The paper shows that a plain Transformer can serve as an effective graph learner by adding three lightweight modifications: simplified L₂ attention, adaptive RMS normalization, and an MLP-based positional encoding stem. These changes preserve token magnitude and enable the model to achieve high expressivity on graph benchmarks, outperforming more complex graph transformer variants. The results suggest that plain Transformers can act as a unified backbone for multimodal learning across language, vision, and graph domains.
By Liheng Ma, Soumyasundar Pal, Yingxue Zhang, Philip H. S. Torr, Mark Coates
CounterVid introduces a scalable counterfactual video generation framework that creates videos differing only in actions or temporal structure while keeping scene context intact. The approach uses multimodal LLMs for action proposals and diffusion models for editing, producing a synthetic dataset of ~26k preference pairs for action recognition and sequence ordering. With the MixDPO optimization method, the authors demonstrate significant improvements in action recognition and temporal ordering on Qwen2.5‑VL and InternVL3 backbones, while maintaining overall video understanding.
By Tobia Poppi, Burak Uzkent, Amanmeet Garg, Lucas Porto, Garin Kessler, Yezhou Yang, Marcella Cornia, Lorenzo Baraldi, Rita Cucchiara, Florian Schiffers
CARE: Causally-Aligned Reasoning Exploration for Medical Large Language Models proposes a new framework to improve medical reasoning in LLMs. It introduces two key conditions—Causal Sufficiency and Proximal Learnability—to curate high-quality training trajectories, using agreement-based self-verification and dynamic entropy bounds. Experiments on medical multimodal and text-only benchmarks show that CARE outperforms competitors, reducing incorrect reasoning and enhancing training stability.
By Yucheng Zhou, Peng Luo, Qianning Wang, Chengzhong Xu, Jianbing Shen
The paper introduces Composed Gait Retrieval (CoGR), a task that retrieves a target gait sequence using a reference sequence and a natural language modification query. To support this, the authors create the first gait-language datasets—Language‑Augmented CCPG and Language‑Augmented CASIA‑B—via an automated annotation pipeline powered by large vision‑language models. They propose ComposeGait, an identity‑anchored composition framework with a Part‑aware Identity Adapter that injects identity tokens into a shared Q‑Former, achieving state‑of‑the‑art retrieval performance on both benchmarks.
By Jingchen Fei, Zengbin Wang, Yukun Liu, Muyi Sun, Shibiao Xu, Man Zhang
CRAM (Centroid‑Routing and Adaptive MoE) is a method for Multimodal Continual Instruction Tuning that isolates task‑specific patterns into independent modules to reduce catastrophic forgetting. It uses adaptive‑rank instantiation to allocate only the necessary parameters for new tasks, and centroid‑guided routing with an orthogonality penalty to reuse existing experts while preventing interference. Experiments on diverse benchmarks show CRAM outperforms existing approaches.
By Jun-Tao Tang, Zhen-Hao Xie, Yu-Cheng Shi, Da-Wei Zhou
The paper introduces HealthCUES, a real‑time streaming pipeline that extracts and analyzes cough and throat‑clearing events from live spoken conversations. It detects coughs within sub‑second latency, distinguishes cough subtypes (dry, wet, barking, whooping), differentiates coughing from throat clearing, and estimates temporal boundaries, all while gating alerts based on conversational context. The system, built on Qwen3Omni, achieves high accuracy (93% F1 for cough detection) and low latency (340 ms) and has been validated by healthcare professionals for telehealth use.
By Tanmay Laud, Herprit Mahal, Subhabrata Mukherjee
The paper introduces Vision-Free Adaptation (VFA), a method that separates multilingual language enhancement from visual alignment in multimodal large language models. VFA fine‑tunes a base LLM on multilingual text to create a multilingual task vector, which is then merged with the vision‑aligned task vector of an existing MLLM. Experiments on five MLLMs and six multilingual benchmarks show consistent gains while preserving multimodal and text‑only performance, and using less than 2% of text data narrows the performance gap to fully multimodal‑trained models.
