Machine Learning Project development step1: Supervised or Unsupervised learning

 

Machine Learning Project development step1: Supervised or Unsupervised learning

First step 1: Machine learning project development: Identify whether the project is coming under supervised learning or unsupervised learning.

Machine learning

Machine learning is a part of artificial intelligence (AI). It provides

Model  which learns from data (experience ) and predict, classify or group the data at the output. Applications of machine learning are as shown in figure1.

                                        Figure1:Applications of machine Learning

Machine learning algorithms are often categorized as supervised or unsupervised and reinforcement.

Therefore first step of machine learning  project is, identify whether the project  is coming under supervised learning or unsupervised learning.

If you have a labeled data then it is supervised learning. Supervised learning needs input and output to train the model.

If no labeled data, only inputs data available and need to group the similarities and for finding output then go for unsupervised learning.

Example of  supervised and unsupervised learning are as shown in Figure2

                                       Figure2: Supervised and Unsupervised Learning

In the first picture for  person identification model needs to train with a face as input and identity of the person as an output. Therefore it will come under supervised learning.

In the Face  recognition or thumb recognition for security or attendance purpose is an example of Supervised learning

In the second picture  person identity is not provided to the model. It will cluster similar persons in one group.

Photo tagging of an individual person on Facebook is an example of machine learning. Supervised and unsupervised learning is further divided into clustering classification or regression as shown in figure 3.

 

                    Figure 3. Supervised vs Unsupervised Machine Learning Problems

Supervised Machine Learning

Ø In Supervised learning, predictive  model has been trained  using data set which is well "labeled".

Ø Data set is available with input and output(numbers or labels).

Ø Trained   model   helps us to predict outcomes for unforeseen data. 

Based on the outcome/response or dependent variable, Supervised learning problems are categorized into "regression" and "classification" problems.

Regression: When the outcome or response variable is a continuous variable (numeric or number), it can be called as regression problems. In this input variables are mapped  to some continuous function

For Example:

1) To predict the house price from training data. The input variables will be locality, size of a house, etc.

2) To predict  things as future demand for their products. The input variables are previous experience (forecasting and optimization)

Classification: When the outcome or response the variable is a discrete variable (labels), it is called as classification problems. In this  input variables are mapped  into discrete categories.

For Example:

1.Character recognition, Pattern recognition, Handwriting, Biometric recognition e.g. face, finger etc,

2. Document classification

Clssification	Regression
In classification problem data is labeled into two or more classes	In regression te system attempt to predict the value for an input based on past data
I t is used to predict discrete class lable	It is used to predict continuous quantity
It is binary or Multi class  classification	A regression problem with multiple input variables is called a multivariate regression problem
e.g Classifying an email as spam or non spam	e.g predicting the price of stock over a period of time

In supervised learning training data set requires both Input and Output 

Input	Output	Application
Home features	Prize	Real estate
Image	Object Recognition	Photo Tagging
Audio	Text transcript	Speech Recognition
English	Japanese	Language translator
Radar info	Position of car	Autonomous driving car

Unsupervised learning

Ø Unsupervised learning is a machine learning technique, allow the model to work on its own to discover information.

Ø It mainly deals with the unlabeled data.

Unsupervised learning problems further grouped into clustering and association problems

Clustering: Clustering is an important concept when it comes to unsupervised learning. It mainly deals with finding a structure or pattern in a collection of uncategorized data. Clustering algorithms will process your data and find natural clusters (groups) if they exist in the data. Clustering is grouping a set of objects in such a manner that objects in the same group are more similar than those objects belonging to other groups

-Unsupervised machine learning finds all kinds of unknown patterns in data.

-Unsupervised methods help you to find features that can be useful for categorization

e.g Spam filter-You know the junk folder in your email inbox? It is the place, where emails have been identified as spam by the algorithm. 

Supervised Learning	Unsupervised Learning
In a supervised learning model, input and output variables will be given.	In unsupervised learning model, only input data will be given
Algorithms are trained using labeled data	Algorithms are used only input data which is not labeled
Supervised learning model uses training data to learn a link between the input and the outputs.	Unsupervised learning does not use output data.
Support vector machine, Neural network, Linear and logistics regression, random forest, and Classification trees	Unsupervised algorithms can be divided into different categories: like Cluster algorithms, K-means, Hierarchical clustering, etc.