In this chapter, we will introduce basic concepts relating to ML. To understand these topics, a basic knowledge of algebra and mathematical modeling is needed. A working knowledge of the MATLAB environment is also required.

ML is based on the idea of providing computers with a large amount of input data, together with the corresponding correct answers or labels, and allowing them to learn from this data, identifying patterns, relationships, and regularities within them. Unlike traditional programming approaches, in which computers follow precise instructions to perform specific tasks, ML allows machines to independently learn from data and make decisions based on statistical models and predictions.

One of the key concepts of ML is the ability to generalize. This means that a model trained on information in the training dataset should be able to make accurate predictions about new data that it has never seen before. This allows ML to be applied across a wide range of domains.

How to define ML

To better understand the basic concepts of ML, we can start from the definitions formulated by the pioneers in this field. According to Arthur L. Samuel (1959) – “ML is a field...

Learning is based on the idea that perceptions should not only guide actions but also enhance the agent’s ability to automatically learn from interactions with the world and the decision-making processes themselves. A system is considered capable of learning when it has an executive component for making decisions and a learning component for modifying the executive component to improve decisions. Learning is influenced by the components learned from the system, by the feedback received after the actions are performed, and by the type of representation used.

ML offers several ways of allowing algorithms to learn from data, which are classified into categories based on the type of feedback on which the learning system is based. Choosing which learning category to use for a specific problem must be done in advance to find the best solution. It is useful to evaluate the robustness of the algorithm, such as its ability to make...

In the previous section, we explored the various types of ML paradigms in detail. So, we have understood the basic principles that underlie the different approaches. At this point, it is necessary to understand what the elements that allow us to discriminate between the different approaches are; in other words, in this section, we will understand how to adequately choose the learning approach necessary to obtain our results.

Selecting the ML paradigm

Selecting the appropriate ML algorithm can feel overwhelming given the numerous options available, including both supervised and unsupervised approaches, each employing different learning strategies.

There is no universally superior method, nor one that fits all situations. In large part, the search for the right algorithm involves trial and error; even seasoned data scientists cannot determine whether an algorithm will work without testing it. Nonetheless, the algorithm choice is also influenced by factors...

Up until now, we have acquired knowledge about the functions and capabilities of ML algorithms. We have also gained an understanding of how to identify various types of algorithms, select the appropriate solution for our requirements, and establish an effective workflow. Now, it is time to delve into the process of executing these tasks within the MATLAB environment.

With MATLAB, the process of solving ML problems becomes remarkably straightforward. The comprehensive set of tools and functionalities provided by MATLAB empowers users to leverage various algorithms and techniques effortlessly. Whether you are a beginner or an experienced practitioner, MATLAB equips you with the necessary resources to dive into the world of ML with confidence.

MATLAB is a software platform specifically designed to address scientific problems and facilitate design processes. It offers an integrated environment where calculations, visualizations, and programming seamlessly...

ML, as a modern innovation, has revolutionized numerous industrial and professional processes, enhancing various aspects of our daily lives. Intelligent systems powered by ML algorithms possess the capability to learn from historical data or past experiences. By leveraging this knowledge, ML applications can generate outcomes and insights.

The fields of study in which ML is used cover many types of problems. The main ones are as follows:

• The representation of knowledge and reasoning that aims to reproduce the way of reasoning of the human brain through the definition of symbolism and languages to create machines capable of performing automatic reasoning
• Planning and coordination dealing with the development of systems that, given an application domain, have the objective of predicting future results and making decisions to achieve these objectives and maximize their benefits
• Robotics, for studies related to the movement of mechanical...

In this chapter, we embarked on an exciting journey into the world of ML, exploring a range of popular algorithms to find the best fit for our specific needs. We learned the importance of conducting a preliminary analysis to determine the most suitable algorithm and gained insights into the step-by-step process of building ML models.

Furthermore, we delved into the powerful capabilities of MATLAB for ML, including its support for classification, regression, clustering, and deep learning tasks. We discovered the convenience of using MATLAB apps for automated model training and code generation, streamlining our workflow.

We also introduced the Statistics and Machine Learning Toolbox and the Deep Learning Toolbox, which provided us with additional tools and functionalities to solve our specific problems. We recognized the significance of statistics and algebra in the field of ML and understood how MATLAB could assist us in leveraging these concepts effectively.

Looking...