When technical terms become buzzwords, it's not just IT experts who roll their eyes. Here is one such case. AI, machine learning, and deep learning - the three terms overlap and are also easy to confuse.
In this , we will explain what is the difference between the 3 terms in a simple and uncomplicated way: Artificial Intelligence, Machine Learning, and Deep Learning and how these technologies work.
Artificial intelligence is an umbrella term and describes the rough approach of using machines to imitate intelligent human behavior to solve problems.
This approach is by no means new: the term "Artificial Intelligence" was first introduced in 1956 at the Dartmouth Conference in a workshop. However, AI only became relevant in the last few years, especially since 2015. This is due, on the one hand, to the computing power, which has significantly improved in recent years, and, on the other hand, to the availability of data (e.g. images and videos on the Internet), which has also increased significantly. We'll clarify in a moment exactly why these two points are so important.
A simple, general example of the use of AI is the spam filter in your e-mail inbox: a person would manually sort out certain e-mails based on characteristics (sender address, certain words in the subject line, etc.). This process can also be "artificially" mapped: by writing a program that takes on exactly this task and sorting out the e-mails for us based on the characteristics. Intelligent, human behavior was imitated here - so it is AI.
However, this AI would be very limited in its capabilities. What happens if the senders of the spam e-mails change their sender address or the words in the subject line? Then the spam filter would no longer work.
So, you need a technology that learns by itself and adapts accordingly. This technology is already in use - machine learning.
Machine learning is a technology that is used to achieve artificial intelligence.
First, data (e.g. images, videos, audio files, statistics, etc.) are collected and «fed» to the program – this is called the data input. This data is then passed on to the program, which uses a more complex algorithm to analyze the data and make predictions or make decisions. The specific feature is that machine learning programs learn without human intervention.
To come back to our example with the spam filter: First, you would show the system a lot of different non-spam emails. Then you would show it lots of different spam emails. The system analyzes these e-mails, finds differences and similarities (data mining), and creates its own rules based on which it classifies an e-mail as spam or non-spam. A machine learning system learns and becomes more intelligent with more data. The general science that deals with data, statistics, algorithms, and machine learning is called data science.
Further everyday examples of the use of machine learning are:
- The face recognition function of your smartphone
- The delivery of the results from your Google search
- Weather forecast
- The series and product suggestions on Netflix, Youtube, and Amazon
In general, the following rule applies to machine learning: the more data the algorithm receives, the more precise the result is.
In recent years the significant volumes of data have become available, while the computing power has been optimized through the further development of so-called GPUs (Graphics Processing Units) and the algorithms have been refined further. This further development of machine learning is known as deep learning.
Deep learning is the further development of machine learning. The technology makes use of so-called neural networks (also known as artificial neural networks).
These Artificial Neural Networks (ANN) are originally inspired by how the human brain works:
When the brain receives information, it tries to decipher it by categorizing the information according to characteristics. If it receives new information, this is compared with the information already available to interpret this information. This is a very rough description. In reality, these are highly complex processes that take place within fractions of a second. Although attempts were initially made to imitate this artificially, ANNs are now making use of complicated algorithms that have little in common with the human brain. These deep learning algorithms are far more complex than those used in “traditional” machine learning.
To achieve accurate results, deep learning requires enormous amounts of data and therefore extreme computing power, which we are only just achieving with today's technology. So, the future will be exciting.
With this, we would have worked out the essential background of the first definitions at the beginning of the article. AI refers to devices that have human-like intelligence in some way. There are many techniques for AI, but a subset of that larger list is machine learning - in other words, making algorithms learn from the data. Finally, deep learning is one of the various types of machine learning that uses complex neural networks to solve the most difficult problems for computers.