This paper introduces a novel approach for detecting synthetic fingerprints, framing it as a continual few-shot adaptation problem. It proposes using a combination of binary cross-entropy and supervised contrastive losses with experience replay to enable rapid adaptation to new synthetic data styles while mitigating catastrophic forgetting.

This paper investigates how to effectively utilize Vision-Language Models (VLMs) for image forgery detection and localization (IFDL). It introduces IFDL-VLM, a novel two-stage pipeline that decouples the core IFDL task from VLM-based explanation generation, leveraging localization masks to significantly enhance VLM interpretability and achieving state-of-the-art results across multiple benchmarks.

This paper introduces an artificial intelligence system (AIOC) designed to accurately detect fetal orofacial clefts from ultrasound images and simultaneously enhance the expertise of radiologists. Trained on a large, multi-center dataset, AIOC demonstrates expert-level diagnostic performance and significantly improves junior radiologists' sensitivity, offering a scalable solution for early diagnosis and medical education in resource-limited settings.

This paper introduces AIRT, an end-to-end deep-learning framework that generates deliverable single-arc VMAT prostate treatment plans in less than one second. The method achieves plan quality comparable to clinical systems, significantly enhancing the efficiency and consistency of radiation therapy planning.

This paper introduces an MLOps-assisted framework leveraging Red Hat OpenShift AI (RHOAI) for the automated design of anomalous reflector metasurfaces. It employs a conditional Generative Adversarial Network (cGAN) with a surrogate model to efficiently create high-quality freeform metasurface designs.

This paper provides a comprehensive review of deep learning-based multimodal data fusion in astronomy. It discusses the motivation for integrating diverse astronomical data, outlines various data sources and modalities, and introduces representative deep learning models and fusion strategies to enhance understanding of the universe.

This paper introduces a transformer-based AI system utilizing the RT-DETR model to accurately detect fungal elements in potassium hydroxide (KOH) microscopy images for dermatophytosis, effectively distinguishing them from common artefacts. The system demonstrated high sensitivity and accuracy, suggesting its potential as a reliable automated screening tool in dermatomycology.

This paper proposes a secure and distributed system for lung disease diagnosis, specifically COVID-19 and Pneumonia, using a hybrid federated learning-enabled ensemble model. It combines established CNN architectures with the SWIN Transformer to enhance diagnostic accuracy and ensure patient data privacy through federated learning.

This paper investigates the use of eXplainable Artificial Intelligence (XAI) methods, Integrated Gradients (IG) and SHAP, to interpret fault diagnosis decisions made by a Long Short-Time Memory (LSTM) classifier in a chemical process. The study highlights how XAI can help identify the faulty subsystem, improving the trustworthiness of deep learning models in industrial settings.

This paper introduces an explainable deep-learning framework for the accurate classification of microplastic and natural microfibers using polarization-resolved digital holographic microscopy. By extracting 72 polarization-based features and employing a deep neural network with SHAP analysis, the method achieves high accuracy and identifies key optical properties for distinguishing fiber types.