Fast.ais

Practical Deep Learning for Coders (fast.ai courses) These are the lecture materials from Practical Deep Learning for Coders. Two important parts of the course are our online forums and our wiki. If you are encountering an error, we recommend that you first search the forums and wiki for a solution. If you can’t find the answer there, the next step is to ask your question on the forums. See this advice on how to ask for help in a way that will allow others to most quickly and effectively be able to help you. »

Lesson 4 This week’s links The lesson video Lesson 4 notes Lesson 4 Timeline The notebooks: State Farm sample State Farm full Collaborative filtering with Movielens (dataset link below) The spreadsheets: Collaborative filtering Convolution example SGD implementations Movielens small dataset Movielens full dataset Kaggle fish classification competition; also, a Fortune article discussing the background behind this competition Information about this week’s topics Collaborative filtering: »

Lesson 5 Tips to get 98.94 acc on Cats and Dogs Redux 00:00:00 [TODO] Summarize the model, techniques and hyperparameters Introducing Batch Normalization into a Pre-Trained Model 00:01:55 Before we move on from CNN’s, we wanted to talk about how we can introduce Batch Normalization into a pre-trained convolutional neural network that doesn’t already have it. Batch Norm Review As a reminder, batch normalization has become something of a standard now because it increases training speed and tends to reduce overfitting. »

Lesson 6 This week’s links The lesson video and timeline Brief lesson 6 notes The notebooks: Lesson 6 shows the RNNs we built in keras and theano The use of ‘pseudo-labeling’ with the new MixIterator will be shown in the Lesson 7 notebook applied to the Fisheries competition. Information about this week’s topics Introduction to Theano Theano convolution tutorial Designing great data products - Jeremy’s paper on how to use predictive modelling to optimize for actions A Simple Way to Initialize Recurrent Networks of Rectified Linear Units - Geoffrey Hinton et al. »

Lesson 7 Resnet: We’re going to start by introducing one of the most important developments in recent years: Resnet. Resnet won the Imagenet competition in 2015, and it’s an incredibly simple and intuitive concept. To start, we’re going to use Resnet to perform things we did with Vgg16, such as image classification. There are different size Resnets, but for our purposes we’re going to use 50 because it works well. »

AWS Setup Setup source aws-alias.sh aws-start aws-ssh Shutdown aws-stop »

tmux tmux allows you to create multiple tabs and panes within a single terminal window. It can also keep programs running after you disconnect (e.g. when you ssh into an AWS instance) Tutorials A Quick and Easy Guide To Tmux Cheatsheet #1 Cheatsheet #2 Terms Session - Saved collection of windows and panes and programs you left running inside Attaching - Enter a running session Detaching - Exit a running session (without killing it) Window - Similar to a tab in your browser. »

Kaggle CLI

For lesson 1, competition is dogs-vs-cats-redux-kernels-edition: pip install kaggle-cli kg config -g -u `username` -p `password` -c `competition` kg download unzip train.zip unzip test.zip Links http://wiki.fast.ai/index.php/Image_Datasets »