CIE555---Deep-Learning-and-Neural-Networks

This repo contains all my work for Deep Learning and Neural Networks class in Spring 2019.

The projects were implemetned using the following techologies:

• Python
• Numpy
• Pandas
• OpenCV
• Keras
• TensorFlow
• Pytorch

Table of content

1. Model Validation
2. TensorFlow
3. Pytorch
4. Keras Eager Execution
5. Keras Optimization Techniques
6. DenseNet Architecture
7. RNN Time Series Predicting Stock Prices
8. RNN on IMDB and GloVe pretrained weights Rating System

Model Validation

In this project, numpy and pandas were explored in details. Also, forward propagation and backward propagation were implemented from scratch and used on Iris dataset.

TensorFlow

In this project, tensorflow modules were used to train a shallow neural network on both Iris dataset, and Boston dataset for classification and regression.

Pytorch

In this project, Pytorch was explored in details. Moreover, a 2-layer neural network was trained on Iris dataset using Pytorch modules.

Keras Eager Execution

In this project, Keras Eager Execution was used on Breast Cancer dataset by training a 3-layers neural networks.

Keras Optimization Techniques

In this project, Keras modules were used on MNIST dataset withouth CNN to train a shallow neural networks using different optimizers (RMSProp, Adam,SGD).

DenseNet Architecture

In this project, DenseNet101 Architecture was implemented and defined completely.

RNN Time Series Predicting Stock Prices

In this project, RNN model was developed to predict to the stock market prices for the next week with accuracy of 90%.

RNN on IMDB and GloVe pretrained weights Rating System

In this project, RNN model was developed to rate the review based on word embedding. The model was first randomly initialized, then GloVe pre-trained weights were used to boost the accuracy up to 92%.