This article is an adaptation, the original can be found here
Much is said about artificial intelligence (AI), machine learning (machine learning) and deep learning, all are complementary terms that are part of a whole and increasingly help market research in obtaining insights. By presenting complementary functions, their definitions are often not known in a different way.
Artificial Intelligence refers to a machine’s ability to mimic human intelligence behavior, such as making chats, playing chess, or even performing a medical diagnosis. One concept to note is Alan Turing’s famous Turing test – the “imitation game”.
A machine and a human being are prepared for an interrogation, if the interrogator (C) can’t distinguish the machine and the human during the interaction process, the machine behaves intelligently and is therefore approved in the Turing test.
Machine Learning is a subset of Artificial Intelligence
AI demonstrates intelligence by producing appropriate responses to the environment, as humans do. For example, the ability of a self-directed automobile to detect a pedestrian crossing the highway and slowing down instead of advancing on it.
To make such intelligent responses, the machine must learn to sense the environment and make the judgment appropriate. And the learning process is machine learning.
Why is the ability to learn without the need for explicit programming important?
Before machine learning, computers only followed instructions written in a code, everything that was mentioned and was part of the code, allowed the computer to react. But if there was something to be discussed that did not have in the code, the computer lost the ability to discussion not knowing what was to be done, since there were no instructions that could be followed.
Let’s go back to the example of the auto-steer, before we had the machine learning, we needed to write the instructions for the computer to distinguish the pedestrians. How could we describe pedestrians? How thin / fat could he / she be? Every pedestrian the car found would be different, making a complete pedestrian feature listing impossible.
Machine learning offers a different approach, we can build a model, providing some situations where the pedestrian walks and we decelerate. And talk to the computer “Look, here’s the data. You need to learn the concept of pedestrian.” And the computer will learn by itself.
This is of crucial importance because the world is so complex that it becomes impossible to list all possibilities for machines. Machine learning solves this problem by giving the machines the ability to grasp.
Deep Learning is the best current method for Machine Learning
Our most advanced cognitive machine is the human brain, and the brain is essentially a network of neurons that fire together. Deep learning mimics the way our neural network works.
Deep learning is called this way because it acts on many simultaneous layers, it is the deep structured learning of complex and / or large-scale data sets. In summary Deep Learning is a sub-category of machine learning that relates to learning opportunities with the use of neural networks, such as speech recognition, computer vision and natural language processing.
Deep learning is responsible for the production of a “training” of a computational model capable of deciphering the natural language. The model relates terms and words to infer meaning since it is fed with large amounts of data.
As previously seen machines are already “taught” to read the documents and can answer questions about their content, but their knowledge bases are usually limited by the size of the documents. As the amount of algorithms online does not stop growing the Deep Learning approach comes to make systems can make use of a greater number of natural language, giving you a deeper understanding of universal themes.
Artificial intelligence is already present and continues to be a trend for the near future. Its impact is noticeable in many sectors and industries and in market research the technology each comes from is assisting in increasing human capabilities for data analysis and innovation.