Azyri.Advancing science.Across AI and Humanity.
Research
Computer Science · Computer Vision and Pattern Recognition · 2026
Decoding the Past: An Uncertainty-Aware Deep Learning Framework for Sex Attribution in Prehistoric Hand StencilsJournal: Journal of Archaeological Science
Karel Becerra · Boris Mederos · Dean Snow · Ramón A. Mollineda
- Uncertainty-aware computer vision pipeline.
- Large-scale DNN training + cross-domain inference.
- Deep ensemble with CNN + Transformer architectures.
- Ensemble learning explicitly models epistemic uncertainty.
- Representation learning + UMAP + k-NN provides independent ML validation.
- Computer vision, uncertainty quantification, deep ensembles, CNN/Transformer architectures, representation learning, manifold learning, non-parametric ML, and XAI
Research
Computers in Biology and Medicine · 2025
Sex classification from hand X-ray images in pediatric patients: How zero-shot Segment Anything Model (SAM) can improve medical image analysis
Ramón A. Mollineda · Karel Becerra · Boris Mederos
- A zero-prompt strategy for hand segmentation in X-ray images based on zero-shot SAM.
- A thorough sex classification study from original and segmented hand X-ray images.
- A low-cost deep learning solution for sex classification from hand X-ray images.
- Visual explanation of model decisions from the aggregation of attention maps.
- Novel use case in biological anthropology: sex estimation of prehistoric handprints.