The paper investigates how different multimodal design choices affect the performance of misinformation detection systems. Using over 3,375 experiments across three benchmark datasets and various pre‑trained vision and language models, the authors systematically compare design options and conduct robustness analyses. The study offers practical guidance on which choices improve detection, when they may fail silently, and which pipeline components most influence model behavior, addressing four key research questions.
By Akshit Sharma, Prashant W. Patil
ActKV is a new KV cache compression framework designed for agentic large language model (LLM) inference. It prioritizes cache entries that contribute to action generation, using action-oriented eviction, confidence-driven budget allocation, and page-aware compression to reduce memory usage while preserving accuracy. In long-trace tasks, ActKV retains 98.53% of FullKV’s accuracy using only 25.98% of its peak memory and boosts token and task throughput by 3.97× and 3.58×, respectively.
By Zihan Wang, Cheng Tang, Lei Gong, Chao Wang, Wenqi Lou, Teng Wang, Xuehai Zhou
SkillFlow is an open, multi-stage retrieval system that helps AI agents selectively load relevant skills from a large library of community-contributed SKILL.md definitions. The pipeline uses dense retrieval, two rounds of cross-encoder reranking, and LLM-based selection to balance recall and precision. Evaluations on SkillsBench and Terminal-Bench show that SkillFlow improves performance when high-quality skills are available, but retrieval alone does not help if the corpus lacks executable skills for the target domain.
By Fangzhou Li, Pagkratios Tagkopoulos, Ilias Tagkopoulos
The paper introduces a scalable hybrid Quantum Diffusion Model for medical image analysis, leveraging a Discrete-Time Quantum Walk executed on a real quantum device to model forward diffusion dynamics. A classical learning model is employed for the backward denoising step, enabling the processing of large real-world medical data, including grayscale, RGB, and moderate-sized 3D volumes. The authors benchmark their quantum approach against a classical discrete-state diffusion model, demonstrating competitive generation performance across three state-of-the-art image generation metrics.
By Francesco Aldo Venturelli, Stefano Martina, Marco Parigi, Filippo Caruso, Alba Cervera-Lierta, Miguel A. Gonz\'alez Ballester
NBAvatar is a method for realistic rendering of head avatars that handles non‑rigid deformations caused by hand‑face interaction. It introduces a hybrid implicit‑explicit representation, combining explicit oriented planar primitives with implicit neural rendering, and uses a geometry‑aware training scheme to jointly optimize these representations. The approach achieves up to 53% LPIPS reduction compared to Gaussian‑based avatar methods, improves PSNR and SSIM, and surpasses the state‑of‑the‑art InteractAvatar in structural similarity for novel‑view and novel‑pose rendering.
By David Svitov, Mahtab Dahaghin, Pietro Morerio, Alessio Del Bue
The paper introduces a reliability-regulated trajectory optimization framework for progressive COLMAP‑free 3D Gaussian Splatting (3DGS). It uses a self‑supervised bidirectional cycle‑consistency mechanism to control camera trajectory estimation through forward motion propagation and retrospective trajectory correction, thereby reducing error compounding without external priors. Experiments on Tanks and Temples and CO3D‑V2 demonstrate improved camera trajectory accuracy and novel‑view rendering quality compared to existing unposed baselines.
By Zijian Wu, Jinliang Wang, Zidian Lin, Ying Song, Ziqian Lu, Hanjie Ma, Zhen Ye, Mingfeng Jiang
The paper introduces an interaction‑centric framework that unifies representations for two‑finger gripper manipulation across different robot embodiments. By using a parameterized universal gripper abstraction and a canonical gripper‑frame representation, the system infers sub‑tasks from language and RGB‑D inputs, grounds interaction triplets, and employs hybrid features and a Flow‑Matching Transformer to generate smooth 7‑DoF action sequences. Experiments in both simulation and real‑world settings show that this approach achieves competitive benchmark performance while enabling extreme cross‑embodiment and cross‑viewpoint zero‑shot sim‑to‑real transfer to heterogeneous robot platforms.
By Guanlin Li, Shifeng Bao, Yihan Zhao, Haitao Shen, Haoyang Li, Chen Zhao, Tong Yang, Jie Tang, Jing Zhang
Spectral Feedback is a new algorithm for aligning discrete diffusion models at test time by iteratively revisiting and editing token positions rather than only steering the reverse process. It selects edit-sets—groups of token positions to re-mask and re-sample—using sparse Fourier representations of edit-set value functions, enabling efficient optimization of which tokens to revisit. The method is model-agnostic and improves alignment performance across pretrained, test‑time aligned, and fine‑tuned diffusion models, achieving significant gains in protein stability for inverse folding tasks.
By Shai Dickman, Mert Cemri, Landon Butler, Kannan Ramchandran
The paper investigates the use of Zero Cost Proxies (ZCPs) to identify high‑performing wearable Human Activity Recognition (HAR) models without full training. Eight ZCPs were evaluated across six benchmark HAR datasets, showing that the top‑predicted architectures achieve performance within 7% of fully trained models, and training the top‑10 predictions reaches within 2% of full training. This demonstrates that ZCPs can significantly reduce computational costs while maintaining competitive accuracy in sensor‑based HAR tasks.
By Richard Goldman, Varun Komperla, Thomas Ploetz, Harish Haresamudram
The paper introduces SRBench, a benchmark dataset comprising 45,064 labeled entries from 32 curated secondary studies, designed to evaluate large language model performance in systematic review screening while addressing class imbalance. It also presents PromptSR, a tool that facilitates prompt experimentation, experiment management, and result analysis for LLM-based screening. A use case demonstrates the practical application of both SRBench and PromptSR.
