Machine learning is a topic that has gained more and more attention in recent times. If you still do not know what it is, we can tell you that this is a kind of artificial intelligence that allows software applications to be quite accurate in predicting results, even without being explicitly programmed for it.
The learning that gives a name to the expression “machine learning” consists in the execution of algorithms that automatically create models of knowledge representation based on a set of data. The idea behind this learning is that we must train the machines, giving them access to historical data, to one or more measures of performance, and let the algorithm “learn”, that is, to adjust the model of knowledge representation so that it improves its performance. After this training, the model has the potential to make quality forecasts in future situations that are related to historical standards. This strategy can even be used to access the Internet and learn continuously with all the data that is sought.
In fact, this is a method of data analysis that is so established in our day-to-day life that we are hardly aware of its use, given all our familiarity with it. For example, Amazon recommendations, Web searches, and automatic translations of the Google service are based on machine learning algorithms. With machine learning computers make our lives easier, acting quickly and subtly, although they often require a large amount of data and processing in their training phase.
Application of machine learning
If you are wondering where we often come across the use of machine learning, here are some examples:
- Online recommendations for products on sales websites
- Real-time ads on websites
- Custom results displayed on Facebook Feed
- Netflix series and movies recommendations
- Optimization of online research and its results
- Spam filtering in email
- Fraud and intrusion detection
- Speech recognition and semantics (natural language processing)
- Object recognition
- Text recognition
Machine learning and artificial intelligence
This field of study began with the study of pattern recognition and evolved into something more elaborate like artificial intelligence, reaching other levels of development, such as autonomous driving, with cars driving alone on the roads without a driver.
Thus, in addition to simplifying tasks for man, machine learning may raise an additional essential question: can the development of this science lead to artificial intelligence reaching levels similar to those of humans? Doing physical and mental tasks faster and better (and cheaper) than humans?
By developing very precise forecasts, it is possible to make better decisions and even act without human intervention.
According to The Baltimore Sun, in 2016 the machines were able to diagnose lung cancer with 50% more accuracy than radiology specialists.
If the machine is developed at a high level, one arrives at an old hypothesis fueled by several famous works of science fiction: the replacement of man by the machine.
This replacement began in the industry, for example in automobile construction, and has achieved other functions such as telephone marketing. Soon we will have autonomous vans and mailmans (estimated to be in the year 2025 in the US), machine lawyers who know by “heart” all the complex legislation and nurse robots who administer medication in hospitals to patients.
What awaits us in the future?
It is in services that the machines promise to take jobs from humans in the near future. A new report released by McKinsey & Company indicates that, by 2030, about 800 million workers around the world could be replaced in robot work.
While some are still reluctant to rely entirely on machines (Elon Musk), others are quite optimistic about them (Mark Zuckerberg). However, nobody wants to lose the autonomization train, investing in this sector. What awaits us in the future is how the lyrics of a song tell us: “only time will tell”.
Machine learning in CCG
In CCG – Center for Computer Graphics – machine learning is already a reality of the present. It is being developed in the domain of applied research EPMQ, by researcher Paulo Cortez, with Ana Lima as development coordinator. In the CVIG domain, the machine learning is also worked, especially in terms of image processing.
It may interest you the scientific publication: “Water quality of Danube Delta systems: ecological status and prediction using machine-learning algorithms“.