what is an artificial neural network

A neural network is a set of algorithms that try to find basic relationships in a dataset using a process that mimics the work of the human brain. Consists of artificial neurons or nodes – Wikipedia, the free encyclopedia

what-is-an-neural-network

What’s a neural network?

Neuron means neuron. Simply put, neurons are the basic building blocks of the human brain. These neurons determine all of your life experiences, feelings, feelings, your entire personality.

Every decision you make in everyday life, big or small, is controlled by these neurons. So, a neural network is a network of neurons. Here it is, when we discussed neural networks in data science, it was mostly inspired by the structure of the human brain.

Another important point: individual neurons cannot achieve anything on their own. This is a collection of neurons in which the real magic happens.

Neural network types

Types of neural networks There are several types of neural networks. The most popular methods are:
1. Convolutional neural network (CNN): used for image recognition and classification.
2. Artificial neural network (ANN): used to compress images.
3. Restricted Boltzmann Machine (RBM): Used for various tasks such as regression classification and size reduction.
4. GAN Generative Adversarial Network: Used for fake message detection and face recognition.
5.Repetitive Neural Network (RNN): Used for speech recognition.
6. SOM (Self Organizing Map): Used for topology analysis.

Neural network applications

1)Face unlock
Almost all new Android phones these days use some type of face unlock to speed up the unlocking process. Basically, it uses CNN to identify individuals. Thus, over time, the more often you use face unlock, the better your results will be.


2)Google Assistant
Say “Ok Google” and think about how the Google Assistant wakes up. Don’t speak out loud. You can call another Google Assistant. :). It uses used RNN to find out the word “resurrection”.


3)Adaptive battery for Android OS
If your Android phone is running Android 9.0 or later, when you enter the Settings menu, you will see the Adaptive Battery option under the Battery section.

This is a very nice feature. This feature mainly uses Convolutional Neural Networks (CNN) to see which apps on your phone are using more power and limits those apps based on that.

Neural network and deep learning

In fact, deep learning is the new name for an artificial intelligence approach known as neural networks that have been around for over 70 years. Neural networks were first proposed by Warren McCall and Walter Pitts in 1944.

Warren McCall and Walter Pitts joined the Massachusetts Institute of Technology in 1952, opening the first cognitive science departments from time to time.

In a deep learning network, each layer of nodes is trained on a separate set of functions based on the output of the previous layer.

The more developed the neural network, the more complex the ability of a node to recognize when combining and recombining functions from previous levels.


Also read: Google Search testing new ‘Short videos’ Search carousel


Convolutional Neural Networks

Convolutional Neural Networks (CNN) are deep learning algorithms that can record input images and assign value to various aspects/objects of an image and distinguish them from each other.

ConvNet requires much less initial processing than other classification algorithms. Raw methods have designed and learned filters by hand, but ConvNets can examine these properties.

The ConvNet architecture is similar to the neural communication model of the human brain and is based on the organization of the visual cortex. Individual neurons respond to stimuli only in a limited area of the visual field, known as the receiving field. These sets of planes overlap each other and cover the entire visual area.

Neural network algorithm

There are various algorithms used to train neural networks with too many variations. Imagine an artificial neural network (ANN) to figure out how an artificial neural network works. Today we all know that there are 3 layers of neural networks.
The input layer

Hidden Layer

The output layer

Algorithms

Feedforward algorithm

Sigmoid – A common activation algorithm

Cost function

Backpropagation

Gradient descent – Applying the learning rate


Also read: Robots In Human’s life


 

Share the knowledge

Leave a Reply

Your email address will not be published. Required fields are marked *