Artificial Intelligence vs Machine Learning
Artificial intelligence is the ability of digital beings to perform tasks commonly associated with intelligent beings, that is, intelligence demonstrated by machines, as opposed to natural intelligence displayed by animals including humans.
Machine Learning is a field of AI that allows machines to learn patterns without being specifically programmed to.
We can classify AI models in two main types:
- Supervised: These models are built with algorithms in which the domain of the output is fed beforehand, so the model learns according to that information. To build a model that classifies images of cats and dogs, previously someone must reveal to the algorithm which image is a cat or a dog.
- Unsupervised: The unsupervised algorithms are expected to group data according to their similar features. For example, clustering of clients of a shop on their most common purchases.
How do we create an AI model?
When building a model, two main parameters must be considered: the dataset and the architecture.
- The dataset is the actual data that we want to learn from. This data can be images, text or tabular.
- The architecture refers to the mathematical operations that will be performed on the dataset in order to learn.
To create a model, the dataset is fed into the architecture and by iterating, the algorithm can distinguish patterns on the data. Once the model created, it can either classify, predict or create new data.
Hidden in the original dataset there can be many biases that would spoil predictions. Learn more about biases in our blog 4 Types of Biases. Data-scientists must find these biases and fix them.
To ease this task, Explainable Artificial Intelligence (XAI) proves very helpful. EXPAI offers a XAI solution that helps to detect these biases and makes your models interpretable. With our services, everyone in the team can analyse a model and find potential biases. If you want to know how, Click here!