Overview
Whenever you search for your favorite dish recipees on search engine, you get desired outputs with related images. After watching a DSA course on youtube, you are recommended with more DSA and interview questions. After listening to pop music by an artist on Spotify, you are recommended with more pop music. This behaviour mapping of an individual account is followed by algorithm written by ML engineers. This pattern mapping and reasonable output comes under domain of Machine Learning.
Applications of Machine Learning Machine learning engineer can operate in the domain of machine learning in two ways: 1.Applied ML & 2.Research ML Applied ML, as name suggests is a field where real world problems are identified and solved using Machine learning methods which are already formulated. Research ML, as name suggests is a field where the new nuances of Machine learning domain are explored. Reasearch goes deep in mathematics of how the model works and how it can be better.
What is Machine Learning? "Field of study that gives computers ability to learn without being explicitly programmed" ~Arthur Samuel
Types of Machine Learning:
- Supervised ML
- Unsupervised ML
- Reinforcement ML
- Recommender System
A. Supervised Learning You have an input X and an output Y. Take an example of guessing housing prices, where the input is sqmetre of house and output is $. We have a dataset which can give us an idea about how the prices of house are varying with increasing size and then we can fit the model with a graph that is bestfit. This is also called Regression. Supervised learning is a way that is most popular and widely used. Another problem can be of identifying answers in a binary format i.e. 0/1 , yes/no. Take an example of predicting a person has a benign or malignant cancer with available data of size of tumour. In this we can identify clusters and label them. This type of supervised learning is known classification model.
B. Unsupervised Learning In this type, the dataset are not given labels that are to be identified and is called clustering. Inputs X are present but Y are not present. For eg: Google news that shows similar keywords result when searched for.