By Yixia Li, Yaqing Shi, Zhiwen Ruan, Dongdong Zhang, Lingjie Jiang, Shaohan Huang, Yun Chen, Guanhua Chen, Furu Wei
The paper introduces a zero‑shot video restoration and enhancement framework that leverages a text‑to‑image latent diffusion model along with multi‑modal references. It employs dual prompt tuning inversion and sampling to cut inference time to about one‑third of the original, while also strengthening performance and temporal consistency. Additional techniques such as texture‑aware video token merging, referenced self‑attention, and referenced token merging further improve temporal coherence across frames.
By Cong Cao, Huanjing Yue, Xin Liu, Jingyu Yang
The paper introduces Checkerboard, a clean‑label backdoor attack that uses a closed‑form, data‑independent trigger design based on an input‑space Fisher‑separability objective and a ridge four‑neighbor local‑smoothness prior. This approach yields a pixel‑wise checkerboard trigger without requiring data access, surrogate model training, or iterative optimization, and it outperforms existing norm‑bounded clean‑label attacks across four benchmark datasets. On CIFAR‑10, poisoning 20 samples with a 10/255 perturbation achieves a 95.72% attack success rate, while on IN‑100 a 0.4% global poisoning rate yields over 83% ASR without harming clean accuracy, and the attack remains robust against state‑of‑the‑art defenses.
By Yi Yang, Jinyang Huang, Binbin Liu, Feng-Qi Cui, Xiaokang Zhou, Haiming Jin, Zhi Liu, Jie Zhang, Meng Li
The paper proposes a multi‑modal AI framework for real‑time queue prediction, management, and resource optimisation in border control systems. It integrates heterogeneous data sources using LSTM networks for forecasting and applies Model Predictive Control and scheduling optimisation to generate actionable policies for officers. Evaluation on synthetic traffic data shows up to 35% reduction in prediction error, 30% lower average waiting time, and nearly 20% higher throughput compared to ARIMA and rule‑based methods.
By Varvara Mama, Eleni Veroni, Nikolaos Kapsalis, Christos D. Nikolopoulos, Anargyros T. Baklezos
ET‑Prune is a training‑free framework that dynamically allocates visual token budgets in multimodal large language models based on question‑conditioned evidence. It protects text‑like spatial regions, converts evidence uncertainty into a token floor, and progressively prunes concentrated evidence while retaining diffuse or text‑dense tokens. In six backbone‑benchmark comparisons, ET‑Prune matches or outperforms other pruned methods while keeping roughly half the visual tokens, achieving notable gains on OCRBench‑v2 and MMBench v1.1.
By Zizhong Ding, Junxian Li, Kai Liu, Shaoqiu Zhang, Xiao Xiao, Linghe Kong, Yulun Zhang
SPAR-K is a scheduled periodic alternating early‑exit framework for interleaved spoken language models that reduces decoding depth for speech tokens while maintaining quality. It lets most speech positions exit at a fixed intermediate layer and inserts periodic full‑depth refresh steps to counter distribution shift. Experiments on Step‑Audio‑2‑mini and GLM‑4‑Voice show up to 11 % depth reduction with less than 0.82 % drop in question‑answering accuracy and negligible impact on MOS and WER.
By Hsiao-Ying Huang, Cheng-Han Chiang, Hung-yi Lee
TransMeme introduces a multi‑agent framework for cross‑cultural meme transcreation, addressing the unique challenges of preserving intent, adapting cultural meaning, and maintaining multimodal consistency. The system coordinates specialized agents for cultural adaptation, text rewriting, revision, and visual adjustment, and is evaluated on Chinese‑English meme pairs. Human and LLM‑based evaluations show that TransMeme outperforms baselines, achieving a 33.1% average improvement in human scores and a 60% Top‑1 ranking rate in LLM judgments.
By Jingyi Zheng, Yule Liu, Zifan Peng, Tianyi Hu, Yuemeng Zhao, Xinhu Zheng, Xinlei He
CultureVidBench is a new benchmark that evaluates how well text‑to‑video generation models capture cultural details. It contains 1,000 prompts spanning 12 countries, 6 continents, 8 cultural regions, and 14 cultural aspects, grouped into material culture, social practice & performance, and ritual & ceremony. Human studies and automated assessments show that while current models perform well on semantic adherence and visual quality, they often miss fine‑grained cultural details, especially for underrepresented regions and multimodal cues.
