Share this post

## Machine Learning

Machine Learning is a one-directional learning algorithm, that can learn from experience and improve its performance without being explicitly programmed.

Machine Learning algorithms classified into 4 types. They are

- Regression Algorithms
- Classification Algorithms
- Clustering Algorithms
- Regularization Algorithms

### 1. Regression Algorithms

The Regression Algorithms find out the relation between input variables in mathematical geometric equations. A line is fit to the data in such a way that maximum data points are on the line. When we give new input to data, it predicts output by replacing variables as input values in the equation.

The following figure represents a regression model. All data points are distributed along with coordinates of the X-axis and Y-axis in a random manner. A line is drawn on points to fit and train the algorithm on the data.

Image Courtesy of Unsplash

Regression Algorithms again classified into 6 types. They are

- Simple Linear Regression
- Multiple Linear Regression
- Polynomial Regression
- Support Vector Regression(SVR)
- Decision Tree Regression
- Random Forest Regression

### 2. Classification Algorithms

Classification algorithms make a threshold point and divide all data points into two or more classes. For example, if a combination of men and women images are given to the model, then the model divides persons into separate classes.

In the above classification algorithms, the threshold value is 0.5. The Algorithm divided based predicted value ranges.

If predicted value >0.5 then that will be labeled as one class and for the values, less than 0.5 are labeled as another class.

There are a total of 7 classification algorithms in Classification algorithms. They are

- Logistic Regression
- K-Nearest Neighbours Algorithm(KNN)
- Support Vector Machines(SVM)
- Kernel SVM
- Naive Bayes Classification
- Decision Tree Classification
- Random Forest Classification

### 3. Clustering Algorithms

Clustering is described as grouping all data in the set of groups. This clustering can be done using 4 types of distances in Machine Learning. They are

- Euclidian Distance
- Hamming Distance
- Minkowski Distance
- Manhattan Distance

Based on these distances clustering algorithms classified into two types. They are

- K Means clustering
- Hierarchical clustering

In the above figure, there are three clusters labeled as green, red, and blue. All green, red, and blue colored labels belong to individual classes.

### 4. Regularization Algorithms

The main parameters to be considered for the fitting curve on data in Machine learning algorithms are bias and variance. There may be a possibility of overfitting as well as underfitting. Overfitting causes due to imbalanced variance and bias. To avoid these imbalances regularization is introduced. Regularization makes coefficients or parameters of the learning curve to zero. The main regularization techniques are

- Lasso Regression or L2 Regression
- Ridge Regression or L1 Regression
- Dropout
- Data Augmentation
- Early Stopping

## Comments

## Post a Comment