It is my project based on Python 3, Tensorflow, and Scikit-learn for the course Hybrid Systems of Center of Informatics of UFPE (2016.1). This course is related to my Master's Degree in Deep Learning. The idea is to extract features from the MNIST dataset and then use SVM to classify their images. We show that it improves the accuracy compared to the regular CNN and is faster than ELM of worst accuracy.
To run it, you should type:
python main.py