There are different definitions of machine learning given by different authors.
Arthur Samuel(1958) has defined the machine learning as the ability of a computer to action without explicit programming.
Tom Mitchell(1998) has defined it as, the ability of computer program to learn from experience ‘E’ with respect to Task ‘T’ and performance measure ‘P’. The performance on task improves the machine experience.
Broadly machine learning can be classified into two categories:
Supervised learning is based on data set and its already known that how the output is going to look alike. There is a mapping between input and output.
There are two types of problems in supervised learning. Classification and regression.
Classification: Example of classification is Birds identifications. To train the machine on this we require dataset of bird images labeled with their species and identifying characteristics. A classification model drives some conclusion from observed values.
Regression: In regression, the output is a continuous or real value. E.g age of a person, income, predicting the price of a house based on area.
Note: Main difference between the two types of supervised learning is that the dependent attribute is numerical for regression and categorical for classification.
In the unsupervised learning, there is input but there is no mapping to the output i.e. no labeling of data exists. In this type of learning first data points are defined and then data is grouped into clusters to get the output. Let us try to understand this with social media data. In social media platforms, user data is grouped into approx. 70 clusters basis on age, demographics, sex, interests and life stages etc. This segregated data further is used to do Ad campaigns or email campaigns. Another real-life example is to recognize the handwritten alphabets. Suppose the alphabets are black and white in color and have size 30×30 Pixel. Thus the digits lie in the dimensions 30 x 30-=900 Pixels. Here with the cluster, we can group the images that are close together and conclude that they represent the same digit. Thus we can recognize the alphabets.
“K Means clustering” is one of the algorithms used for this grouping of data.
Deep learning is a kind of machine learning, focused on algorithms inspired by the structure and function of the brain called artificial neural networks. Deep learning emulates the approach which human beings use to gain knowledge.