Pinned Loading

1. Face-Recognition Face-Recognition Public

   This is a Conv2D model based on MobileNetV2 to first create a facial attribute clasificator which then was used as a pretrained model for a face recognition code. It has an accuracy of ~85% (using …

   Jupyter Notebook 1

2. Functions-ODE-CustomLayers Functions-ODE-CustomLayers Public

   These are a serie of codes which were used to practice the creation of custom layers using keras: To create a B&W filter; to train a ANN to compute functions, get the coeficients of a polynomial an…

   Jupyter Notebook

3. Higgs-Boson-SignalnBackground-Classificator Higgs-Boson-SignalnBackground-Classificator Public

   Here I present a signal and noise Residual ANN model classificator for particle collisions Delphes sumulations. I explain how to work with .ROOT data files using the Uproot library and how to creat…

   Jupyter Notebook 1

4. Muon-M3 Muon-M3 Public

   Here I present a code to implement the new Google Research's proposed optimizator Multi-Scale Momentum (M3) which goal is to support the effectiveness of continuum Memory System in optimizers. I tr…

   Python

5. Plant-Pathology-Classificator Plant-Pathology-Classificator Public

   These are two CNN models trained to classificate images from a dataset with different diseased and healthy leaves. One is a usual Conv2D (88.27% acc) and the other is based in a ResNet101 pretained…

   Jupyter Notebook