Depending on the amount and type of supervision received during training, machine learning can be classified into the following four types: supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning.
supervised learning
In supervised learning, the training data that is provided to the algorithm, containing the desired solution, becomes the labels, or markers.
Simply put, supervised learning includes independent variables and dependent variables (with Y), and can be used for both classification and regression.
Common algorithms:
- K nearest neighbor algorithm
- linear regression
- logistic regression
- Support Vector Machines (SVMs)
- Decision Trees and Random Forests
- Neural Networks
unsupervised learning
The training data for unsupervised learning is unlabeled, and the algorithm learns automatically without guidance.
Simply put, the training data has only independent variables and no dependent variables (that is, there is no Y).
Common algorithms:
- Clustering Algorithm
- K-means algorithm (K-means)
- Density-based clustering method (DBSCAN)
- Maximum Expectation Algorithm
- Visualization and Dimensionality Reduction
- principal component analysis
- Kernel Principal Component Analysis
- association rule learning
- A priori
- Glow
semi-supervised learning
Some algorithms can handle partially labeled training data, usually a lot of unlabeled data and a small amount of labeled data, which is called semi-supervised learning.
Photo recognition is a good example. Online photo albums can be specified to identify photos of the same person (unsupervised learning). When you add a tag to these same people, new photos with the same person will automatically be tagged for you.
Most semi-supervised learning algorithms are a combination of unsupervised and supervised algorithms. For example, Deep Belief Networks (DBNs) are based on unsupervised components stacked on top of each other.
reinforcement learning
Reinforcement learning is a very different kind of algorithm, a learning system that observes the environment, makes choices, performs actions and is rewarded, or punished in the form of negative rewards. It has to learn on its own what is the best strategy to get the greatest reward over time.
For example, many robots learn how to walk through reinforcement learning algorithms. The AlphaGo project is also a good example of reinforcement learning.