What Is Machine Learning And Why It Matters?
Machine learning is a subject within the field of artificial intelligence. It is defined as the ability of machines to learn automatically by themselves.
When we say ‘learn’, we mean a computer that has access to millions of data. From them, its algorithm is capable of establishing behavior patterns. And even it can make predictions of trends that will occur in the mere future. But the most interesting thing is that the computer machine, from these data analyzes, can learn by itself, without a human being intervening in its programming. The more data it examines, the more ‘smart’ it becomes.
A little history about Machine Learning:
Arthur Samuel is considered the father of machine learning. In 1952, he developed a checkers program that was capable of improving game by game.
Soon after, in 1958, Frank Rosenblatt creates Perceptron. It is an artificial neural network that wants to function like the human brain.In the year 1967, when the Nearest Neighbor algorithm is developed. It was the first program that used pattern recognition. Gerald Dejong achieved another milestone in 1981. It is the birth of Explanation Based Learning.
In Machine learning the computer has ability of analyzing data and creating a rule that discards the useless things.
In the 90s, the data driven concept was developed. We are talking about programs that draw conclusions from the information they analyze. In 1997, one of the best-known events for public opinion occurs: IBM’s Deep Blue computer wins the world chess championship, Gary Kasparov.
Currently, the race for the advancement of machine learning has increased up speed: IBM launches its artificial intelligence, Watson, in 2010; In 2014, Facebook developed DeepFace, capable of recognizing people. In 2015, Amazon introduced its own machine learning platform, a tool that wants to be accessible to all developers.
The key is in the algorithm without the hand of a man. Algorithms can turn machines into “computer engineers” who program by themselves. Each new data helps to make the computer’s operation more complex, effective and more precise.
Different classification techniques are used. But one of the most useful is the decision tree. Constructed from diagrams, it develops like a map. It offers all the probabilities that would result from making every possible decision. Those actions are compared to each other based on costs, eventualities and benefits.
In machine learning, there are several types of algorithms:
Supervised Machine learning:
Detailed and precise information and data are provided to the machine. It is the basic knowledge for your analyzes. The different examples serve to make further generalizations.
Unsupervised Machine learning:
It is more like the way our brain works. The computer does not receive prior information about the data. You have to crawl your own database and establish patterns through understanding and abstraction.
Machine learning benefits:
Machine learning has many applications in the business world. According to the Cloudera platform, 87% of larger companies have invested in machine learning.
And a third of them say they have already experienced an (ROI) return on investment.
Why do you go for machine learning?
- It helps to better understand consumers, their tastes and needs through the analysis of their behavior.
- It favors innovation and the development of new products. Their in-depth analysis helps uncover patterns and solutions that have never been thought of before.
- It is based on the adaptability of the processes. Optimize the logistics of the company thanks to its solid database.
- Quick and effective decision making. It allows the company to be up to date by offering trusted information.
- Machine learning changes humans
- Machine learning systems (just one example of AI that affects people directly) recommend new movies based on your score to others and after comparing your preferences with other users. Some systems are already quite good at it.
- A movie recommendation system changes your preferences over time and limits them. Without it, you would face the nightmare of having to watch bad movies and movies you don’t want. But with AI, you will hit the nail on the head with each movie and, in the end, you will stop investigating and only consume what it offers you.
- It should be noted that we do not usually realize how these algorithms manipulate us. This example from the movies is not scary, but think about the news or advertising.
Strong and weak artificial intelligence
First, we have to differentiate two concepts: strong and weak AI. Strong AI is a hypothetical machine that is capable of thinking and aware of its own existence, furthermore, it can not only solve the tasks for which it is programmed, but it can also learn new things.
Weak AI already exists, it is found in all applications created to solve specific problems, such as recognizing images or driving a car. Weak AI is what we commonly know as “machine learning”.
It is still unknown when strong AI can be invented. According to the studies of the experts, we will have to wait another 45 years, that is, “someday”. They also claim that fusion power will be on the market in 40 years, which is exactly the same as they said 50 years ago.