Machine Learning: Definition, Types, Algorithms, Applications

Photo by Fabian Grohs on Unsplash

definition:

Machine Learning is the science of programming computers so they can learn from data.

We provide training data to the system then we pass this data through an algorithm which creates a model now the new input(test data) is passed through the model which in turn gives predicted output so basically the model is trained using this training data.

The model is a mathematical representation of a real-world process. Suppose we take an example of the general equation of a straight line y=mx+c, so here y is the output of the machine learning model and x is the input variable which the user will be providing m and c are the parameters which will be obtained by training the model.

Types of Machine Learning:

There are mainly three types

1. Supervised Learning
2. Unsupervised Learning
3. Reinforcement Learning

Supervised Learning

Here the algorithm takes both input and output of those data and trains the model. The model makes predictions based on the given data and the output is corrected by training data. This process continues until the algorithm achieves an acceptable level of performance.

The supervised learning is of two categories:

Classification:

When the output variable is categorical or discrete it is considered a classification problem

Some cases where classification is used: To find whether an email received is spam or not, handwriting recognition.

Regression:

When the output variable is continuous it is considered a regression problem.

Regression is used in predicting housing prices, predict rainfall using different atmospheric parameters like temperature.

Some important supervised learning algorithms are:

1. K-nearest neighbours
2. Linear Regression
3. Logistic Regression
4. SVM( Support Vector Machine )
5. Decision Tree and Random Forests
6. Neural Networks

Unsupervised Learning

In Unsupervised learning, there is only input data and no corresponding output data and there is no “supervisor” to teach the algorithm, unlike the supervised learning. Here the algorithm is left on its own to find a pattern in the data and then it gives the output. In Unsupervised learning, we work with “unlabelled” data which means no extra effort is needed to convert that unlabelled data into machine-readable form hence we can work with the larger dataset.

Unsupervised learning is of two categories:

Clustering:

In this process, data is grouped based on the similarity of the data-points. Clustering is used for identifying cancer cells, it is used in google news.

Dimensionality Reduction:

Dimensionality reduction refers to techniques that reduce the number of input variables in a dataset. It helps in visualizing data

Association Rule Learning:

The goal of Association Rule Learning is to dig into large amounts of data and discover interesting relations between data. It is used in data analysis.

Some important algorithms of unsupervised learning are:

1. Hierarchical Cluster Analysis (HCA)
2. k-Means
3. Principal Component Analysis (PCA)
4. Apriori

supervised learning and unsupervised learning

Reinforcement Learning:

The learning system called an agent which observe the environment, select and perform actions, and get rewards or penalties in return. It must then learn by itself what is the best strategy, called a policy, to get the most reward over time. A policy defines what action the agent should choose when it is in a given situation. Here the learning takes place using the hit and trial method.

Reinforcement Learning is used in Robotics, games.

Summary: