Today's students extensively use Artificial Intelligence to ask specific questions, facilitate the completion of projects, and even solve problems. However, most High Education Degrees do not include advanced AI tools in their syllabus. The concept of Machine Learning involves the use of algorithms to identify patterns in (usually massive) data and make predictions. These algorithms require prior learning with the available data to make predictions automatically. The advantage of these algorithms is that they are much cheaper, faster, and more direct than simulators.

The simulation of processes and systems has been, for many decades, the (on many occasions the only) option to predict a response to a given problem (input data). However, simulation is costly in terms of time and money. For example, in the chemical engineering field, chemical industrial process simulators have become widespread in both industry and academia. The licenses to use these simulators are expensive, as they require a lot of previous theoretical and practical work in addition to their implementation in software. On the contrary, neural networks do not require prior theoretical knowledge of the actual process. They do not use complex equations (not even simple ones), do not need to feed on characteristic parameters of the physical process, and do not know about process control systems. Neural networks only need data to learn, and once trained, the neural network needs validation using more available data. After validation, the neural network can respond appropriately to specific input data (other than training and validation data for which the response is unknown). The advantage of using neural networks for Machine Learning in process industry positions is increasingly widespread. When faced with a vacant job, the company will preferably hire a candidate with knowledge and practice in Machine Learning. Furthermore, the salaries of these workers are higher than those who do not have this skill.

Given these facts, it is essential to schedule teaching activities related to neural networks for undergraduate students to make them aware of such emerging technology and prepare them for their future endeavours. In a previous work [1], the Matlab Neural Network Fitting function was studied with some data (reaction rates) to predict values from several input values. The training of the neural network was unsupervised. In this work, an exercise has been designed using this tool and applied to students in the sixth semester of an engineering degree. A survey was also prepared to determine the students’ opinions and acceptance. The positive results have encouraged the lecturers to continue and deepen the activity in the coming years. Although the use of neural networks can never replace the knowledge and concepts of theoretical-practical operation of the processes studied in the Degrees’ subjects, it can constitute a quick and helpful tool to extract information and responses from a system, saving money and effort for a company. Therefore, students' Machine Learning knowledge is highly valued in the job market and constitutes an essential content in Higher Education.

References:
[1] E. Romero, J. Remón, Neuronal Networks as emerging tools for data fitting in Higher Education, 19th International Technology, Education and Development Conference (INTED2025) Proceedings, pp. 7316-7322, 2025. DOI: 10.21125/inted.2025.1894.

Keywords: Artificial Intelligence, neural networks, Machine Learning, Matlab, training data.