The article introduces an open image dataset for active‑fire segmentation in satellite imagery, comprising 2,148 image‑mask pairs from 25 California wildfires captured between July 2020 and August 2026. Each 512×512 pixel, three‑channel image is a Sentinel‑2 Level‑2A composite of bands B12, B11, and B8A, with a fixed linear rendering applied uniformly. Masks distinguish background, short‑wave‑infrared rule‑based active fire, and invalid observations, and the dataset includes chip‑level metadata, an incident‑disjoint split, and a mask‑blind analyst review of 233 test chips.
By Shreyan Mitra, Mohammadreza Narimani, Parastoo Farajpoor
The paper introduces FastFlowGS, a streaming 4D Gaussian Splatting method that reconstructs fast-moving subjects from a sparse set of external cameras, and Monaco4D, a photorealistic Unreal Engine 5 benchmark featuring Formula 1 sequences with dense ground truth. FastFlowGS combines sparse matches, semi-dense tracks, and dense optical flow using a Kalman-style temporal update, achieving significant performance gains over existing baselines on both CMU-Panoptic and Monaco4D datasets. The benchmark provides varied illumination and viewpoints from trackside, onboard, and drone cameras, enabling evaluation of high-speed outdoor reconstruction.
By Saswat Subhajyoti Mallick, Riu Cherdchusakulchai, Marc Ruiz Olle, Albert Mosella-Montoro, Jose Ribeiro-Gomes, Francisco Vicente Carrasco, Fernando De la Torre
QART is a quantum‑classical hybrid architecture that augments a language model with quantum encoding, CIM‑based QUBO optimization, and quantum decoding to improve long‑horizon reasoning. The authors claim that, under certain assumptions, QART can maintain a non‑zero probability of recovering an optimal reasoning path while traditional autoregressive LLMs see their acceptance probability drop to zero as cumulative risk grows. Experiments on six benchmarks with three backbone models show that QART outperforms the baselines in 14 of 15 pairings, with relative gains up to 84.0% on SciCode.
By Lehao Lin, Yuheng Cheng, Guolong Liu, Yao Li, Xuning Tan, Xiyuan Zhou, Ruixi Zou, Shi Wang, Huan Zhao, Wenxuan Liu, Haifeng Wu, Junhua Zhao
The paper introduces Style‑Debiased DPO (SD‑DPO), a method that refines large language models’ ability to retrieve stored knowledge by using preference optimization that corrects for style differences while preserving factual accuracy. SD‑DPO evaluates on the EntiGraph storing‑side framework and outperforms baseline CPT on the QuALITY reading‑comprehension benchmark, achieving higher accuracy with far fewer training tokens. In a knowledge‑editing setting (AToKE), SD‑DPO attains an overall accuracy of 0.982, correctly answering queries with either new or old facts based on the requested time period.
By Takayuki Yamamoto, Daisuke Kawahara
EviScope is a new paired counterfactual benchmark that evaluates grounded language models by fixing the question while manipulating evidence—adding, removing, distracting, or contradicting it. The v1.1 dataset includes 40 four‑condition quartets with repaired counterfactual claims and span‑level support labels for automated assessment. Experiments on Qwen2.5‑7B, Llama 3.1 8B, and Gemini 3.5 Flash show that paired metrics reveal grounding behaviors hidden by simple answer accuracy, such as unsupported answers, conflict blindness, and incorrect non‑answer actions.
By Suryadeep Singh Deswal
Lit3R is a system developed by tus-nlp for the LitTraceQA shared task, which focuses on evidence-grounded question answering over scientific literature. The system combines off-the-shelf retrieval, reranking, and large language model components without task-specific training, using an iterative retrieval process that merges BM25-based sparse and dense retrieval, cross-encoder reranking, and LLM verification, along with paper-to-paper expansion. In the official test set, Lit3R achieved a 4th place ranking on the leaderboard.
