Supervised vs Unsupervised Learning – We have all been dazzled at some point by robot movies: intelligent machines that are capable of speaking and responding to us but can also solve problems and learn on their own. Although machine learning is here to stay, this means another step in the world of Artificial Intelligence. We tell you what Machine Learning is and two operating models, Unsupervised Learning.
What is Machine Learning or machine learning?
When we talk about the meaning of Machine Learning, we refer to an area of knowledge within Artificial Intelligence where computers apply statistical learning techniques to identify patterns in data automatically. For this reason, we can define automatic learning as automated learning or machine learning, an exciting world within Big Data and Business Intelligence Unsupervised Learning.
The generated algorithms can generalize behaviours from the data supplied as examples.
Difference Between Supervised and Unsupervised Learning
As we said, the main difference between these two types of algorithms lies in the data we use in their training.
In supervised learning, the results obtained from the model known in advance. For example, in an algorithm that predicts the validity of an email, we know that the output result must spam or not spam, and its training process carried out with hundreds of thousands of example emails labelled as spam and not spam.
However, in unsupervised learning, training not carried out with previously labelled data; instead, said labelling of the data is discovered during the learning process itself.
Types of supervised learning
Supervised learning aims to predict future responses by training the algorithm with known data from the past. The problems that led learning solves, in general, are of two types:
Regression problems. In this type of problem, what s sought is to infer (predict) a continuous numerical response based on a set of input variables. There are numerous real-world problems where the variable expected or estimated is a numerical variable, for example, a person’s income, the sale price of a property, or something as diverse as the load that a metal structure will support.
Types of unsupervised learning Supervised vs Unsupervised Learning
In the case of unsupervised learning, the problem consists of testing and determining the existing structure in the data, but without using a previous label.
These algorithms also known as grouping or clustering algorithms since they group instances of the data based on the variables of the data sets. That is, this type of algorithm looks for patterns in the data to find clusters.
Unsupervised learning algorithms can divided into two large groups according to their internal operation:
Other types of algorithms
Specific algorithms such as anomaly detection or reinforcement learning algorithms would not fit so clearly into supervised or unsupervised.
Reinforcement learning focuses on rule-based learning processes capable of learning from their environment, in which machine learning algorithms are provided with information from the background about what is or is not appropriate.
Machine Learning Example: Siri
For Machine Learning to improve our quality of life, we will have to accept a change in our relationship with technology. We have a close example: Siri is a passive servant that waits for us to ask to give us the information we need. We like this question-answer paradigm; we would not understand that Siri gave us advice left and right every time we faced a dilemma, even if we had not asked any questions.
Supervised and unsupervised learning
The number of algorithms used in data mining enormous, and we run the risk of getting lost in a sea of acronyms that initially only add confusion. An excellent way to approach it will be to organize them according to the learning they use: supervised and unsupervised.
Supervised learning assumes that we start from a previously labelled data set; that is, we know the value of the objective attribute for our data set. Unsupervised learning starts from previously unlabeled data.
What is classification in supervised learning?
There are two main kinds of supervised learning; classification and regression. Variety where an algorithm is trained to classify input data into discrete variables. The algorithms receive training input data with a ‘classification’ tag during training. For example, the training data might consist of the latest credit card bills, including e-card payments, for a set of customers, labelled whether or not they made a future purchase. When a new customer’s card balance presented to the algorithm, the algorithm will classify the customer into the “will buy” or “won’t buy” group.
How does unsupervised learning work?
To understand unsupervised learning, we will first need to understand supervised learning. If a computer were learning to identify fruits in a supervised learning environment, it would given sample images of labelled fruits, called input data. For example, labels would say that bananas are long, curved, and yellow, apples are round and red, while orange is spherical, waxy-looking, and orange. After a suitable amount of time, the machine should be able to identify which fruit is which based on those descriptors confidently. If you were presented with an apple, for example, you could safely say that it not orange, therefore not an orange, but also that it not yellow and long, hence not a banana. So,
Importance of supervised learning
Supervised learning offers solutions to process and convert data into accurate information. It helps companies predict situations of interest, letting them know which decisions favour their objectives and, conversely, which are the decisions that harm them.
This virtue offers companies a huge advantage over the rest of their competitors since they can anticipate unfavourable events and, therefore, easily avoidable.
Knowing this, organizations focus on those situations that do provide considerable benefits.