In Vino Veritas, Artificial Intelligence Approves

Artificial intelligence (AI) – you probably have heard something about it as it recently gains more popularity. Movies tend to portray artificial intelligence as robots in sci-fi and media are trying to attract attention by fancy headlines such as: “Artificial Intelligence: The end of the human race?”, “Machines Are Getting Smarter – Now They Should Explain Themselves” or “Does ‘Super Intelligence’ Pose an Existential Threat to Humans?”.

However, these depictions are slightly misleading. Even though a main goal of AI is to create a thinking machine, we are far from that. Nowadays AI is about learning, reasoning, planning or making predictions. I would like to look beneath the surface, more in-depth. I like to say that when anything is broken down to fundamental pieces, only math remains. So, we can’t leave out mathematics while working with AI, but time for math will come in later posts.


That’s why I chose artificial intelligence as the topic of my semester project. The exact title is “Classification of Wines using Machine Learning”. AI is a broad field of study and machine learning is one of the approaches. We can say machine learning is a tool that gives computers the ability to learn and act without being explicitly programmed. Basically, machine is trying to improve itself.

To understand basic principles of machine learning, it is good to start with classification problems. But why is my semester project about wine? Because as a chemist, we drink a lot should already know something about chemical composition and how it can affect quality of wine.


At the beginning, there is a data set. Scientists have measured properties of different wines, which belong to three quality classes. The chemical analysis determined value of 13 attributes related to wine. Attributes are, for example, amount of alcohol, alkalinity, or color intensity. Some of them are more important than others, some of them do not affect quality at all. The question is, which ones?

And what if there is a new wine? We don’t know a class where a new wine belongs to, but we can measure its properties. Then it is a task for machine learning – according to attributes - to predict the wine class. So, this is also a task for my semester project and a solution will be explained in the next post.