Andrew Ng introduces why studying case studies of convolutional neural networks (convnets) is useful for gaining practical intuition. He motivates reusing effective architectures across vision tasks, outlines upcoming case studies (LeNet-5, AlexNet, VGG, ResNet, Inception), and previews that ResNet enables very deep (e.g., 152-layer) networks and that these ideas generalize beyond computer vision.
Hello and welcome back. This week the first thing we'll do is show you a number of case studies of effective convolutional neural networks. So why look at case studies? Last week you learned about the basic building blocks such as convolutional layers, pooling layers and fully connected layers of convnets. It turns out a lot of the past few years of computer vision research has been on how to put together these basic building blocks to form effective convolutional neural networks. And one of the best ways for you to gain intuition yourself is to see some of these examples. I think just as many of you might have learned to write code by reading other people's code, I think that a good way to gain intuition on how to build confidence is to read or to see other examples of effective confidence. And it turns out that a net neural network architecture that works well on one computer vision task on often works well on other tasks as well, such as maybe on your task. So if someone else has trained a neural network, has or has figured out a neural network architecture that's very good at recognizing cats and dogs and people, but you have a different computer vision task, like maybe you're trying to build a self driving car, you might well be able to take someone else's neural network architecture and apply that to your problem. And finally, after the next few videos, you'll be able to read some, some of the research papers from the field of computer vision. And I hope that you might find it satisfying as well. Uh, you don't have to do this for this class, but I hope you might find it satisfying to read, you know, some of these seminal computer vision research paper and, and see yourself able to understand them. So with that, let's get started. As an outline for what we'll do in the next few videos, we'll first show you a few classic Networks. Um, the Learnet 5 network, which came from I guess the 1980s AlexNet, which is often cited in the VGG network. And these are examples of pretty effective neural networks. And some of the ideas laid the foundation for modern computer vision. And you see ideas in these papers that are probably useful for your own, and you see ideas from these papers that will probably be useful for your own work as well. Then I want to show you the RESNET or called residual network. And you might have heard that neural networks are getting deeper and deeper. The RESNET neural network trained a very, very deep 152 layer neural network and had some very interesting tricks, interesting ideas for how to do that effectively and then finally you also see a case study of the inception neural network. After seeing these neural networks, I think you have much better intuition about how to build effective convolutional neural networks. And even if you end up not working computer vision yourself, I think you find that a lot of the ideas from some of these examples such as residency, inception network, many of these ideas are cross fertilizing or making their way into other disciplines. So even if you don't end up building computer vision applications yourself, I think you find some of these ideas very interesting and helpful for your work.
