Since the appearance of artificial intelligence (AI) at the end of the first half of the 20th century, the technology has evolved greatly. During its development it has lived golden moments, characterized by the conviction that this technology had great potential and the arrival of large amounts of money to carry out research, and moments of crisis, called "AI winters", in which the lack of computing power did not allow technology to develop as expected and research funds disappeared.
During one of those golden moments, in the 1980s, a branch of AI was born: machine learning (ML). ML uses mathematical algorithms that allow machines to learn. Machine learning is an analytical way of solving problems through identification, classification or prediction. Algorithms learn from entered data and then use this knowledge to draw conclusions from new data.
Already in the 21st century, in 2011, a branch of machine learning called deep learning (DL) appeared. The popularity of machine learning and the development of the computing capacity of computers enabled this new technology. Deep learning as a concept is very similar to machine learning but uses different algorithms. While machine learning works with regression algorithms or decision trees, deep learning uses neural networks that function very similarly to the biological neural connections of our brain.
Supervised and unsupervised learning
The learning carried out by the algorithms can be supervised or unsupervised depending on whether they require human assistance or not.
An unsupervised learning system is based on algorithms that learn from data with unlabeled elements looking for patterns or relationships between them: input data are provided (not output data). Unsupervised learning does not require human intervention.
Unsupervised learning algorithms can be clustering algorithms, i.e., discovery of groups in the data, such as, for example, grouping customers according to their purchases, and association algorithms, i.e., the discovery of rules within the data set. For example, they can discover that customers who buy a car also take out insurance.
In contrast, a supervised system is based on algorithms that learn from data with tagged elements. Expected input and output data are provided. This type of learning requires human intervention.
In this case, the algorithms can be classification algorithms, to classify objects within classes, such as sick patients or spam mail, or regression algorithms, to predict a numerical value such as the price of a house, occupancy demand, weight or height.
AI has developed other branches since its appearance, such as natural language processing (NLP), robotics or speech and image recognition and machine vision, among others.
What is machine learning?
ML algorithms are mathematical algorithms that allow machines to learn by imitating the way humans learn, although machine learning is not only algorithms, it is also the approach from which the problem is approached. Machine learning is basically a way to get artificial intelligence.
What is deep learning?
Deep learning (DL) is part of machine learning. In fact, it can be described as the new evolution of machine learning. It is an automatic algorithm that mimics human perception inspired by our brain and the connection between neurons. DL is the technique that comes closest to the way humans learn.
Most deep learning methods use neural network architecture. That is why deep learning is often referred to as "deep neural networks. It is known as "deep" in reference to the layers that these neural networks have.
What's the difference between the two?
Simply explained, both machine learning and deep learning mimic the way the human brain learns. Its main difference is therefore the type of algorithms used in each case, although deep learning is more similar to human learning as it works with neurons. Machine learning usually uses decision trees and deep learning neural networks, which are more evolved. In addition, both can learn in a supervised or unsupervised way.