Machine Learning Theory: An introduction #MachineLearning #DataScientist
In our day to day life, Machine Learning (ML) used in different applications to provide the intelligence using the data analytics field. Different techniques used in ML are natural language processing, image recognition and recommendation systems etc. in different domains. Machine Learning simply defined as a computer program which set to learn from the experience with the given training set. Different Use cases for ML are fraud detection, identifying the patterns, predicting the house price in a particular area in a time, recommendations engines, social media data analysis, IOT and new generation bots.
The ML broadly classified into Super vised and unsupervised machine learning. Super vised machine learning is to train on the predefined given data set. Unsupervised learning is to take data and then identify patterns or classification in the data set.
Supervised learning – The supervised learning which is to train based on the given data set, classified into
- Regression - Produces continuous output by mapping the given input variables.
- E.g.1.: Predicting house price given the different input values house size and price.
- E.g.2.: Predicting age of person in the given picture
- Classification – Produces the discrete outputs means mapping the input variable into different discrete values of output.
- E.g.1.: Predicting the house price whether the house sells for more or less than a particular price
- E.g.2.: Predicting the person is from high school, college or graduate from the given picture.
Unsupervised learning – Unsupervised learning is typically predicting the patterns and relations in the given data without any training input. These unsupervised learning algorithms uses clustering, dimensionality reduction and correlation techniques and completely different from supervised learning algorithms.
Neural Networks – When we have huge number of input variables, it is difficult to use any of supervised learning techniques, then neural networks can be used effectively.