Radiology report generation (RRG) aims to automatically produce clinically faithful reports from chest X-ray images. Prevailing work typically follows a scale-driven paradigm, by multi-stage training over large paired corpora and oversized backbones, making pipelines highly data- and compute-intensive. In this paper, we propose Oracle-educated GRPO {OraPO) with a FactScore-based reward (FactS) to tackle the RRG task under constrained budgets. OraPO enables single-stage, RL-only training by converting failed GRPO explorations on rare or difficult studies into direct preference supervision via a lightweight oracle step. FactS grounds learning in diagnostic evidence by extracting atomic clinical facts and checking entailment against ground-truth labels, yielding dense, interpretable sentence-level rewards. Together, OraPO and FactS create a compact and powerful framework that significantly improves learning efficiency on clinically challenging cases, setting the new SOTA performance on the CheXpert Plus dataset (0.341 in F1) with 2--3 orders of magnitude less training data using a small base VLM on modest hardware.

Recent unified models such as GPT-5 have achieved encouraging progress on vision-language tasks. However, these unified models typically fail to correctly understand ECG signals and provide accurate medical diagnoses, nor can they correctly generate ECG signals. To address these limitations, we propose UniECG, the first unified model for ECG capable of concurrently performing evidence-based ECG interpretation and text-conditioned ECG generation tasks. Through a decoupled two-stage training approach, the model first learns evidence-based interpretation skills (ECG-to-Text), and then injects ECG generation capabilities (Text-to-ECG) via latent space alignment. UniECG can autonomously choose to interpret or generate an ECG based on user input, significantly extending the capability boundaries of current ECG models. Our code and checkpoints will be made publicly available at https://github.com/PKUDigitalHealth/UniECG upon acceptance.

ABSTRACT Fine-tuning large pre-trained models for downstream tasks has become a fundamental approach in natural language processing. Fully fine-tuning all model parameters is computationally expensive and memory-intensive, especially in resource-constrained environments. Existing parameter-efficient fine-tuning methods reduce the number of trainable parameters but typically overlook the varying sensitivity of different model layers and the importance of training data. In this work, we propose TsqLoRA, a novel method that integrates data-quality-driven selection with sensitivity-aware low-rank adaptation, consisted of two main components: a quality-aware sampling mechanism for selecting the most informative training data, and a dynamic rank allocation module that adjusts the rank of each layer based on its sensitivity to parameter updates. The experimental results demonstrate that TsqLoRA improves fine-tuning efficiency while maintaining or even improving performance on a variety of NLP tasks.

Multimodal semantic learning plays a critical role in embodied intelligence, especially when robots perceive their surroundings, understand human instructions, and make intelligent decisions. However, the field faces technical challenges such as effective fusion of heterogeneous data and computational efficiency in resource-constrained environments. To address these challenges, this study proposes the lightweight LCMF cascaded attention framework, introducing a multi-level cross-modal parameter sharing mechanism into the Mamba module. By integrating the advantages of Cross-Attention and Selective parameter-sharing State Space Models (SSMs), the framework achieves efficient fusion of heterogeneous modalities and semantic complementary alignment. Experimental results show that LCMF surpasses existing multimodal baselines with an accuracy of 74.29% in VQA tasks and achieves competitive mid-tier performance within the distribution cluster of Large Language Model Agents (LLM Agents) in EQA video tasks. Its lightweight design achieves a 4.35-fold reduction in FLOPs relative to the average of comparable baselines while using only 166.51M parameters (image-text) and 219M parameters (video-text), providing an efficient solution for Human-Robot Interaction (HRI) applications in resource-constrained scenarios with strong multimodal decision generalization capabilities.

Large Language Models (LLMs) have demonstrated strong performance in information retrieval tasks like passage ranking. Our research examines how instruction-following capabilities in LLMs interact with multi-document comparison tasks, identifying what we term the "Ranking Blind Spot", a characteristic of LLM decision processes during comparative evaluation. We analyze how this ranking blind spot affects LLM evaluation systems through two approaches: Decision Objective Hijacking, which alters the evaluation goal in pairwise ranking systems, and Decision Criteria Hijacking, which modifies relevance standards across ranking schemes. These approaches demonstrate how content providers could potentially influence LLM-based ranking systems to affect document positioning. These attacks aim to force the LLM ranker to prefer a specific passage and rank it at the top. Malicious content providers can exploit this weakness, which helps them gain additional exposure by attacking the ranker. In our experiment, We empirically show that the proposed attacks are effective in various LLMs and can be generalized to multiple ranking schemes. We apply these attack to realistic examples to show their effectiveness. We also found stronger LLMs are more vulnerable to these attacks. Our code is available at: https://github.com/blindspotorg/RankingBlindSpot

Porous materials exhibit vast structural diversity and support critical applications in gas storage, separations, and catalysis. However, predictive modeling remains challenging due to the multiscale nature of structure-property relationships, where performance is governed by both local chemical environments and global pore-network topology. These complexities, combined with sparse and unevenly distributed labeled data, hinder generalization across material families. We propose the Interaction Topological Transformer (ITT), a unified data-efficient framework that leverages novel interaction topology to capture materials information across multiple scales and multiple levels, including structural, elemental, atomic, and pairwise-elemental organization. ITT extracts scale-aware features that reflect both compositional and relational structure within complex porous frameworks, and integrates them through a built-in Transformer architecture that supports joint reasoning across scales. Trained using a two-stage strategy, i.e., self-supervised pretraining on 0.6 million unlabeled structures followed by supervised fine-tuning, ITT achieves state-of-the-art, accurate, and transferable predictions for adsorption, transport, and stability properties. This framework provides a principled and scalable path for learning-guided discovery in structurally and chemically diverse porous materials.

