New to Machine Learning? A beginner’s guide

Machine learning term was coined by Arthur Samuel and defined it as the field of study that gives computers the ability to learn without being explicitly programmed.

It falls under the study of artificial intelligence and focuses on algorithms that enable computer programs to learn from data and tasks it performs. It uses the application of some of the most advanced computer science and mathematical theories.

As learning itself is an integral part of artificial intelligence, Application of machine learning algorithms goes into almost every AI software system. Machine learning itself is a combination of statistics, mathematics, physics, theoretical computer science and more.

In the 1990’s a more formal definition of Machine learning was accepted which was defined by Tom Mitchell as follows.

A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E.

An Example

Consider a sorting algorithm, comprised of multiple algorithms like merge sort, quicksort etc which is continuously fed with unsorted data and is continuously storing the data in sorted order.

That task T will be

1.      Check the data if its valid and can be sorted?

2.      Check the type of data

3.      Select the best algorithm to run on this data

The performance P of the algorithm will be

a)      Correctness of sorting

b)     Time that is taken to sort the data

The experience E will be what the algorithm learns when performing task 1,2 and 3.

If the algorithm will be learning,

1.      It could predict if the data can be sorted, without really invoking a sort algorithm.

2.      It could predict a position by looking at the data it is sorting

3.      It can pick up the best algorithm to be used for sorting the data.

Although there are thousands of algorithms that can be used to accomplish the task, in today’s world a class of algorithms related to Neural Networks is widely used. And to use these algorithms, frameworks like tensorflow, Scikit, Keras are used. For extensive computational needs, cloud technologies, server clusters and CUDA is used.

For a small set of data, a PC is sufficient, but as the complexity and size of data grow, especially in scientific research and enterprise environments, specialized tools, techniques and infrastructure are used.

The most commonly used programming language for machine learning is Python along with its libraries. For simulation, modeling, and testing of algorithms, Matlab and Octave are used. For working with large data sets, big data technologies like Hadoop and Spark are used.

Common application of machine learning

·        Optical character recognition

·        Face detection

·        Spam filtering

·        Natural language processing

·        Medical diagnosis

·        Fraud detection

·        Weather prediction

·        Stock Analysis

·        and many more

Find more of my articles on Machine Learning and Data Science at https://njtrainingacademy.com/blog-streampage/