Artificial intelligence (AI) are systems that mimic human intelligence to perform tasks and improve those actions based on the information gathered. More information on what AI is can be found here.
Artificial intelligence and machine learning are not the same thing, but they are closely connected.
Machine learning is a subset of artificial intelligence:
Machine learning
It is understood that machine learning is a branch of artificial intelligence and computer science, focusing on algorithms that mimic the way humans learn, improving their accuracy through experience.
With this technology, the system can improve itself based on the data provided to it.
The way machine learning works depends on the input data and the algorithms used.
Machine learning methods
There are 4 basic machine learning techniques:
- supervised machine learning – uses data with labels for learning, the machine learns from examples:
Example of this learning technique: when we want the machine to recognize a picture of a dog – we show the machine pictures of different dogs, which we label as “dog”. The machine by looking at more and more pictures of the dogs can extract certain characteristics of the dog.
After the training is complete, we can show the machine the pictures of different animals – including dogs – without labels.
The computer then compares the images and recognizes the dog. It is crucial, therefore, that the machine learns at the beginning to set a certain pattern that can be used in similar cases.
2. unsupervised machine learning – the machine itself analyzes and groups data sets without labels, detects patterns without human intervention.
Example: if we showed the machine pictures of different animals, it would search for characteristics and create sets of specific animal species.
3. semi – supervised learning – uses a smaller data set with labels to group a larger data set without labels.
4. reinforcement learning – involves the algorithm not being trained using sample data, but learning by trial and error. The machine is given only a set of rules and assertions on the basis of which it acts – it uses them to achieve certain effects. These effects can be positive or negative. It is rewarded for positive effects, and punished for negative ones.
Machine learning vs deep learning
Deep learning is a subset of machine learning. The difference between the two lies in the way each of these algorithms learns. Deep learning does not require human control – it can occur in an unsupervised manner, thus using larger data sets. Machine learning, on the other hand, was mentioned is human-dependent and requires data that is already structured.
Artificial neural networks are used for deep learning.
The learning process is called deep because the structure of the aforementioned networks consists of a large number of layers – input (which take input parameters), hidden layers (whose function is to learn) and output (provide the results).
Neural networks
To teach a computer human reasoning, it is common to use neural networks, which are algorithms that are modeled on the human brain – mimicking the way biological neurons communicate with each other.
They learn from vast amounts of data.
Neural networks can be used to create images, make predictions, or recognize faces and speech.
Where is machine learning being used?
Machine learning is used:
- in predictive analytics – to predict trends;
- in speech recognition – to identify words;
- in facial recognition
- in manufacturing – to increase operational efficiency.
Author: Agata Konieczna
Credits:
Dog, cat, bear icons, circle icons created by Freepik – Flaticon
Supervisor icons created by lutfix – Flaticon
Machine learning icons created by Icongeek26 – Flaticon
Automation icons created by Becris – Flaticon
Research icons created by ultimatearm – Flaticon
Access icons, exit icons created by Pixel perfect – Flaticon
