ACALBERTO CASTELO

Projects

Showcase your projects with a hero image (16 x 9)

Neuroevolution of Bayesian Neural Networks

Using Neuro Evolution to train and architecture-search in Bayesian Neural Networks (Feed-forward) for Supervised Learning tasks: Regression and Classification.

Learn more →

Bayesian Deep Learning for Autonomous Image-based Medical Diagnosis

In this project, I was curious to compare interesting approaches in Bayesian Deep Learning for Computer Vision: * Use a pretrained ResNet model and stack a final Gaussian Process layer to provide the uncertainty estimation. This model was build on top of GPytorch. * Consider a Convolutional Bayesian Neural Network and train it from zero using Bayes-to-backprop method.

Learn more →