Batch normalization: adjust the parameterization of a model in order to make the loss surface smoother.Data augmentation: modify model inputs during training in order to effectively increase data size.Dropout: remove activations at random during training in order to regularize the model.Lesson 6 - Regularization Convolutions Data ethicsĭiscuss some powerful techniques for improving training and avoiding over-fitting: Look inside the weights of an embedding layer, to find out what our model has learned about our categorical variables.Īlthough embeddings are most widely known in the context of word embeddings for NLP, they are at least as important for categorical variables in general, such as for tabular data or collaborative filtering. Lesson 5 - Back propagation Accelerated SGD Neural net from scratch Work with the fastai.tabular module to set up and train a model.Ĭollaborative filtering (recommendation systems).Īn “embedding” is simply a computational shortcut for a particular type of matrix multiplication (a multiplication by a one-hot encoded matrix e.g. Then fine-tune this model for the final classification taskĬover tabular data (such as spreadsheets and database tables). Remove the encoder in this fine tuned language model, and replace it with a classifier.Fine-tune this language model using your target corpus.Create (or use pretrained) language model (predict the next word of a sentence).Here's a popular science article on the model Predict whether a movie review is positive or negative using ULMFiT. Lesson 4 - NLP Tabular data Collaborative filtering Embeddings Predict face keypoints (interesting areas) Image segmentation - process of labeling every pixel in an image with a category that shows what kind of object is portrayed by that pixel. Use the data block API to get the data into shape (more info here). Lesson 3 - Data blocks Multi-label classification Segmentation Using the model to find and fix mislabeled or incorrectly-collected images.Ĭreate a model and our own gradient descent loop. Lesson 2 - Data cleaning and production SGD from scratch ![]() ![]() Set the most important hyper-parameter when training neural networks: the learning rate, using Leslie Smith’s fantastic learning rate finder method.įeatures that fastai provides for allowing you to easily add labels to your images. The videos can be found in a YouTube playlist. The course uses pytorch and the fastai wrapper. Image localization (segmentation and activation maps).Notes on how to setup the course in Azure here: Notes on how to setup the couse in GCP here: Seven lessons, each around 2 hours long, and you should plan to spend about 10 hours on assignments for each lesson. And no document has been created in this tutorial, only the equation is presented in front of you.A blog post on what you need for deep learning: ![]() Hopefully, you understand how to use the plus-minus or minus plus symbol with the help of latex. For example, when you determine the square root of a number, you have to use the plus-minus sign.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |