The projects in this repository were created while working on the course Machine Learning and Natural Language Processing for my postgraduate studies. They include the following:
-
Supervised learning (Classification): examining different financial performance indicators of Greek companies and creating a classification model to identify those that will go bankrupt.
-
CNN architecture: development of a CNN model to be used on the CIFAR-10 dataset. The development includes experimentation with different loss functions and a more complex model structure.
-
Unsupervised learning (Clustering): development of deep learning models and clustering techniques to be used on the fashion-mnist dataset (https://github.com/zalandoresearch/fashion-mnist). Distinction between clustering of raw data and the use of high level feature values extracted using stacked autoencoders.
-
NLTK Project (Creation of sentences and classification of movie reviews): Use of the Python NLTK library to create sentences using bigrams and trigrams. In addition, using the NLTK movie_reviews dataset, training a classifier to correctly classify positive and negative reviews.