Overview of Udacity’s Deep Learning Course

Ok so here it is, the full program. I’m so excited it’s hard to describe.

Program Structure

The Deep Learning Nanodegree Foundation program is divided into five parts covering various topics in deep learning. The first two parts are available immediately upon starting the program. The other parts are released every two weeks.


The first part is an introduction to the program as well as a couple lessons covering tools you’ll be using. You’ll also get a chance to apply some deep learning models to do cool things like transferring the style of artwork to another image.

We’ll start off with a simple introduction to linear regression and machine learning. This will give you the vocabulary you need to understand recent advancements, and make clear where deep learning fits into the broader picture of ML techniques.

Neural Networks

In this part, you’ll learn how to build a simple neural network from scratch using Numpy. We’ll cover the algorithms used to train networks such as gradient descent and backpropagation.

The first project is also available this week. In this project, you’ll predict bike ridership using a simple neural network.

You’ll also learn about model evaluation and validation, an important technique for training and assessing neural networks. We also have guest instructor Andrew Trask, author of Grokking Deep Learning, developing a neural network for processing text and predicting sentiment.

Convolutional Networks

Convolutional networks have achieved state of the art results in computer vision. These types of networks can detect and identify objects in images. You’ll learn how to build convolutional networks in TensorFlow.

You’ll also get the second project, where you’ll build a convolutional network to classify images of frogs, planes, cars, and more.

You’ll also use convolutional networks to build an autoencoder, a network architecture used for image compression and denoising. Then, you’ll use a pretrained neural network (VGGnet), to classify images of flowers the network has never seen before, a technique known as transfer learning.

Recurrent Neural Networks

In this part, you’ll learn about Recurrent Neural Networks (RNNs) — a type of network architecture particularly well suited to data that forms sequences like text, music, and time series data. You’ll build a recurrent neural network that can generate new text character by character.

From Andrej Karpthy's [blog post on RNNs](http://karpathy.github.io/2015/05/21/rnn-effectiveness/)

From Andrej Karpthy’s blog post on RNNs

Then, you’ll learn about word embeddings and implement the Word2Vec model, a network that can learn about semantic relationships between words. These are used to increase the efficiency of networks when you’re processing text.

You’ll combine embeddings and an RNN to predict the sentiment of movie reviews, an example of common tasks in natural language processing.

In the third project, you’ll use what you’ve learned here to generate new TV scripts from episodes of The Simpson’s.

The fourth project will have you training a network that can translate from English to French using a sequence-to-sequence model. This is a special type of RNN that reads in a sequence of data and generates another sequence.

Generative Adversarial Networks

Generative adversarial networks (GANs) are one of the newest and most exciting deep learning architectures, showing incredible capacity for understand real-world data. The networks can be used for generating images such as the CycleGAN project.

GAN example from the [pix2pix project](https://github.com/phillipi/pix2pix).

GAN example from the pix2pix project.

The inventor of GANs, Ian Goodfellow, will show you how GANs work and how to implement them. You’ll also learn about semi-supervised learning, a technique for training classifiers with data mostly missing labels.

In the fifth project, you’ll use a deep convolutional GAN to generate completely new images of human faces.

Images created for the fifth project by a student!

Images created for the fifth project by a student!

Additional Lessons

In this part, you’ll learn about topics such as TensorBoard for visualizing TensorFlow graphs, as well as training a neural network to play computer games with reinforcement learning. This part will also contain future lessons on interesting topics in deep learning.

Karl Niebuhr

Karl Niebuhr

I'm Karl! I like to read, learn and share knowledge
Karl Niebuhr

Also published on Medium.