Below are the types of Machine Learning.
1) Supervised Learning
Definition: The model is trained in supervised learning using a labeled dataset, meaning each training example is coupled with an output label or result.
Goal: It learns mapping inputs to the right output such that it generalizes well and can accurately predict new, unseen data.
Example: The prediction of house prices according to characteristics such as size, location, and number of bedrooms, where the right prices are already known in the training data.
Algorithms: Linear Regression, Logistic Regression, Support Vector Machines (SVM), k-Nearest Neighbors (k-NN), Decision Trees, Neural Networks.
2) Unsupervised Learning
Definition: The training of the model in unsupervised learning is performed with data that have no labeled outputs. It discovers covert patterns or groupings.
Goal: The model seeks an underlying structure in the data without any prior knowledge of the desired output.
Example: Customer segmentation, where a model groups customers based on their purchasing behavior with no predefined labels for a group.
Algorithms: K-means Clustering, Hierarchical Clustering, Principal Component Analysis (PCA), and Anomaly Detection.
3) Semi-supervised Learning
Definition: It is a hybrid approach when the model is trained on a combination of labeled and unlabeled data. This is done where labeled training data is costly or time-consuming to obtain.
Example: Labeling only a small portion of a vast dataset and using a model to make predictions on the unlabeled data, as with medical image classification.
Algorithms: Semi-supervised SVMs, Self-training algorithms.
4) Reinforcement Learning (RL)
Definition: Reinforcement learning is a type of machine learning in which an agent learns through interaction with the environment and gets feedback either in the form of rewards or penalties.
Goal: The ultimate goal of the agent is to maximize cumulative rewards by trying a variety of actions and learning from them.
Example: Teaching an AI to play a game, wherein the AI learns from its actions—namely, winning or losing—and in time changes its strategy.
Algorithms: Q-learning, Deep Q Networks (DQN), Policy Gradient methods.
1) Supervised Learning
Definition: The model is trained in supervised learning using a labeled dataset, meaning each training example is coupled with an output label or result.
Goal: It learns mapping inputs to the right output such that it generalizes well and can accurately predict new, unseen data.
Example: The prediction of house prices according to characteristics such as size, location, and number of bedrooms, where the right prices are already known in the training data.
Algorithms: Linear Regression, Logistic Regression, Support Vector Machines (SVM), k-Nearest Neighbors (k-NN), Decision Trees, Neural Networks.
2) Unsupervised Learning
Definition: The training of the model in unsupervised learning is performed with data that have no labeled outputs. It discovers covert patterns or groupings.
Goal: The model seeks an underlying structure in the data without any prior knowledge of the desired output.
Example: Customer segmentation, where a model groups customers based on their purchasing behavior with no predefined labels for a group.
Algorithms: K-means Clustering, Hierarchical Clustering, Principal Component Analysis (PCA), and Anomaly Detection.
3) Semi-supervised Learning
Definition: It is a hybrid approach when the model is trained on a combination of labeled and unlabeled data. This is done where labeled training data is costly or time-consuming to obtain.
Example: Labeling only a small portion of a vast dataset and using a model to make predictions on the unlabeled data, as with medical image classification.
Algorithms: Semi-supervised SVMs, Self-training algorithms.
4) Reinforcement Learning (RL)
Definition: Reinforcement learning is a type of machine learning in which an agent learns through interaction with the environment and gets feedback either in the form of rewards or penalties.
Goal: The ultimate goal of the agent is to maximize cumulative rewards by trying a variety of actions and learning from them.
Example: Teaching an AI to play a game, wherein the AI learns from its actions—namely, winning or losing—and in time changes its strategy.
Algorithms: Q-learning, Deep Q Networks (DQN), Policy Gradient methods.