Silence is Not Consensus: Disrupting Agreement Bias in Multi-Agent LLMs via Catfish Agent for Clinical Decision Making

Large language models (LLMs) have demonstrated strong potential in clinical question answering, with recent multi-agent frameworks further improving diagnostic accuracy via collaborative reasoning. However, we identify a recurring issue of Silent Agreement, where agents prematurely converge on diagnoses without sufficient critical analysis, particularly in complex or ambiguous cases. We present a new concept called Catfish Agent, a role-specialized LLM designed to inject structured dissent and counter silent agreement. Inspired by the ``catfish effect'' in organizational psychology, the Catfish Agent is designed to challenge emerging consensus to stimulate deeper reasoning. We formulate two mechanisms to encourage effective and context-aware interventions: (i) a complexity-aware intervention that modulates agent engagement based on case difficulty, and (ii) a tone-calibrated intervention articulated to balance critique and collaboration. Evaluations on nine medical Q&A and three medical VQA benchmarks show that our approach consistently outperforms both single- and multi-agent LLMs frameworks, including leading commercial models such as GPT-4o and DeepSeek-R1.

Delving into RL for Image Generation with CoT: A Study on DPO vs. GRPO

Recent advancements underscore the significant role of Reinforcement Learning (RL) in enhancing the Chain-of-Thought (CoT) reasoning capabilities of large language models (LLMs). Two prominent RL algorithms, Direct Preference Optimization (DPO) and Group Relative Policy Optimization (GRPO), are central to these developments, showcasing different pros and cons. Autoregressive image generation, also interpretable as a sequential CoT reasoning process, presents unique challenges distinct from LLM-based CoT reasoning. These encompass ensuring text-image consistency, improving image aesthetic quality, and designing sophisticated reward models, rather than relying on simpler rule-based rewards. While recent efforts have extended RL to this domain, these explorations typically lack an in-depth analysis of the domain-specific challenges and the characteristics of different RL strategies. To bridge this gap, we provide the first comprehensive investigation of the GRPO and DPO algorithms in autoregressive image generation, evaluating their in-domain performance and out-of-domain generalization, while scrutinizing the impact of different reward models on their respective capabilities. Our findings reveal that GRPO and DPO exhibit distinct advantages, and crucially, that reward models possessing stronger intrinsic generalization capabilities potentially enhance the generalization potential of the applied RL algorithms. Furthermore, we systematically explore three prevalent scaling strategies to enhance both their in-domain and out-of-domain proficiency, deriving unique insights into efficiently scaling performance for each paradigm. We hope our study paves a new path for inspiring future work on developing more effective RL algorithms to achieve robust CoT reasoning in the realm of autoregressive image generation. Code is released at https://github.com/ZiyuGuo99/Image-Generation-CoT

EchoInk-R1: Exploring Audio-Visual Reasoning in Multimodal LLMs via Reinforcement Learning

Multimodal large language models (MLLMs) have advanced perception across text, vision, and audio, yet they often struggle with structured cross-modal reasoning, particularly when integrating audio and visual signals. We introduce EchoInk-R1, a reinforcement learning framework that enhances such reasoning in MLLMs. Built upon the Qwen2.5-Omni-7B foundation and optimized with Group Relative Policy Optimization (GRPO), EchoInk-R1 tackles multiple-choice question answering over synchronized audio-image pairs. To enable this, we curate AVQA-R1-6K, a dataset pairing such audio-image inputs with multiple-choice questions derived from OmniInstruct-v1. EchoInk-R1-7B achieves 85.77% accuracy on the validation set, outperforming the base model, which scores 80.53%, using only 562 reinforcement learning steps. Beyond accuracy, EchoInk-R1 demonstrates reflective reasoning by revisiting initial interpretations and refining responses when facing ambiguous multimodal inputs. These results suggest that lightweight reinforcement learning fine-tuning enhances cross-modal reasoning in MLLMs. EchoInk-R1 is the first framework to unify audio, visual, and textual modalities for general open-world reasoning via reinforcement learning. Code and data are publicly released to facilitate further research.

Unveiling Deep Shadows: A Survey on Image and Video Shadow Detection, Removal, and Generation in the Era of Deep Learning

Shadows are formed when light encounters obstacles, leading to areas of diminished illumination. In computer vision, shadow detection, removal, and generation are crucial for enhancing scene understanding, refining image quality, ensuring visual consistency in video editing, and improving virtual environments. This paper presents a comprehensive survey of shadow detection, removal, and generation in images and videos within the deep learning landscape over the past decade, covering tasks, deep models, datasets, and evaluation metrics. Our key contributions include a comprehensive survey of shadow analysis, standardization of experimental comparisons, exploration of the relationships among model size, speed, and performance, a cross-dataset generalization study, identification of open issues and future directions, and provision of publicly available resources to support further research.

Video Instance Shadow Detection

Video instance shadow detection aims to simultaneously detect, segment, associate, and track paired shadow-object associations in videos. This work has three key contributions to the task. First, we design SSIS-Track, a new framework to extract shadow-object associations in videos with paired tracking and without category specification; especially, we strive to maintain paired tracking even the objects/shadows are temporarily occluded for several frames. Second, we leverage both labeled images and unlabeled videos, and explore temporal coherence by augmenting the tracking ability via an association cycle consistency loss to optimize SSIS-Track's performance. Last, we build $\textit{SOBA-VID}$, a new dataset with 232 unlabeled videos of ${5,863}$ frames for training and 60 labeled videos of ${1,182}$ frames for testing. Experimental results show that SSIS-Track surpasses baselines built from SOTA video tracking and instance-shadow-detection methods by a large margin. In the end, we showcase several video-level applications.