By Akira Ise, Kotaro Kumagai, Yuta Yamaguchi, Hisanori Ozaki, Yukio Uematsu, Ikuya Yamada
The paper introduces a panoramic aerodynamic performance prediction framework for turbomachinery cascades that first predicts basic Navier-Stokes parameters (temperature, pressure, density) and then derives key turbine stage performance metrics. A transformer‑enhanced neural operator (TNO) is trained on Rotor 37 blades to accurately predict transonic compressor blade performance, outperforming state‑of‑the‑art deep learning operators like FNO and DeepONet. The TNO also supports downstream tasks such as sensitivity analysis and optimization, achieving CFD‑like results while cutting computational cost by four orders of magnitude.
By Qineng Wang, Zhendong Guo, Liming Song, Tianyuan Liu
arXiv:2609. 16298v1 Announce Type: new Abstract: Despite recent advances in the verification of nonlinear neural feedback systems, scalability remains the central obstacle, as state-of-the-art solvers do not yet handle the network sizes and nonlinear dynamics of autonomy applications.
By I. Samuel Akinwande, Mykel J. Kochenderfer, Clark Barrett
HoloAegis is a minimally parametric topological inference framework that uses frozen representations to map text onto the unit sphere and makes decisions via Gibbs‑Boltzmann free‑energy differences over pre‑computed anchor centroids. On a frozen three‑benchmark protocol, it matches WildGuard‑7B on toxicity, outperforms it on harmful behaviors, but underperforms on oversafety detection, while ShieldGemma‑2B fails on indirect harms. The study demonstrates that geometric guardrails can substitute for LLM judges in some cases and must defer to them in others, with anchor banks reducing score variance and boundary displacement.
By Tak Ho Alex Li, Kaijie Liu, Lik-Hang Lee, Kin Chung Ho, Ping Shum, Michael K. Ng
SyncVoice is a new automatic video dubbing framework that adds a lightweight Text‑Visual Fusion Module to a pretrained TTS system, aligning visual features with linguistic representations to produce temporally synchronized speech. The approach avoids complex architectural changes and achieves state‑of‑the‑art performance on the LRS3 dataset in zero‑shot dubbing. When further trained on a large bilingual audio‑visual corpus, SyncVoice improves vocal fidelity while maintaining synchronization, enabling a single model to dub both Chinese and English videos.
By Kaidi Wang, Yi He, Wenhao Guan, Weijie Wu, Peijie Chen, Hongwu Ding, Xiong Zhang, Di Wu, Meng Meng, Jian Luan, Lin Li, Qingyang Hong
The paper investigates intersectional biases in multimodal clinical predictions using Electronic Healthcare Records (EHR). It introduces datasets MIMIC-Eye1 and MIMIC-IV ED, applies unified text representations from pre‑trained clinical language models, and benchmarks bias mitigation at the intersectional subgroup level. Results show that subgroup‑specific mitigation is robust across datasets, subgroups, and embeddings, effectively addressing intersectional biases in multimodal settings.
By Ayaazuddin Mohammad, Kishore Sampath, Resmi Ramachandranpillai
arXiv:2609. 16406v1 Announce Type: cross Abstract: Machine learning-based partial differential equations (PDEs) solvers have attracted significant attention in recent years.
By Chi-An Chen, Chunyang Liao, Ming Zhong
The paper re‑examines the impact of removing the encoder from Action Chunking Transformers (ACT), a model used for robot manipulation learning. Contrary to the original claim that encoder removal drops success rates from 35% to 2%, the authors find no such dramatic effect in their re‑runs, though minor variations remain uncertain. They attribute discrepancies to training length and checkpoint selection, and note that the encoder’s latent variable offers little reconstruction benefit on the tested benchmark, while its removal speeds up training.
By Bo Kang
The paper proposes a cooperation mechanism for multiple small neural network agents that share predictions during training to reduce model complexity while maintaining performance. By incorporating shared predictions into the loss function, agents influence each other's weight updates through strategies such as voter, majority, and weighted average models. Experiments on standard benchmarks show that several small agents can outperform a single large model, achieving comparable accuracy with fewer parameters and lower computational cost.