By Xianjing Han, Yuhan Su, Yang Deng, Dong Ma, Wee Peng Tay, Bin Zhu
Aphanta is an automated framework that diagnoses how well image editors can produce task‑aligned visual intermediates for multimodal large language models (MLLMs). It evaluates three reasoning conditions—direct, editor‑generated, and idealized intermediate—to distinguish visual potential from practical editor performance across 20 tasks and various editor–MLLM pairs. The study finds that image editing benefits certain tasks like visual cue injection and grounding, but is less reliable for symbol‑sensitive or structural tasks, and demonstrates measurable performance gains with a Qwen pipeline.
By Hengyuan Xu, Wei Cheng, Yumeng Ji, Xuanyang Zhang, Xianfang Zeng, Gang Yu, Xingjun Ma
The paper introduces Omni-Interactive Universal Embedder (OmniUE), a unified embedding framework that learns a single representation space for text, video, and audio using learnable tokens and intermediate-layer representations. OmniUE supports omni-interactive querying, allowing users to input text, visual regions, or audio spans, which are processed by segmenters and an omni-LLM to generate user-conditioned embeddings. The authors evaluate OmniUE on the new OmniCHOIR benchmark and other multimodal tasks, reporting significant performance gains over state‑of‑the‑art baselines across textual, audio, and visual interactive settings.
By Wei-Yao Wang, Kazuya Tateishi, Shuyang Cui, Christian Simon, Takashi Shibuya, Shusuke Takahashi, Yuki Mitsufuji
The paper introduces PuMVR, a benchmark of 1,000 Punjabi image‑text pairs spanning three scripts—Gurmukhi, Shahmukhi, and Roman—to evaluate Vision‑Language Models (VLMs). Testing ten state‑of‑the‑art VLMs reveals a significant Script Gap: models perform well in one script but poorly in another, with accuracy differences up to 16%. The authors propose the Script Consistency Rate (SCR) as a new metric, noting it can be as low as 24.8% on their benchmark, and argue that current multilingual VLMs are not truly multi‑script.
By Prabhjot Singh, Bhushan Pawar, Madhu Reddiboina, Rajvee Sheth
The paper introduces a new benchmark that evaluates large language models (LLMs) on their agentic mathematical reasoning rather than just final answers. It aligns problem‑solving behaviors with a taxonomy of reusable mathematical atomic capabilities and includes planning, action, and feedback tasks in both textual and multimodal settings. Experiments show that models with similar end‑to‑end accuracy can have very different agentic profiles, highlighting the importance of process‑level evaluation.
By Jiayi Kuang, Yinghui Li, Yunze Song, Keyu Chen, Zhifeng Shen, Yangning Li, Yidong Wang, Di Yin, Ruizhi Qiao, Xing Sun, Kai Jin, Ying Shen, Liang Lin, Philip S. Yu
Visual Information-Guided Parallel Decoding for Diffusion Multimodal Large Language Models introduces the VIG‑Sampler, a method that prioritizes tokens for decoding based on their attention to image tokens and penalizes redundancy in image‑attention distributions. The approach aims to improve the quality of multimodal generation by selecting more informative tokens during diffusion decoding. Experiments on seven captioning and VQA benchmarks with three open‑source dMLLMs show that VIG‑Sampler outperforms the Info‑Gain Sampler by an average of 19.3 CIDEr points and achieves better COCO Caption results using only half as many decoding steps.
By Insu Lee, Wooje Park, Wonseok Shin, Jinwoo Son, Byonghyo Shim
DocTalkBN is a large-scale multimodal dataset of authentic expert telemedicine conversations in Bengali, comprising 557.63 hours of paired audio and text, 1,515 multi-turn patient calls, and 10,274 host–doctor question–answer exchanges across 26 medical specialties. The dataset contains 1.7 million tokens and preserves the spontaneity and contextual richness of real medical interactions in a low-resource language. Three downstream tasks—medical triage classification, advice safety evaluation, and medical named entity recognition—are constructed to benchmark large language models and encoder-based baselines, demonstrating DocTalkBN’s practical usefulness for clinically grounded reasoning.
By Anik Saha, Fahmida Sultana Naznin, Sadatul Islam Sadi, Ananya Shahrin Promi, Wahid Al Azad Navid, Rifat Shahriyar