fast-ai

Tags: #course #machine-learning #programming

• ==Deep learning is a computer technique to extract and transform data by using multiple layers of neural networks==
• What I’ll learn in this class
  • How to train models that achieve state-of-the-art results in computer vision, natural language processing, tabular data, and collaborative filtering
  • How to turn and deploy models into web applications
  • Why and how deep learning models work
  • Latest deep learning techniques that really matter in practice
  • Implement stochastic gradient descent and a complete training loop from scratch
  • Techniques
    • random forests and gradient boosting future
    • affine functions and nonlinearities future
    • parameters and activation future
    • random initialization and transfer learning future
    • SGD, momentum, Adam, and other optimizer future
    • convolutions future
    • batch normalization future
    • [ ] dropout future
    • [ ] data augmentation future
    • [ ] weight decay future
    • [ ] image classification and regression future
    • [ ] entity and word embedding future
    • [ ] recurrent neural networks (RNNs) future
    • segmentation future
• Tools
  • Will be learning these concepts using fastai, fastcore, pytorch and the gradient environment
• Lesson 1