Photoplethysmography (PPG) sensor in wearable and clinical devices provides valuable physiological insights in a non-invasive and real-time fashion. Specialized Foundation Models (FM) or repurposed time-series FMs are used to benchmark physiological tasks. Our experiments with fine-tuning FMs reveal that Vision FM (VFM) can also be utilized for this purpose and, in fact, surprisingly leads to state-of-the-art (SOT A) performance on many tasks, notably blood pressure estimation. We leverage VFMs by simply transforming one-dimensional PPG signals into image-like two-dimensional representations, such as the Short-Time Fourier transform (STFT). Using the latest VFMs, such as DINOv3 and SIGLIP-2, we achieve promising performance on other vital signs and blood lab measurement tasks as well. Our proposal, Vision4PPG, unlocks a new class of FMs to achieve SOT A performance with notable generalization to other 2D input representations, including STFT phase and recurrence plots. Our work improves upon prior investigations of vision models for PPG by conducting a comprehensive study, comparing them to state-of-the-art time-series FMs, and demonstrating the general PPG processing ability by reporting results on six additional tasks. Thus, we provide clinician-scientists with a new set of powerful tools that is also computationally efficient, thanks to Parameter-Efficient Fine-Tuning (PEFT) techniques. 1 Introduction

Antidepressant nonadherence is pervasive, driving relapse, hospitalization, suicide risk, and billions in avoidable costs. Clinicians need tools that detect adherence lapses promptly, yet current methods are either invasive (serum assays, neuroimaging) or proxy-based and inaccurate (pill counts, pharmacy refills). We present the first noninvasive biomarker that detects antidepressant intake from a single night of sleep. A transformer-based model analyzes sleep data from a consumer wearable or contactless wireless sensor to infer antidepressant intake, enabling remote, effortless, daily adherence assessment at home. Across six datasets comprising 62,000 nights from >20,000 participants (1,800 antidepressant users), the biomarker achieved AUROC = 0.84, generalized across drug classes, scaled with dose, and remained robust to concomitant psychotropics. Longitudinal monitoring captured real-world initiation, tapering, and lapses. This approach offers objective, scalable adherence surveillance with potential to improve depression care and outcomes.

AI-driven crop health mapping systems offer substantial advantages over conventional monitoring approaches through accelerated data acquisition and cost reduction. However, widespread farmer adoption remains constrained by technical limitations in orthomosaic generation from sparse aerial imagery datasets. Traditional photogrammetric reconstruction requires 70-80\% inter-image overlap to establish sufficient feature correspondences for accurate geometric registration. AI-driven systems operating under resource-constrained conditions cannot consistently achieve these overlap thresholds, resulting in degraded reconstruction quality that undermines user confidence in autonomous monitoring technologies. In this paper, we present Ortho-Fuse, an optical flow-based framework that enables the generation of a reliable orthomosaic with reduced overlap requirements. Our approach employs intermediate flow estimation to synthesize transitional imagery between consecutive aerial frames, artificially augmenting feature correspondences for improved geometric reconstruction. Experimental validation demonstrates a 20\% reduction in minimum overlap requirements. We further analyze adoption barriers in precision agriculture to identify pathways for enhanced integration of AI-driven monitoring systems.

Software vulnerabilities remain a persistent risk, yet static and dynamic analyses often overlook structural dependencies that shape insecure behaviors. Viewing programs as heterogeneous graphs, we capture control- and data-flow relations as complex interaction networks. Our hybrid framework combines these graph representations with light-weight (<4B) local LLMs, uniting topological features with semantic reasoning while avoiding the cost and privacy concerns of large cloud models. Evaluated on Java vulnerability detection (binary classification), our method achieves 93.57% accuracy-an 8.36% gain over Graph Attention Network-based embeddings and 17.81% over pretrained LLM baselines such as Qwen2.5 Coder 3B. Beyond accuracy, the approach extracts salient subgraphs and generates natural language explanations, improving interpretability for developers. These results pave the way for scalable, explainable, and locally deployable tools that can shift vulnerability analysis from purely syntactic checks to deeper structural and semantic insights, facilitating broader adoption in real-world secure software development.

Multimodal large reasoning models (MLRMs) are rapidly advancing vision-language reasoning and are emerging as a foundation for cross-modal intelligence. Hallucination remains a persistent failure mode, manifesting itself as erroneous reasoning chains and misinterpretation of visual content. In this study, we observe that attention heads exhibit a staged division: shallow heads predominantly serve perception, while deeper heads shift toward symbolic reasoning, revealing two major causes of hallucination, namely perceptual bias and reasoning drift. To address these issues, we propose a lightweight and interpretable two-step plugin, Functional Head Identification and Class-conditioned Rescaling, which locates perception- and reasoning-oriented heads and regulates their contributions without retraining. Evaluations on three real-world MLRMs (Kimi-VL, Ocean-R1, R1-Onevision), six benchmarks across three domains, and four baselines show that our plugin achieves an average improvement of 5% and up to 15%, with only <1% additional computation and 9% of baseline latency. Our approach is completely model-agnostic and significantly enhances both the reliability and interpretability of the off-the-shelf MLRMs, thereby enabling their safe deployment in high-stakes applications. Our code is available at https://anonymous.4open.science/r/Functional-Attention-Control.

Clinical reasoning in medicine is a hypothesis-driven process where physicians refine diagnoses from limited information through targeted history, physical examination, and diagnostic investigations. In contrast, current medical benchmarks for large language models (LLMs) primarily assess knowledge recall through single-turn questions, where complete clinical information is provided upfront. To address this gap, we introduce VivaBench, a multi-turn benchmark that evaluates sequential clinical reasoning in LLM agents. Our dataset consists of 1762 physician-curated clinical vignettes structured as interactive scenarios that simulate a (oral) examination in medical training, requiring agents to actively probe for relevant findings, select appropriate investigations, and synthesize information across multiple steps to reach a diagnosis. While current LLMs demonstrate competence in diagnosing conditions from well-described clinical presentations, their performance degrades significantly when required to navigate iterative diagnostic reasoning under uncertainty in our evaluation. Our analysis identified several failure modes that mirror common cognitive errors in clinical practice, including: (1) fixation on initial hypotheses, (2) inappropriate investigation ordering, (3) premature diagnostic closure, and (4) failing to screen for critical conditions. These patterns reveal fundamental limitations in how current LLMs reason and make decisions under uncertainty. Through VivaBench, we provide a standardized benchmark for evaluating conversational medical AI systems for real-world clinical decision support. Beyond medical applications, we contribute to the larger corpus of research on agentic AI by demonstrating how sequential reasoning trajectories can diverge in complex decision-making environments.

The performance of Large Language Models (LLMs) often degrades when crucial information is in the middle of a long context, a "lost-in-the-middle" phenomenon that mirrors the primacy and recency effects in human memory. We propose that this behavior is not simply a flaw indicative of information loss but an adaptation to different information retrieval demands during pre-training: some tasks require uniform recall across the entire input (a long-term memory demand), while others prioritize the most recent information (a short-term memory demand). Consistent with this view, we show that this U-shaped performance curve emerges when LLMs (GPT-2 and Llama variants) are trained from scratch on two simple human memory paradigms simulating long-term and short-term memory demands. Our analysis reveals that while the recency effect directly aligns with short-term memory demand in the training data, the primacy effect is induced by the uniform long-term memory demand and is additionally influenced by the model's autoregressive properties and the formation of attention sinks. Our main findings from simple human memory paradigms also generalize to a sequence completion task, which more closely resembles the next-token prediction process in LLM pre-training. Together, our findings reveal how information retrieval demands, model architecture, and structural attention dynamics during model training can jointly produce positional bias observed in LLMs.

Unlike regular tokens derived from existing text corpora, special tokens are artificially created to annotate structured conversations during the fine-tuning process of Large Language Models (LLMs). Serving as metadata of training data, these tokens play a crucial role in instructing LLMs to generate coherent and context-aware responses. We demonstrate that special tokens can be exploited to construct four attack primitives, with which malicious users can reliably bypass the internal safety alignment of online LLM services and circumvent state-of-the-art (SOTA) external content moderation systems simultaneously. Moreover, we found that addressing this threat is challenging, as aggressive defense mechanisms-such as input sanitization by removing special tokens entirely, as suggested in academia-are less effective than anticipated. This is because such defense can be evaded when the special tokens are replaced by regular ones with high semantic similarity within the tokenizer's embedding space. We systemically evaluated our method, named MetaBreak, on both lab environment and commercial LLM platforms. Our approach achieves jailbreak rates comparable to SOTA prompt-engineering-based solutions when no content moderation is deployed. However, when there is content moderation, MetaBreak outperforms SOTA solutions PAP and GPTFuzzer by 11.6% and 34.8%, respectively. Finally, since MetaBreak employs a fundamentally different strategy from prompt engineering, the two approaches can work synergistically. Notably, empowering MetaBreak on PAP and GPTFuzzer boosts jailbreak rates by 24.3% and 20.2%, respectively.

Backdoor attacks are a significant threat to large language models (LLMs), often embedded via public checkpoints, yet existing defenses rely on impractical assumptions about trigger settings. To address this challenge, we propose \ourmethod, a defense framework that requires no prior knowledge of trigger settings. \ourmethod is based on the key observation that when deliberately injecting known backdoors into an already-compromised model, both existing unknown and newly injected backdoors aggregate in the representation space. \ourmethod leverages this through a two-stage process: \textbf{first}, aggregating backdoor representations by injecting known triggers, and \textbf{then}, performing recovery fine-tuning to restore benign outputs. Extensive experiments across multiple LLM architectures demonstrate that: (I) \ourmethod reduces the average Attack Success Rate to 4.41\% across multiple benchmarks, outperforming existing baselines by 28.1\%$\sim$69.3\%$\uparrow$. (II) Clean accuracy and utility are preserved within 0.5\% of the original model, ensuring negligible impact on legitimate tasks. (III) The defense generalizes across different types of backdoors, confirming its robustness in practical deployment scenarios.

This study explores the application of deep learning techniques in the automated detection and segmentation of brain tumors from MRI scans. We employ several machine learning models, including basic logistic regression, Convolutional Neural Networks (CNNs), and Residual Networks (ResNet) to classify brain tumors effectively. Additionally, we investigate the use of U-Net for semantic segmentation and EfficientDet for anchor-based object detection to enhance the localization and identification of tumors. Our results demonstrate promising improvements in the accuracy and efficiency of brain tumor diagnostics, underscoring the potential of deep learning in medical imaging and its significance in improving clinical outcomes.