tda-nn-analysis
A collection of experiments that use persistent homology and topological data analysis to study how knowledge transfers across CNN architectures and optimization landscapes.
Pablo Olivares
ML engineer · data alchemist · tinkerer
I turn raw data into insight, build things with PyTorch and curiosity, and cook up experiments in the lab. Currently engineering ML at Santander.
Loc. Madrid, ES Ver. 2.0 Date 2026.07.28 Scale 1:1
System status Currently reading: The Book of Why
Featured experiments
All experiments → 
feature-weight-metaheuristics
Bioinspired optimization algorithms tuned to learn feature weights that improve 1-nearest-neighbour classification accuracy.
granada-parking-segmentation
Deep segmentation models applied to aerial imagery for detecting parked vehicles across Granada, comparing U-Net, PSPNet and DeepLab v3+.