Machine Learning Introduction

Content

• History

• What is Machine Learning

• Types of Machine Learning 

History

In 1940s neurophysiology Warren McCulloch and mathematician Walter Pitts wrote paper on neurons and the working of neurons. So they think to made model using electric circuits that work same as neurons and the born of Neural Network comes.

In 1952 Arthur Samuel, an American pioneer in the field of computer gaming and artificial intelligence created one game named checkers which learns its self when run. Arthur played that game hundred times against machine.

Arthur Samuel, coined the term "Machine Learning" in 1959 while at IBM. In 1960s the representative book of Machine Learning Research was the Nilsson's book on Learning Machines, that dealing mostly with machine learning for pattern classification. Future more in 1980s the reports was given on using teaching strategies so that a neural network learns to recognize 40 characters (26 letters, 10 digits, and 4 special symbols) from a computer terminal.

Now in 21st century, many businesses have accepted that machine learning will increase their business goal and revenue potentially. Because of this many Giant companies make heavy research in Machine Learning and Neural Network. So they can survive in today's competition.

Here is some Large Project and Research of ML:

• GoogleBrain (2010) - Google Brain is a deep learning artificial intelligence research team at Google. Formed in the early 2010s, Google Brain combines open-ended machine learning research with systems engineering and Google-scale computing resources.
• AlexNet (2012) - AlexNet is the name of a convolutional neural network (CNN), designed by Alex Krizhevsky, and published with Ilya Sutskever and Krizhevsky's doctoral advisor Geoffrey Hinton.
• DeepFace (2014) - This is a Deep Neural Network created by Facebook, which they claimed can recognize people with the same precision as a human can.
• DeepMind (2014) - This company was bought by Google, and can play basic video games to the same levels as humans. In 2016, it managed to beat a professional at the game Go, which is considered to be one the world’s most difficult board games.
• OpenAI (2015) - This is a non-profit organization created by Elon Musk and others, to create safe artificial intelligence that can benefit humanity.
• Amazon Machine Learning Platform (2015) - This is part of Amazon Web Services, and shows how most big companies want to get involved in machine learning. They say it drives many of their internal systems, from regularly used services such as search recommendations and Alexa, to more experimental ones like Prime Air and Amazon Go.

What is Machine Learning ?

Artificial Intelligence is field of computer science which enables machine to act like human and perform human related work. Machine Learning is application or sub-field of Artificial Intelligence which learns from data without any extra code.

Machine Learning is a type of  Artificial Intelligence(AI) that provide machines and computer to learn without any extra programming logic. Machine Learning is computer program which grows and change automatically when expose to new data.

According to Arthur Samuel : Machine Learning is a field of study that gives computers the ability to learn without being explicitly programmed.

Tom Mitchell said that Machine Learning is a computer program is said to learn form experience E with respect to task T and some performance measure P, if its performance on T,as measure by P,improves with experience E.

The field of machine learning is concerned with the question of how to construct computer programs that automatically improve with experience.

Here,
Task (T) : To Playing Cheese.
Performance measure (P) : Percentage of games won against opponent.
Experience (E) : Number of time game has been played and practice

Ex.
Suppose your email program watches which email you do or you do not mark as a spam, and based on that how to better fit the spam. What is the task T  in this setting?

T = Classify email is spam or not spam.
E = Watching you label emails as spam or not spam.
P = Then number of emails is correctly classified as spam or not spam.

Types of Machine Learning

There are two types of Machine Learning.

1. Supervised Learning 

2. Unsupervised Learning

 

Supervised Learning :

As we know that Machine Learning learns from new and existing data, so when data is already labeled then it call supervised learning. In machine learning the training data is already labeled the system know the answer of particular data and trained on that data.

In supervised ML the independent and depended both data has been given to model. The main objective of this algorithm to be able to learn by comparing its actual output with the taught outputs to find errors, and change the model accordingly. Supervised ML algorithms use patterns and matrix to predict the value on unlabeled data.

There are two types of supervised learning algorithms:

• Regression
• Classification

Regression

Regression is the problem where the result value is different or a real value. Regression doesn't give any labels from training data. It is used for continues data.

Ex. In house price prediction where we can calculate price of the house from the area of the house.




Classification

  Classification is a problem where result will always come from classified label. The output is come from predefined category. Ex. Yes or No, True or False, 0 or 1

Ex. In given graph X = spam mails, Y = not spam mail. Now any mail arrive then it will apply the algorithm and classify that the mail is Spam or Not Spam.







Unsupervised Learning

Now in many businesses all need to know the customer interest for their business, which market we have to target for our business, recommendation for users. To solve these problems unsupervised machine learning is used. In unsupervised learning there is no label with training data. Algorithm it self need to classify the data in different different cluster by identifying patterns from training data. The data is unlabeled and uncategorized in unsupervised learning. The output is depended on code algorithm.

Bellow the types of unsupervised learning

• Clustering
• Association

Clustering

 Clustering is the task of dividing the population or data points into a number of groups such that data points in the same groups are more similar to other data points in the same group and dissimilar to the data points in other groups. In clustering we group data in nearest matching cluster. 

In Machine Learning, data that are in the same group should have similar properties and/or features, while data in different groups should have highly dissimilar properties and/or features.

Ex.  
Classify user in different different cluster to provide them meaning full and usable health care data.
Classify market to sell particular product in particular market.


Association 

Association rule mining finds interesting associations and relationships among large sets of data items. This rule shows how frequently a item set occurs in a transaction. A typical example is Market Based Analysis.(from here)

Association rule will define relationship between a large data set and show that which item is there and which item is not there. 

Ex. Following  grocery data set
{milk, bread, paneer, egg, cake, diaper, baby cream}
{milk, bread, cake}
{paneer, egg}
{diaper, baby cream} 
{cake, diaper, baby cream}
{milk, bread}
{egg, cake}

In given data set if the customer buy milk then he/she will also buy bread in all cases same if customer buy diaper then he/she also buy baby cream. So there is the relation between diaper and baby cream.

Market Based Analysis is one of the key techniques used by large relations to show associations between items.It allows retailers to identify relationships between the items that people buy together frequently.(from here)