Deep Learning Nanodegree Foundation

Sentiment Analysis with Andrew Trask

This is a project from a course on how to build a neural network to predict sentiment, positive or negative in movie reviews. The couse was lectured by Andrew Trask, a Phd student at the university of Oxford and author of the book Grooking Deep Learning, in the Deep Learning Nanodegree Foundation from Udacity.

Contents

The course is separated in 6 mini projects:

• Mini Project 1 - Correlate the words to labels
• Mini Project 2 - Creating the input/output data
• Mini Project 3 - Building the neural network
• Mini Project 4 - Reducing the noise in the input data
• Mini Project 5 - Improving the efficiency of the network
• Mini Project 6 - Further noise reduction

Methods and Frameworks utilized

• NumPy - a fundamental package for scientific computing in python
• Pandas - an ease-to-use python library for manipulating data structures and performing data analysis
• Jupyter Notebook - tool that allow the creation of documents with live code
• Bag of Words & Word2vec - methods for transforming text into vectors in order to be used as input in the network

Installation

Install Anaconda and run the code below.

conda env create -f environment.yaml
activate sentiment_andrew_intro
jupyter notebook