Hi, and welcome to Math for Machine Learning and Data Science. Concepts like vectors, matrices, gradients, optimization, priority distributions, p-values, occur a lot in machine learning. When you learn about these concepts, you gain a deeper understanding of how and why machine learning algorithms work. And when you understand the math behind machine learning, you'll be able to better go beyond just using out-of-the-box machine learning algorithms to build and customize the models more, and you'll also be better able to judge when to apply which technique. Through learning the math of machine learning, you also become more effective at debugging machine learning algorithms. When I train a machine learning algorithm, it pretty much never works the first time or the first few times I run it. And so the math has frequently helped me, and I'm sure helped you too, to make better decisions about how to efficiently improve the performance of your machine learning algorithms. And who knows, maybe with the mathematical foundation that you gain from this specialization, I hope that you might someday use that skill to even invent new machine learning algorithms. I'm thrilled to introduce to you the instructor for this specialization, Luis Serrano. When DeepLearn.ai set out to create this specialization, I hoped to find an instructor who could bring math to life with visual examples that convey the intuition behind these math concepts. We spoke with many people to try to find the best instructor for these concepts, and I was thrilled when we found Luis, who turned out to be a fantastic fit. Luis is a machine learning engineer, researcher, and an educator with a PhD in pure math. He was a machine learning engineer at Google, where he worked on the YouTube recommended system, and he was also a lead AI educator at Apple. Luis also holds a popular YouTube channel called Serrano Academy, where he puts together math and machine learning concepts in a really illustrative, really delightful visual style. He's also the author of a best-selling book, Rocking Machine Learning. With Luis teaching the specialization, I'm confident you're in good hands, and you enjoy learning these materials about math and machine learning and data science. So I'm thrilled to have you with us, Luis. Well, thank you very much, Andrew. It's great to be here, and it's actually funny because it was about 10 years ago that I started hearing the term machine learning, and I remember I got my computer and Google searched introduction to machine learning courses. I thought it was going to be a very complicated thing, but I looked, and the first hit was your class. So I started taking your class, and I was blown away by how wonderful and simple it was. So thank you for that. And for me, it's very interesting that 10 years later, I'm on the other side of the camera talking to you about machine learning courses. So it's great to be here. Thank you. Thank you. So glad to have you here. And in fact, I did not know that story, but it actually makes me reflect. You know, when you're watching this, learning math from Luis, maybe a few years from now, maybe 10 years, maybe even faster, if you find it in you to teach something online, I think that could be very cool too. Yeah, you'll be on the side of the camera. We have another chair. So this specialization has three courses, and the first one, which is just four weeks long, is on linear algebra. Do you want to say a bit more about what the first course of the three is about? Yes. So the first course is linear algebra, and in this course, we cover anything regarding vectors, matrices, linear transformations, systems of equations, determinants, etc. But we like to see it in a different light. So for example, a matrix can be seen as an array of numbers, but that's like seeing a book as an array of letters. Books have a lot of interesting stories and can take you to places in your mind, and matrices are the same. They can be seen in much, much deeper ways. In fact, one thing I've heard you say is I think that if you look at how a neural network or some learning algorithm manipulates data, a lot of a neural network is built on top of matrix operations to take the data set, rotate it, turn it, warp it, and multiple times with multiple layers of a neural network, so that eventually you get out some answer, like predicting, you know, is this a cat or a dog? And that's what neural networks do, right? They're just a bunch of matrices with activations in the middle that warp space. Yeah. So I think, you know, given that so many of the most important learning algorithms, including neural networks, but many others, are built on top of matrix operations, gaining that deeper intuition about how it all works will give you a better sense of how neural networks and many other learning algorithms actually, you know, do the magic that they do. Exactly. And then the second course is on calculus. Do you want to say a bit more about that? Yes. So calculus is very important in machine learning for many things, but one that's fundamental is maximizing and minimizing functions. Yeah. In fact, the vast majority of learning algorithms are created by creating some cost function and then minimizing it. So knowing how to do that is fundamental to a lot of learning algorithms. And so I found that when I'm using a minimization algorithm, you know, be it gradient descent or some more advanced algorithm like Adam, I think that if you can picture in your head what's the derivative calculation doing, then if it isn't working as well as you want, I find that I'm better able to tune the algorithm to make it do a better job solving this critical minimization task. Yeah, absolutely. Knowing a lot about the calculus and the derivatives can really help demystify these optimizers that at first may look obscure. And then finally, the third course of the specialization is on probability and statistics and even things like hypothesis testing and p-value. So tell us more about the third course. Yeah, the third course is very interesting because a lot of machine learning happens to be probabilities, right? Many times a model outputs a probability and probability can be used to train models. For example, maximum likelihood estimation is very important. In maximum likelihood estimation, what you want to do is let's say you have some evidence and you want to find the scenario that most likely generated this evidence. That's machine learning. Your evidence is your data and you want to find the model that most likely generated this data. So you want to maximize the probability that the model generated that data. So sometimes I've said when it comes to math, I sometimes said, don't worry about it. And actually, stand by that. When you're learning machine learning for the first time, to get it to work, sometimes you don't need to worry about the intricacies of exactly how the math works. But as Lewis was saying, to get to that next level of expertise, when you can then start to gain that deeper understanding of that math as well, you can then learn a better, deeper mastery of the algorithms that you're using in building. And then one exciting element of this specialization is, you know, learners don't just learn the math. You also get to see it in Python code and see it run. Do you want to say more about that? Yes, that's absolutely right. You will be able to put these algorithms in practice. And for that, we have a bunch of Python labs. So we assume some basic knowledge in Python to be able to run them. But we're also going to have the resources for you to get up to speed. Awesome. So this specialization will go from, you know, high school, not very advanced level of math all the way up to the core concepts of math for machine learning and data science. And so with that, I'm excited to have you jump in to this specialization. So please go on to the next video to dive into linear algebra.