By Gauransh Kumar, Luciano Marchezan, Guillaume Genois, K\'evin Delcourt, Eugene Syriani
The paper introduces a recursive self-improvement framework for language models that replaces an external teacher with a frozen copy of the student, enabling dynamic co-evolution (DCE) and self-refined concise learning (SRCL). DCE allows the privileged teacher to evolve alongside the student, while SRCL trains on shorter, verified rewrites to reduce verbosity. Experiments show that the combined DCE+SRCL approach outperforms traditional on‑policy self‑distillation across multiple model sizes and math benchmarks, achieving significant accuracy gains and shorter outputs.
By Shangjian Yin, Zehao Zhao, Kavosh Asadi, Rui Liu, Yuchen Lu, Shike Mei, Hang Cui, Luke Simon, Zhouxing Shi, Hamed Firooz
The paper proposes a new way to evaluate language models on cultural understanding by focusing on interpretive depth rather than just factual recall. It argues that literary interpretation, where scholars can disagree yet still assess the quality of evidence, provides a useful framework for testing how models handle cultural references, reuse, and transformation across texts. The authors suggest linking evidence-centered benchmarks, preserving scholarly disagreement, and conducting model-development experiments on literary data, with Danish literature as a starting point for broader applications.
By Daniel Hershcovich, Alexander Conroy, Jens Bjerring-Hansen
RupeeBias is a new benchmark that audits demographic bias in large language models (LLMs) when they provide economic guidance in India. It contains 39,150 prompts across four use cases—salary estimation, salary increment estimation, counter‑offer recommendation, and service pricing recommendation—varying 87 India‑specific demographic identifiers such as caste, religion, regional identity, gender, disability, and urban‑rural location. Evaluations of nine LLMs show that outputs differ by an average of 20.2% when only the demographic identifier changes, revealing systematic disparities across all six axes.
By Pavithra P M Nair, Bhavik Talaviya, Shourya Bhushan, Rahul Pankajakshan, Seema Guruvadoo, Avinash Agarwal, Gilad Gressel, Krishnashree Achuthan
The paper introduces "stale‑document poisoning," a temporal alignment failure where outdated retrieval evidence causes retrieval‑augmented generation models to produce incorrect answers even when the model could answer correctly without retrieval. A benchmark of 317 verified knowledge reversals across medicine, law, software, and platform policy shows that outdated evidence flips 30–37% of Llama and Qwen answers, rising to 66–75% when models are explicitly instructed to trust the document. The study demonstrates that providing explicit validity information and a recency‑aware re‑ranker can substantially reduce poisoning, highlighting the need for models to assess whether retrieved evidence remains applicable.
By Md Shamim Ahmed, Lukas Galke Poech, Richard R\"{o}ttger
LeakScale is an interventional framework that estimates the causal effect of benchmark exposure on model performance. By creating new executable tasks that require private, family‑specific information not present in the public task, LeakScale controls access to that information and measures the resulting change in accuracy. Across 2,048 families, two model families, two domains, and 262,144 generations, exposure consistently improves accuracy, with gains ranging from +7.17 to +27.31 percentage points, thereby distinguishing whether benchmark contact occurred from how strongly a reported score depends on it.
By Divyansh Singh
The paper presents an all-in-one framework for restoring polarization images affected by multiple coupled degradations. It identifies the normalized Stokes representation as effective for separating intensity and polarization, and introduces a dual-branch architecture that uses pretrained general restoration priors for the intensity branch while transferring knowledge to the polarization branch via cross-domain feature transform. A composite-degradation benchmark is also established to support future research.
By Chenggong Li, Jinhao Liu, Caiyun Wu, Yidong Luo, Junchao Zhang, Degui Yang
The paper introduces Timo, a kinematics-aware multimodal diffusion transformer designed for human motion generation. Timo employs fully shared multimodal attention, flow matching, and geometric/rotational-kinematics supervision to better coordinate articulated motion, and uses a two-stage curriculum to align motion with text captions. The authors also present a new benchmark of 40,025 clips from six datasets, showing that Timo outperforms state‑of‑the‑art methods, achieving a 40.8% relative improvement over Kimodo on average.
By Zhao Wang, Jiangtao Hu, Jack Yu, Tao Yu
The paper introduces PIMDE, a self‑supervised monocular depth estimation framework that decomposes input images into perceptual feature maps, each encoding a specific visual cue. Separate depth branches process these maps to produce individual depth estimates, which are then fused explicitly. Experiments on the KITTI benchmark show that PIMDE matches the accuracy of existing self‑supervised methods while offering clearer insight into how each perceptual cue contributes to depth prediction.
By Zain Ul Abidin, George Dimas, Dimitris K. Iakovidis
arXiv:2609.31193v1 Announce Type: new
Abstract: Current Audio-Visual LLMs (AVLLMs) struggle with reasoning over videos featuring multi-speaker dialogues. In such videos, resolving "who says what" is...
By Jihoo Jung, Youngjoon Jang, Joon Son Chung
arXiv:2609.31198v1 Announce Type: new
Abstract: We present Light Field Primitives (LFP), a formulation for novel view synthesis that replaces the dense ray database with a compact set of differentiab...
By Liang Chen, Jiahui Ning, Xun Jiang, Xing Xu, Jimmy Ren, Fenglei Fan, Heng Tao Shen