Chapter1Notes.Rmd

---
title: "Chapter1"
author: "Julin Maloof"
date: "11/3/2020"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

## Book Notes

## Chapter 1: what is deep learning?

### AI, Machine Learning, Deep Learning

__AI__ automate intellectual tasks normally performed by humans.

Early efforts attempted to make a bunch of rules for the machine to follow. e.g. early chess programs, etc.

But these fail as the tasks get harder.

__Machine Learning__

Can a computer learn the rules from the data without us telling us what the rules are?

Input data and answers and have the computer figure out the rules, instead of the classic paradigm of input data and rules and compute the answer.

Need to find appropriate _Representation_ of the data in order to learn from it.

Learning: automate the search for better representations.

__DEEP learning__

successive layers of representations

Number of layers == _depth_

The layers in deep learning are usually neural networks

Each layer has _weights_ which are essentially parameters for the layer.  There can be millions of weights.

We use a _loss function_ to determine how far away the predictions are from the target (truth).  We will want to minimize the loss function.

The _optimizer_ adjusts the weights to improve the loss function via backpropogation

## History of Machine Learning