Python for Data Science Advanced: Fundamentals of Machine Learning

Transcript of the free preview lesson. Remaining lessons unlock with the full course.

Hello, my name is Nizhar Dajani, and in these lessons, you will learn about the fundamentals of machine learning, its history, the different types, as well as the machine learning process and data properties. You will also learn about some of Python's popular machine learning modules, like scikit, pytorch, and tensorflow. Now in this lesson, you will learn the fundamentals of machine learning. So let's begin by defining what machine learning is in a very simple way. Machine learning is like teaching a computer to be really smart and learn things on its own, just like you and I learn new things. For example, imagine you have a robot friend and you want to teach it to recognize different fruits, like apples and bananas. At first, you show the robot many pictures of apples and bananas, and you tell it which one is which. The robot looks at the pictures and tries to find things that are the same in all the apple pictures and different from the banana pictures. So it might notice that apples are usually red or green and they have a round shape, while bananas are usually yellow and long. So machine learning is like teaching the robot to figure things out on its own by showing it many examples and it learns from these examples and can make smart guesses even when it sees something new. And just like you and I learn from our experiences and can recognize things and get better at doing so, machine learning helps computers do the same with the help of lots of data and some special tricks. Now I will share with you some definitions of machine learning from various sources. The first definition is machine learning is the field of study that gives computers the ability to learn without being explicitly programmed. This quoted definition is by Arthur Samuel, who was one of the early researchers and pioneers in machine learning. He highlights the idea that machine learning allows computers to learn on their own without having every detail programmed by a human. The next definition is a computer program is set to learn from experience E with respect to some class of tasks T and performance measure P if its performance at tasks in T as measured by P improves with experience E. So this definition is by Tom Mitchell, who is another pioneer in machine learning and he emphasizes that machine learning is about computers improving their performance on tasks through learning from data. The next definition is by Microsoft and it says machine learning is the practice of using algorithms to analyze and interpret data, discover patterns, and make more informed decisions or predictions. This definition by Microsoft simplifies the concept by focusing on using algorithms to analyze data and make better decisions. And last but not least, the following definition is from Google. Machine learning is the study of algorithms and statistical models that computer systems use to perform tasks without using explicit instructions, relying on patterns and inference instead. So as you can see, Google's definition highlights the reliance on patterns and inference rather than explicit instructions. So all these definitions capture the essence of machine learning as a field focused on enabling computers to learn and make decisions from data, often without explicit programming. It also involves the development of algorithms and models that can improve a computer's performance through experience and data analysis. Now in the field of machine learning, the terms, algorithms, and models are commonly heard and commonly used. They are closely related, but they refer to two different concepts. So I wanted to clarify both of them first. So the computer learns from data using the algorithm and creates a model to help it do things like making predictions. And on a more formal definition, an algorithm in machine learning refers to the specific set of rules, procedures, or computational steps that a machine learning system follows to learn from data and make predictions or decisions. We can also say that machine learning algorithms are responsible for the learning process itself. So in essence, algorithms are the core computational engines that enable machine learning systems to generalize from data and make predictions on new, unseen data. And for the formal definition of a model, we can say that a model in machine learning is the output or result of running a specific machine learning algorithm on a data set. It represents the learned patterns or relationships within the data. In other words, a model is a mathematical or computational representation of the underlining patterns in the training data that the algorithm has learned. This representation can be in the form of equations, graphs, decision boundaries, or neural network architectures depending on the algorithm used. And once a model is trained, it can be used to make predictions or classifications on new unseen data. For example, a trained image classification model can identify objects in images similar to the apples and bananas example I used earlier. So models are what enable machine learning systems to perform tasks and solve real world problems by leveraging the knowledge extracted from the training data. So in summary, we can say that machine learning algorithms are the techniques and procedures used to learn from data, while models are the learned representations of data that are created by running an algorithm on a data set. Models encapsulate the knowledge acquired during the learning process and are used for making predictions or decisions on new data. Different algorithms can produce different types of models, each suited to specific types of tasks and data sets. So in this lesson, you learned about the fundamentals of machine learning, stay tuned for the next lesson where you will learn about the history of machine learning. Thanks for watching.