Background of Machine Learning (ML) you should know before start learn

-

Today I am going to talk about the differences between Machine Learning, AI, ANN, Mathematical models, and few libraries that can use to program using python. In most of the time, we get confused about the terminology of Machine Learning. This is the main reason that we cannot understand Machine Learning even we know how to code them.

Machine Learning (ML) is a subset of Artificial Intelligence (AI). The Machine Learning has subsets such as Artificial Neural Networks (ANN) and Mathematical Models. 

ANN can be categorized in to following two types,

  1. Shallow feed forward Neural Network
  2. Deep Neural Network (Also known as Deep Learning)

Shallow feed forward Neural Network has only one hidden layer while Deep Neural Network has multiple hidden layers. That is shown in the figures bellow.

Shallow feed forward Neural Network




Deep Neural Network (Deep Learning)

The following figure shows how these areas relate to each other,

Mainly there are two types of machine learning libraries can be used to code machine learning systems using python. They are,

  • Scikit Learn
  • TensorFlow

The Scikit Learn can be used to code Machine Learning Mathematical models such as Decision Tree, Random Forest, SVM, and Linear Regression etc. TensorFlow can be used to code Aritificial Neural Networks (ANN) such as Deep Learning.  

From these two types of python libraries, the easiest is the Scikit Learn library. To code using Scikit Learn, the user do not need any special knowledge about machine learning. They can just apply for the user case directly. In terms of TensorFlow, the user should have some clear knowledge about Neural Networks, Activation functions and back-propagation, etc.

There are so many algorithm types of Deep Learning (Deep Neural Network):

  • CNN (Convolutional Neural Network)
  • LSTM (Long Short Term Memory Networks)
  • RNN (Recurrent Neural Networks)
  • MLP (Multilayer Perceptrons)
  • GAN (Generative Adversarial Networks)
  • ..., etc

Convolutional Neural Network (CNN) is a deep learning algorithm that is mainly used for image classification applications. Recurrent Neural Networks (RNN) algorithm is better to used with sequential data. RNN is very similar to neuron activity in the human brain.

Comments

Post a Comment

Popular posts from this blog

Solar System

-
Image
Our solar system is mainly consisting of eight planets, Moons, our Sun and few dwarf planets. These planets are Mercury, Venus, our own planet Earth, Mars, Jupiter, Saturn, Uranus, and Neptune.   Solar System with Dwarf planet Pluto (Image Credit: Pixabay.com and NASA/JPL-Caltech ) And also we have Pluto the dwarf planet. Previously Pluto was believed as a planet but after August 2006, the International Astronomical Union (IAU) down graded the status of the Pluto to Dwarf planets. Also there is a asteroid belt in between Mars and the Jupiter.  In our solar system, the planets are categorized in to two. Inner Planets Outer Planets Inner Planets are innermost planets in other words planets which are closer to the Sun. And all of them are terrestrial planets.They have rocky and compact surface.Mercury, Venus, Earth, and Mars are inner planets. These terrestrial planets are quite smaller compared to other gas and icy giants but larger than dwarf planets. Outer planets are planets ...
Read more

Converting JWST raw FITS images into TIFF images before image process

-
Image
In a previous post , we have learnt about how to download images of James Webb Space Telescope using MAST portal database. Today let’s learn how to convert FITS images in to TIFF image type using a tool called “ FITS Liberator ”.    FITS Liberator Graphical User Interface Note that, this post is for “Pixinsight” users and there is another method to process these images without using “Pixinsight”. The FITS type images cannot be processed using graphics software such as “Adobe Photoshop”. Therefore we need to convert them in to TIFF type images. TIFF images are uncompressed large images without any data lose. Let’s download the FITS Liberator and install. So type fits liberator and click the first link in the search results.     Searching FITS Liberator in the browser. After that, click the link “Download FITS Liberator 4”, as shown in the figure below. Then you will direct in to another page which contains all the download links for any required operating system y...
Read more

Easiest way to download JWST raw image data

-
Image
In this post we are going to learn how to download specific raw image files of a celestial object which was taken by JWST (James Webb Space Telescope). It is easier to just download any image from the NASA or any other website but if you need to image process the raw data by your self, you need to download the raw image data of the JWST or any other mission from a database. MAST portal is one of the best image database you can find in the internet.   AstroView of MAST Portal  First thing you need to do is to search the keyword "mast portal" in the browser search.   Searching "mast portal"   Then click the first link in the search results as shown below. Then you will see the MAST portal website in the browser window.   The first link on the top of the search results     MAST Portal website in the browser window   After that you have to search the required celestial object in the search bar and click "Search" button. In this post I have shown how t...
Read more