GRAM-R$^2$: Self-Training Generative Foundation Reward Models for Reward Reasoning

The paper presents a novel approach called Zero-Training Task-Specific Model Synthesis (ZS-TMS) for few-shot medical image classification. This method addresses the challenge of limited annotated datasets in medical imaging by utilizing a pre-trained generative engine to synthesize parameters for a task-specific classifier. By requiring minimal input, such as a single example image, ZS-TMS aims to enhance the efficiency of medical image analysis, particularly for rare diseases where data is scarce.

International standards for biometric identity documents require strict adherence to pose requirements, particularly the square presentation of a subject's shoulders. This paper introduces a Shoulder Presentation Evaluation (SPE) algorithm that quantifies shoulder yaw and roll using 3D coordinates from two shoulder landmarks. Evaluated on 121 portrait images, the SPE scores showed a strong correlation with human-assigned labels. The algorithm effectively identifies non-compliant samples, providing a lightweight tool for automated compliance checking.

The study explores the scaling laws of deep neural networks in medical anatomical segmentation, revealing that larger training datasets lead to improved performance across various semantic tasks and imaging modalities. It highlights the significance of deformation-guided augmentation strategies, such as random elastic deformation and registration-guided deformation, in enhancing segmentation outcomes. The research aims to address the underexplored area of data scaling in medical imaging, proposing a novel image augmentation approach to generate diffeomorphic mappings.

MMaDA-Parallel is a new multimodal diffusion framework aimed at enhancing thinking-aware generation in AI models. It addresses performance degradation caused by error propagation in existing autoregressive approaches. The framework introduces ParaBench, a benchmark for evaluating text and image outputs, revealing that misalignment between reasoning and generated images contributes to performance issues. MMaDA-Parallel employs supervised finetuning and Parallel Reinforcement Learning to improve interaction between text and images throughout the denoising process.

The paper discusses Zero-shot coordination (ZSC), a significant challenge in multi-agent game theory, particularly in evolving games. It emphasizes the need for agents to coordinate with previously unseen partners without fine-tuning. The study introduces Scalable Population Training (ScaPT), an efficient reinforcement learning framework that enhances zero-shot coordination by utilizing a meta-agent to manage a diverse pool of agents, addressing limitations of existing methods that focus on small populations and computational constraints.

Recent research highlights a new class of attacks in federated learning that compromise model interpretability without impacting accuracy. The study reveals that adversarial clients can apply small color perturbations, shifting a model's saliency maps from meaningful regions while maintaining predictions. This method, termed the Chromatic Perturbation Module, systematically creates adversarial examples by altering color contrasts, leading to persistent poisoning of the model's internal feature attributions, challenging assumptions about model reliability.

As embodied agents navigate complex environments, the ability to perceive and track individual objects over time is crucial, particularly for tasks involving similar objects. In non-Markovian contexts, decision-making relies on object-specific histories rather than the immediate scene. Without a persistent memory of past interactions, robotic policies may falter or repeat actions unnecessarily. To address this, LIBERO-Mem is introduced as a task suite designed to test robotic manipulation under conditions of partial observability at the object level.

BEDLAM2.0 is a new dataset designed to improve the estimation of 3D human motion from video, addressing challenges posed by simultaneous human and camera movement. Unlike its predecessor, BEDLAM, this dataset features a wider variety of realistic camera motions, body shapes, clothing, and environments, including the addition of shoes. This advancement aims to enhance the training of 3D human pose and motion regressors, which are crucial for various applications in computer vision and artificial intelligence.