Hello, and welcome to section four on deep learning. In the last section we met linear regression and logistical regression and used them to solve both classification and regression problem. These models share a lot of elements in common. For example, they are both formulated as a hypothesis with parameters and they're both optimized, they both learn from examples by minimizing a cost function. However, these models have some limits. They don't work well in non-linear cases. So to extend these models and work with more complex cases we'll need other techniques. What is awesome about neural nets is that they provide a unified framework to deal with all cases. So you'll be able to do regression and classification you'll be able to do linear and non-linear problems all within the same framework of neural nets. So we starting this chapter to introduce them with the simplest possible neural network it's called the perceptron. We will see some elements like weights, biases nodes layers and activation function. In the end, we'll use it to perform a simple classification and see how it works. So let's get started with section four, deep learning.