Deep-Learning-Projects

Projects of Neural Networks Graduate Course under Supervision of Prof. Reza Safabakhsh - AUT - Fall 2021

Project 1:

Goal: getting familiar with multi-layer perceptron neural networks.

• In this project we worked on “seeds” dataset. First we examine the linear separability of data. Then we try to classify this data using multi-layer perceptron and our goal is to find best structure for network which reaches highest accuracy. We also get familiar with concept of overfitting and regularization by trying to overfit neural network on dataset. Finally, we operate regularization on dataset in order to remove overfitting effect.

project 2:

Goal: getting familiar with kohonen self organizing maps (SOM Neural Netwok)

• in this project we illustrate rcv1 dataset with SOM (self-organizing map) and we plot a feature map for data.
• in next step we cluster dataset using SOM and find purity for our clustering.

project 3:

Goal: getting familiar with RBF Neural Networks

project 4:

Goal: getting familiar with Convolutional Neural Network

project 5:

Goal: getting familiar with Transfer Learning

project 6:

Goal: getting familiar with Generative Adverserial Networks