There are three main types of machine learning: supervised learning, unsupervised learning, and reinforcement learning. Supervised learning is used when you have labelled data and want to predict a target variable, unsupervised learning is used when you don't have labelled data and want to identify patterns, and reinforcement learning is used when you want to train a system to make decisions based on rewards or penalties.
Supervised Learning: Use supervised learning when you have labelled data, which means you have input features and corresponding output labels. This type of machine learning is used for tasks like classification (where you want to predict a categorical label) and regression (where you want to predict a continuous output).
Unsupervised Learning: Use unsupervised learning when you don't have labelled data and want to find patterns in the data on your own. This type of machine learning is used for tasks like clustering (where you want to group similar data points together) and dimensionality reduction (where you want to reduce the number of features in the data).
Reinforcement Learning: Use reinforcement learning when you want an agent to learn how to take actions in an environment to maximize a reward signal. This type of machine learning is used for tasks like game playing, robotics, and self-driving cars.
Semi-supervised Learning: Use semi-supervised learning when you have some labelled data and a lot of unlabelled data. This type of machine learning combines aspects of both supervised and unsupervised learning to learn from both labelled and unlabelled data.
Transfer Learning: Use transfer learning when you have a pre-trained model on a similar task and want to use it to help with a new task. This type of machine learning is used to transfer the knowledge learned from one task to another and can save time and resources.
Machine learning tasks have been divided into three categories, depending upon the feedback available:
1. Supervised Learning: These are human builds models based on input and output.
2. Unsupervised Learning: These are models that depend on human input. No labels are given to the learning algorithm, the model must figure out the structure by itself.
3. Reinforcement learning: These are the models that are feed with human inputs. No labels are given to the learning algorithm. The algorithm learns by the rewards and penalties given.
The algorithms that can be used for each of the categories are:
Algoritm | Supervised | Unsupervised | Reinforcement |
Linear | 1 | 0 | 0 |
Logistic | 1 | 0 | 0 |
K-Means | 1 | 1 | 0 |
Anomaly Detection | 1 | 1 | 0 |
Neural Net | 1 | 1 | 1 |
KNN | 1 | 0 | 0 |
Decision Tee | 1 | 0 | 0 |
Random Forest | 1 | 0 | 0 |
SVM | 1 | 0 | 0 |
Naive Bayes | 1 | 0 | 0 |
The machine learning functions and uses for various tasks are given in the below table:
â€‹ | Algorithm | Use |
Basic Regression | Linear | Lots of numerical data. |
â€‹ | Logistic | Target variable is categorical. |
Cluster Analysis | K-Means | Similar datum into groups based on centroids. |
â€‹ | Anomaly Detection | Finding outliers through grouping. |
Classification | Neural Net | Complex relationships. Prone to over fitting. |
â€‹ | K-NN | Group membership based on proximity. |
â€‹ | Decision Tree | If/ then/ else. Non-contiguous data. Can also be regression. |
â€‹ | Random Forest | Find best split randomly. Can also be regression. |
â€‹ | SVM | Maximum margin classifier. Fundamental. Data Science algorithm. |
â€‹ | Naive Bayes | Updating knowldege step by step with new info. |
Feature Reduction | T-DISTRIB Stochastic NEIB Embedding | Visual high dimensional data. Convert similarity to joint probabilities. |
â€‹ | Principle Component Analysis | Distil feature space into components that describe the greatest variance. |
â€‹ | Canonical Correlation Analysis | Making sense of cross-correlation matrices. |
â€‹ | Linear Discriminant Analysis | Linear combination of features that separates classes. |
The flowchart given below will help you give a rough guide of each estimator that will help to know more about the task and the ways to solve it using various ML techniques.
Find out more about machine learning with cryptocurrencies and see how Neomony can help you invest in the crypto currency markets !
Comments