An introduction to Machine Learning


Machine Learning- Introduction

Machine learning (ML) is the scientific way of study of algorithms and statistical models that computer systems use in order to perform a particular task efficiently without using explicit instructions and command.

It can be also defined as discipline that interact with the programming systems so as to make them automatically learn and improve with experience

Machine learning is taken as a subgroup of artificial intelligence. This algorithms build a mathematical model based on sample input data, which is also called as  “training data”. This make the predictions without being explicitly programmed to perform the task.

Let’s clear it with example. You are trying to toss a paper to a dustbin. After first try, you understand that you have put too much force in it and after second attempt, you realize that you are closer to your target but you require to increase your throw angle.

You notice the actions happening here is basically after every throw you are learning something and improving the final result. We are programmed to learn from our experience and knowledge.


In the field of information analytics, ML can be used to direct complex models and algorithms that offer themselves to prediction. This is called as predictive analytics. This type of analytical models let researchers, data scientists, engineers, and analysts to “produce reliable, repeatable decisions and results” and discover “hidden insights” through learning from past historical relationships data sets.

Predictive_Analytics_ML

Suppose you want to go for a vacation. You browse through the different travel agency website and search for a hotel and restaurants. When you look at a specific best and hotel, just below the hotel description you will find there a section titled “You might also like these hotels for your vacation”.

This is a common use and implementation of Machine Learning. This is called “Recommendation Engine”.

Suppose you want to predict traffic patterns at a busy intersection, you can run it through a machine learning algorithm with data about past historic traffic patterns  and, if it has successfully “learned”, it will then do better at predicting future upcomming traffic patterns.
The complex nature of many real-life problems, though, often means that inventing particular algorithms that will resolve them perfectly every time is impractical, if not impossible.

There are many examples of ML.Machine learning problems include,

 “Is this cancer?”,

“Total benign tumour?”

“Which of these people are good friends with each other?”,

“Will this specific person like this movie?”

These problems are excellent examples for Machine Learning, and in fact machine learning has been applied such problems with great victory.


Before heading to this section

Read Getting started with Machine Learning

Classification of Machine Learning

Machine learning applications are classified into three major categories, on the basis of the learning “signal” or “response” available to a learning system which are as follows:-

  1. Supervised learning:

    In this type of learning process, algorithm learns from example data of data and associated target replies can consist of numeric values or string labels, such as classes or tags. In order to predict the correct response later when posed with new examples comes under the category of Supervised learning.

Supervised_learning-ML

This method is indeed similar to human learning process under the supervision of a teacher or parent or friends. They provides good examples for the student or there children’s to memorize, and the student then derives general rules from these specific examples.

  1. Unsupervised learning:

    In unsupervised learning process, an algorithm learns from plain examples without any related response, leaving to the algorithm to control and determines the data patterns on its own. This type of algorithm tends to structure the data again into something else, such as data with new features that may represent a class or a new series of un-related values. They are also useful in providing humans with intuitions into the meaning of data and new beneficial inputs to supervised machine learning algorithms.
    As a kind of learning, it represents the methods humans use to work out that certain objects or events are from the same class, such as by observing the degree of similarity between objects. Some reference systems that you find on the web in the form of marketing automation are based on this type of learning.

 

  1. Reinforcement learning: 

When we define Reinforcement learning in the context of artificial intelligence, it is a type of dynamic programming that trains algorithms using a system of prize and punishment.

Reinforcement_learning-ML

A reinforcement learning algorithm learns by interacting with its environment and receives rewards by performing correctly and consequences for performing incorrectly. The algorithm learns without interference from a human by maximizing its reward and minimizing its penalty.

There are fabulous work on applying Reinforcement learning in Robotics. RL In particular, trained a robot to make learn policies and laws to map raw video images to robot’s actions. The RGB pictures were fed to a Convolutional neural networks (CNN) and outputs were the motor torques. The RL section was the guided policy search to generate training data that came from its own state circulation.

 

  1. Semi-supervised learning:

    Semi-supervised learning is a class of machine learning(ML) tasks and techniques that also make use of unlabeled data for training – typically a small amount of labeled data with a huge amount of unlabeled data.

Categorizing on the basis of required Output

On the basis of required output, machine learning can be classified as :

  1. Classification:

    When inputs are divided into two or more classes, and the learner must produce a model that allocates unseen inputs to one or more of these classes. This is typically undertaken in a supervised way. For eg : Spam filtering. In this the inputs are email (or other) messages and the classes are “spam” and “not spam”.
  2. Regression :

    Which is also called as a supervised problem, A case when the outputs are constant rather than discrete.
  3. Clustering :

    When we divides the input into the groups. Unlike in classification, the groups are not known early, making this typically an unsupervised task.

Machine Learning comes into the picture when problems cannot be resolved by means of typical methods.

 

Applications of Machine Learning

Machine learning can be used in the various fields of science. Some of the application of machine learning algorithm are given below:

  • Vision processing
  • Pattern recognition
  • Data mining
  • Language processing
  • Forecasting things like stock market trends, weather
  • Games
  • Robotics
  • Expert systems

machine learning application machine learning applications

What are the steps Involved in Machine Learning

A machine learning project involves the following procedure −

  • Defining a Problem
  • Preparing Data
  • Presenting Results
  • Evaluating Algorithms
  • Improving Results

 

 

Sending
User Review
0 (0 votes)

Leave a Reply