Module 13 - Deep Learning

Overview

During this week, we continue to learn about neural networks, moving from the basics of individual neurons to challenges that arise in deep architectures. We introduce convolutional and recurrent neural networks, discuss the advantages of inductive bias for certain problems, and how to enhace the amount of data using random transformations. Image identification will be the primary problem archetype this week.

Lab 7 is due at the end of the week.

Learning Objectives

  • CNN and RNN Architectures
  • Inductive Bias
  • Data Enhancement
  • Hyperparameter Choices for Stochastic Gradient Descent

Readings

  • ISLP (Introduction to Statistical Learning): 10.3-10.8

Videos