AI Learns to Read Dangerous URLs Character by Character While Showing Its Work
Researchers have developed X-URL, a character-level deep learning framework that classifies malicious URLs into four threat categories with over 98 ...
Researchers have developed X-URL, a character-level deep learning framework that classifies malicious URLs into four threat categories with over 98 ...
A two-stage Vision Transformer–U-Net pipeline restores reliable detection of rare voids in concrete X-ray CT images where single-stage deep learning ...
A new safety audit shows that machine unlearning in medical AI can silently degrade performance on rare diagnoses, and proposes ...
A new Cluster Computing study using stroke prediction shows that federated learning in healthcare will fail unless it confronts severe ...
A new study of three undergraduate courses shows that Moodle engagement predicts grades with up to 97 percent accuracy in ...
A new graph-attention-based generative adversarial network called GARAGE produces realistic synthetic single-cell RNA-seq data that preserves rare cell types and ...
A new classifier called CABLE-kNN caches local neighborhood evidence during training and neutralizes class priors, beating ordinary kNN on imbalanced ...
Researchers have introduced DACD, a dynamic adaptive learning method that detects concept drift in streaming data with a window-based CUSUM ...
A new study finds that the YOLO12s deep learning model detects mitotic figures in whole-slide breast cancer images with F1-scores ...
Researchers have developed CL3AN-GNN, a graph neural network that borrows curriculum learning from education to progressively disentangle features and boost ...
© 2026 Scienmag - Science Magazine
© 2026 Scienmag - Science Magazine