Deep-Learning-ND-Projects

Projects completed related to Udacity Deep Learning Nanodegree Foundations Program

1. Project 1 - Neural Net for Bikeshare usage prediction. Basic 3 layer Feedforward Neural Net with backpropagation built using Python Numpy and Pandas
2. Project 2 - CNN for image recognition on CIFAR-10 dataset coded in Python using Tensorflow. ((Conv-Relu)*2-MaxPool)*3-Flatten-(FC-Dropout)*2-Output architecture. ~67% accuracy of testset after training
3. Project 3 - LSTM RNN to generate Moe's Tavern scripts from The Simpsons based on a training set of scripts from the actual series. Built in Tensorflow.
4. Project 4 - SeqToSeq model based on LSTM cell for translating simple sentences from English to French. Built in Tensorflow.
5. Project 5 - Convolutional GAN to generate realistic faces, trained on CelebA dataset. Built in Tensorflow.