Introduction to Neural Networks

Don't panic. If you're watching this video, then you probably want to learn how neural networks work. And [music] to you they might seem like this uh very complex system >> [music] >> that only people with a PhD can understand or even work with. However, I'm here to show you today that that's not quite the case.

>> [snorts] >> In fact, you yourself can absolutely understand how neural [music] networks work. Now, in this video, the goal is to [music] explain how neural networks work and [music] some terminology that is commonly used in the field. It's not to explain how neural networks [music] are trained. That's a complication that only people

who expect to train neural networks in the future will need to know. And if you're one of those people, then after this video, I highly recommend >> [music] >> watching 3Blue1Brown's video series on backpropagation. However, if your goal is just to work with machine learning practitioners in the future, then this video is exactly

made for you. Now, about me. I train neural networks to solve physics problems, [music] such as accelerating simulations. And in my research, I typically have to work with scientists who might not have any understanding machine learning at all. So, >> [music] >> if I sent this video to you directly, then hello. Now, with introductions out of the way,

welcome to Relu. And let's [music] get started by learning about data types. Now, [music] before we talk about neural networks themselves, we have to talk about the type of data you're using. In the same way you wouldn't use a hammer to tighten a screw, the type [music] of inputs you're feeding to your network changes what