[Neural Networks and Deep Learning] week1. Introduction to deep learning

[TOC]

What is a neural network?

Example: housing price prediciton.

Each neuron: ReLU function

Stacking multiple layers of neurons: hidden layers are concepts more general than input layer — found automatically by NN.

Supervised Learning with Neural Networks

supervised learning: during training, always have output corresponding to input.

Different NN types are used for different problems:

structured data: database, each feature/column has a well-defined meaning. unstructured data: audio/image/text, no well-defined meaning for pixels/tokens

Why is Deep Learning taking off?

scale drives deep learning progress. (scale: both of NN and of data)

trandition methods: pleateaus as amount of data grows further. NN: grows with data.

  • data scales up
  • computation faster
  • new algorithms, e.g. from sigmoid to ReLU, which in turn speeds up computation too.

About this course

This course: implementing NN.

comments powered by Disqus