Dementia is a progressive neurodegenerative disorder with multiple etiologies, including Alzheimer's disease, Parkinson's disease, frontotemporal dementia, and vascular dementia. Its clinical and biological heterogeneity makes diagnosis and subtype differentiation highly challenging. Graph Neural Networks (GNNs) have recently shown strong potential in modeling brain connectivity, but their limited robustness, data scarcity, and lack of interpretability constrain clinical adoption. Explainable Graph Neural Networks (XGNNs) have emerged to address these barriers by combining graph-based learning with interpretability, enabling the identification of disease-relevant biomarkers, analysis of brain network disruptions, and provision of transparent insights for clinicians. This paper presents the first comprehensive review dedicated to XGNNs in dementia research. We examine their applications across Alzheimer's disease, Parkinson's disease, mild cognitive impairment, and multi-disease diagnosis. A taxonomy of explainability methods tailored for dementia-related tasks is introduced, alongside comparisons of existing models in clinical scenarios. We also highlight challenges such as limited generalizability, underexplored domains, and the integration of Large Language Models (LLMs) for early detection. By outlining both progress and open problems, this review aims to guide future work toward trustworthy, clinically meaningful, and scalable use of XGNNs in dementia research.

Compared to traditional models, agentic AI represents a highly valuable target for potential attackers as they possess privileged access to data sources and API tools, which are traditionally not incorporated into classical agents. Unlike a typical software application residing in a Demilitarized Zone (DMZ), agentic LLMs consciously rely on nondeterministic behavior of the AI (only defining a final goal, leaving the path selection to LLM). This characteristic introduces substantial security risk to both operational security and information security. Most common existing defense mechanism rely on detection of malicious intent and preventing it from reaching the LLM agent, thus protecting against jailbreak attacks such as prompt injection. In this paper, we present an alternative approach, LLMZ+, which moves beyond traditional detection-based approaches by implementing prompt whitelisting. Through this method, only contextually appropriate and safe messages are permitted to interact with the agentic LLM. By leveraging the specificity of context, LLMZ+ guarantees that all exchanges between external users and the LLM conform to predefined use cases and operational boundaries. Our approach streamlines the security framework, enhances its long-term resilience, and reduces the resources required for sustaining LLM information security. Our empirical evaluation demonstrates that LLMZ+ provides strong resilience against the most common jailbreak prompts. At the same time, legitimate business communications are not disrupted, and authorized traffic flows seamlessly between users and the agentic LLM. We measure the effectiveness of approach using false positive and false negative rates, both of which can be reduced to 0 in our experimental setting.

Cancer is one of the leading health challenges for women, specifically breast and ovarian cancer. Early detection can help improve the survival rate through timely intervention and treatment. Traditional methods of detecting cancer involve manually examining mammograms, CT scans, ultrasounds, and other imaging types. However, this makes the process labor-intensive and requires the expertise of trained pathologists. Hence, making it both time-consuming and resource-intensive. In this paper, we introduce a novel vision transformer (ViT)-based method for detecting and classifying breast and ovarian cancer. We use a pre-trained ViT-Base-Patch16-224 model, which is fine-tuned for both binary and multi-class classification tasks using publicly available histopathological image datasets. Further, we use a preprocessing pipeline that converts raw histophological images into standardized PyTorch tensors, which are compatible with the ViT architecture and also help improve the model performance. We evaluated the performance of our model on two benchmark datasets: the BreakHis dataset for binary classification and the UBC-OCEAN dataset for five-class classification without any data augmentation. Our model surpasses existing CNN, ViT, and topological data analysis-based approaches in binary classification. For multi-class classification, it is evaluated against recent topological methods and demonstrates superior performance. Our study highlights the effectiveness of Vision Transformer-based transfer learning combined with efficient preprocessing in oncological diagnostics.

The meta-task of obtaining and aligning representations through contrastive pretraining is steadily gaining importance since its introduction in CLIP and ALIGN. In this paper we theoretically explain the advantages of synchronizing with trainable inverse temperature and bias under the sigmoid loss, as implemented in the recent SigLIP and SigLIP2 models of Google DeepMind. Temperature and bias can drive the loss function to zero for a rich class of configurations that we call $(\mathsf{m}, \mathsf{b}_{\mathsf{rel}})$-Constellations. $(\mathsf{m}, \mathsf{b}_{\mathsf{rel}})$-Constellations are a novel combinatorial object related to spherical codes and are parametrized by a margin $\mathsf{m}$ and relative bias $\mathsf{b}_{\mathsf{rel}}$. We use our characterization of constellations to theoretically justify the success of SigLIP on retrieval, to explain the modality gap present in SigLIP, and to identify the necessary dimension for producing high-quality representations. Finally, we propose a reparameterization of the sigmoid loss with explicit relative bias, which improves training dynamics in experiments with synthetic data.