esarant / ntua-neural_networks-21

Repo hosting the notebooks for the assignments of the fall 21 "Neural Networks and Intelligent Systems" course @ NTUA

NTUA-Neural_Networks-21

In this repo we host the notebooks from the "Neural Networks and Intelligent Systems" course of NTUA.

During the semester we worked in teams of 3 and we had to prepare 3 labs. The first was on supervised learning, the second on unsupervised learning and the third on deep learning.

The Labs

Lab 1 - Supervised Learning

On the first lab we studied and optimized classifiers on two specific datasets. The first was the Magic GAMMA Telescope from the UCI ML repository and the second was the CS:GO Round winnder classification. On the first dataset we used skicit-learn functions while on the second we had to create our own functions using existing libraries.

Lab 2 - Unsupervised Learning

On the second lab, we worked on a 5.000 movies subset of the Carnegie Mellon Movie Summary Corpus using a Self-Organizing Map (SOM) Neural Network and tf-idf numerical statistic in order to develop a movie recommender system.

Lab 3 - Deep Learning

On the third and fina lab, we expanded upon the TensorFlow tutorial for Image Captioning with visual attention. We worked on a different dataset than that of the tutorial and the goal was to optimize the captions for the images of the dataset

The team

The team consisted of

• Stathis Sarantinopoulos
• Constantinos Costopoulos
• Constantinos Tsaousis