Deep Learning

This repository contains my work in the field of Deep Learning. Some of the notebooks relate to the Deep Learning Nanodegree Foundation program, which consists of a bunch of tutorial notebooks for various deep learning topics. These notebooks are mostly implemented in Tensorflow.

The other material is notebooks of tutorials, experiments, hacks and code ideas from various sources. These mostly consist of notebooks implemented in Keras with Tensorflow backend. For these notebooks I have used the following sources as my guide:

Udacity   Coursera   Kaggle   Sebastian Raschka   Jason Brownlee   Sebastian Ruder   Kevin Markham  

Table Of Contents

Tensorflow

• Sentiment Analysis with Numpy: Build a sentiment analysis model, predicting if some text is positive or negative.
• Intro to TensorFlow: Starting building neural networks with Tensorflow.
• Weight Intialization: Explore how initializing network weights affects performance.
• Your First Neural Network: Implement a neural network in Numpy to predict bike rentals.

Keras

Deep Learning Basics

• Deep Learning Network Tutorial: Exploring MLP, CNN and RNNs
• SciKit Learn: Using cross validation and grid search on a simple dataset
• Multiclass Classification: Uses famous iris dataset
• Binary Classification: Predict binary target data
• Regression: Regression of House prices and tuning of network topology

Better Deep Learning

• Network Capacity: Investigate impact of changing model capacity on a complex multiclass dataset
• Batch Size and Gradient Descent: Investigating Batch, Stochastic and Minibatch Gradient descent
• Dropout: Investigate Dropout techniques and evaluate performance on Deep learning model
• Learning Rates: Investigate Learning Rate techniques and evaluate performance on Deep learning model
• Checkpoints: Use Keras API to checkpoint and save model weights
• Training History: Use Keras API to display training and test history
• Early Stopping: Use Keras API to employ early stopping on a dataset

NLP

• Preparing Text Data: Preparing and applying scikit-learn and keras vectorizers to text data
• Movie Reviews: Formatting IMDB movie review data ready for analysis
• Sentiment Analysis: Sentiment analysis on the prepared movie reviews using Bag of Words model.
• Model Validation: Model validation of the trained movie reviews
• Language Model: Tring to predict text using LSTMs
• IMDB Sentiment Analysis: Basic CNN and NLP on the IMDB datset
• Quora Kaggle Challenge: First Attempt at Kaggle Quora challenge using word embedding and CNN model

CNN

• Keras Covnets: Using the Keras Framework tools to process images
• Pre-Trained Covnet: Using the Keras Framework tools to process images using pre-trained covnet
• Whale Id: First attempt at Kaggle Whale Id challenge

LSTM

• Prediction: Trying to predict airline passeneger numbers with LSTMs

Audio

• Audio Classification Part1: Looking at the audio classification of street sounds