Course Introduction: Fully Connected Neural Networks with Keras
In this course, we'll build three different neural networks with Keras, using Tensorflow for the backend. Keras is a high level API for building neural networks, and makes it very easy to get started with only a few lines of code.
You don't need to know a bunch of math to take this course, and we won't spend a lot of time talking about complicated algorithms - instead, we'll get straight to building networks that you can use today.
By the end of this course, you will be able to build a neural network, train it on your data, and save the model for later use. Then, you'll be able to load up your model, and use it to make predictions on new data!
Instructor: [00:00] Neural networks can be a powerful way to make predictions with data. Learning how to build them can be extremely confusing, and many courses use a lot of complicated math to explain what is happening. In this course, we'll make three different types of neural networks using the Keras API with TensorFlow on the backend.
[00:17] Keras is a high level API that lets you get started with neural networks quickly. We don't have to review a lot of math or learn a lot of complex algorithms before we start seeing real results with our networks.
[00:30] We'll start with a neural network designed to predict a single continuous number, which could be used to answer questions like, "What price will this house sell for?" or "How many customers will this restaurant get tonight?"
[00:40] Then we'll switch to a binary classification network, which can make predictions between two options, or answer yes or no questions.
[00:49] The network can answer questions like, "Will this customer buy our product?" or "Is this cancer or not cancer?" It can give you a probability score for each answer.
[00:59] Finally, we'll switch to a multiclass classification network, which can be used to distinguish between any number of classes. This is useful to answer questions like, "What type of flower is this?" or "What genre of movie is this?"
[01:13] Along the way, we'll learn how to train these networks, how to split our data into validation and test sets, how to save the model for later use, and how to use the model to make predictions on new data.
[01:23] We'll also cover a variety of options you have when setting up your networks, like how many layers the network has, and which optimizer and learning rates to choose.
[01:32] I'm excited to present this course because with just a little bit of work, it's possible to make a network that can give you real answers with just a small amount of data. You don't need a million data points to get real results. You don't need an advanced degree in math to get started with neural networks.
[01:47] It's amazing how you can make powerful neural networks in just a few lines of code. I'm excited to show you exactly how to do that.