Artificial IntelligenceMachine Learning
Machine learning is linked to the various branches of data analysis, and in particular to big data, because its implementation in a given context depends in most cases on the existence of large amounts of data obtained in a real situation.
In fact, the algorithms used in machine learning are based on a learning phase generally based on data collected beforehand. During this phase, these data are submitted as input to the algorithm which will adjust its internal parameters to reproduce a desired output.
As soon as the learning phase has been able to take place on enough data, the machine learning algorithm can then replace people by proposing "intelligent" decisions in their place, hence the name artificial intelligence.
It is particularly effective in situations where the core of the algorithm cannot simply be divided into logical steps, because it involves phenomena that are sometimes poorly understood or difficult to determine, as is the case in the following examples:
- Internet shopping behavior
- Expert systems in the medical world
- Financial and stock market analysis
- Recognition of patterns in image, video and sound content
The two main objectives of machine learning are to predict phenomena, and to find clusters in data. Many methods exist for carrying out a machine learning process in a given context. For example, we commonly use:
- Genetic algorithms
- Neural networks
- Decision trees