicon covers 82.2 % of out‑of‑vocabulary words and improves NER performance; and at the topological tier, regesta data help locate entry starts with 73.9 % coverage and estimate a character‑error rate of 7.80 % beyond folio 203. The study shows that while antiquarian scholarship cannot replace modern transcription, it can effectively steer, verify, and segment HTR across untranscribed registers.
Tail Cues, Principal Corrections: Plug-and-Play Rectification for Open-Set Test-Time Adaptation introduces TALOR, a lightweight module that rectifies features during non‑stationary target streams by decomposing them into high‑gain principal and low‑gain tail coordinates using the source linear head as a geometric anchor. TALOR applies a soft‑routed affine correction in the principal subspace, subtracts a drift estimate via weighted ridge regression, and reconstructs the feature for the source head. On ImageNet‑C with six csOOD datasets, TALOR surpasses the next‑best UniEnt by 2.2% H‑score, runs 2.5× faster, and uses only 12% of its GPU memory, while also boosting COME to 68.4% H‑score with minimal memory overhead.
ActiveAlign is a framework that uses active perception to close the Real2Sim2Real gap in robot manipulation. It introduces two modules: ActiveAlign‑Waving, which performs hand‑eye calibration via active hand motions and differentiable rendering, and ActiveAlign‑Touching, which refines object alignment with tactile feedback for millimeter‑level precision. The approach enables strong real‑to‑simulation alignment, allowing simulation environments to be curated from uncalibrated datasets and enabling a real Unitree G1 robot to manipulate millimeter‑scale objects such as buttons and sliders.
Backpropagation (BP) remains the predominant algorithm for training deep neural networks, but it is biologically implausible and suffers from the backward locking problem, constraining parallelization...
Contextual dueling bandits learn from pairwise comparisons, the same feedback used to fine-tune language models from human preferences. Standard algorithms explore aggressively and can show users two...