By Jarod Ketcha Kouakep, Sreyvi UANN, Timoteo Carletti
FLAT (Flexible‑Length Aligned Transmodal representations) is a joint multimodal pre‑training framework that learns a shared encoder for images and text, producing 1‑D continuous embeddings that can be directly used by downstream generative decoders. By combining contrastive alignment with bidirectional cross‑modal generative objectives, FLAT yields representations that are both discriminative and generative, enabling cross‑modal retrieval and generation with a single pre‑training stage. The model achieves strong performance on T2I generation (GenEval 71.1), image captioning (BLEU‑4 40.5, CIDEr 138.6), and retrieval tasks (Recall@5 86.8/75.8 on MS‑COCO, 98.3/93.6 on Flickr30K), and supports linear interpolation, latent space arithmetic, and zero‑shot composed retrieval.
By Guangyu Sun, Shlok Kumar Mishra, Wentao Bao, Robert Zhenheng Yang, Xiao Wang, Xiyuan Wang, Yujunrong Ma, Chen Yuan, Max Xiangjun Fan, Jun Xiao, Jianpeng Cheng
The paper introduces Turn-level Multiscale Density Ratio Estimation (tlm‑DRE), a post‑training technique for large language model agents that assigns varying weights to different turns and employs asymmetric token‑level training based on positive‑negative gaps across multiple turns. Unlike conventional alignment methods such as PPO, DPO, DIL, and GRPO, which mainly target single‑turn tasks, tlm‑DRE is designed for complex multi‑turn reasoning. Experiments on diverse agent benchmarks demonstrate that tlm‑DRE competes with traditional alignment methods and enables LLMs to perform robustly in both in‑domain and out‑of‑domain multi‑turn tasks.
By Zishuo Zhao (Alibaba Group), Kai Chen (Alibaba Group), Ao Li (Alibaba Group), Yuan Liu (Alibaba Group)
The paper introduces MedPCFM‑TED, a one‑step distillation framework that uses teacher‑guided endpoint supervision and geometric matching losses to generate cranial implants from point clouds. It outperforms existing one‑step methods on the SkullBreak benchmark, remains competitive on SkullFix, and achieves a generation time of about 0.04 s per sample. The approach demonstrates that rapid, high‑quality implant generation is possible without multiple neural evaluations during inference.
By Kamil Kwarciak, Marek Wodzinski
HUMAID-NER is the first named entity recognition dataset built on the HumAID benchmark, comprising 60,000 English disaster tweets with approximately 175,000 labeled entity spans across ten operationally motivated entity types. The dataset was created using a reproducible three‑stage hybrid pipeline that combines a spaCy transformer model, disaster‑domain EntityRuler patterns, and structured regular expressions with priority‑based overlap resolution. A joint multitask learning framework using a shared RoBERTa‑large encoder and homoscedastic uncertainty weighting achieves an NER span micro‑F1 of 0.841 and classification macro‑F1 of 0.761, and the authors provide a real‑time web dashboard, dataset, models, and pipeline code for reproducibility.
By Aijaz Ali, Nazish Basir, Sarfaraz Nawaz, Danish Nazir Arain, Haris Ali
The paper investigates how rubric-based grading in medical AI benchmarks can fail when the rubrics themselves are flawed. Using the RIFT taxonomy, the authors analyze two clinical benchmarks—HealthBench Professional and LiveMedBench—and discover significant issues: nearly 30% of criteria on HealthBench Professional are non-atomic and over 65% are misaligned or rigid. They demonstrate that correcting bundled criteria can shift scores by up to 15.9 percentage points, revealing that rubric flaws materially affect evaluation outcomes.
By Prithvi Dixit, Pedram Hosseini
The paper introduces three new vision‑centric evaluation benchmarks—temporal frame retrieval, video future prediction, and causal memory distortion—to assess visual question answering in large video models. Unlike traditional benchmarks that rely on text-based multiple choice questions, these tasks require models to reason directly from visual inputs. The authors find that current state‑of‑the‑art models struggle with visual queries, highlighting a gap in visual understanding that future research should address.
By Rwiddhi Chakraborty (Oliver), Yinong (Oliver), Wang, Cheng Zhang, Fan Bai, Zhuoran You, Michael Kampffmeyer, Yong Jae Lee, Fernando De la Torre, Robert Jenssen