Introduction to Neural Networks
This article will give you an basic understanding about how things are originated in Deep Learning and how Neural Network works actually ?
The most fundamental unit of Deep Neural Network is Artificial Neurons. The inspiration behind the word neurons came from biology and (more specically from brain). Our brain is connected with millions of neuron cells(1011) which helps us to take decisions.
Now the question is, how do these neuron cells are so capable to answer our questions and let us take decisions within a second.
For example, if you see a Car your eyes detect it quickly. If you listen to a song anywhere around you, your mind gives you an answer this is the song of a particular artist.
These are the things which all the millions, billions connected neurons perform by taking some input, processing it and finally giving you an output.
Biological neuron structure looks like this,
This is basically a single neuron cell , there are billions of brain cells connected like this.
1) Dendrites are used as input to receive the signals of neurons like we see an object through eyes, listen from ear etc.
2) Axon are used to transmit the output.
3) Nucleus is also called as soma used to process the input. Now you had seen an image but there should be something to process it and decide what actually the input is. Right? This is the function of Nucleus.
4) Synapses are used to connect one neuron to another neuron cell.
Consider the example when our brain neuron has to decide whether it should feel happy when seeing any married couple or not,
In this example, we had given an image of the married couple as an input to eyes, eyes give input to the first layer of neurons, then that first layer of neurons connects with other neurons to take decisions and finally gives an output based on its past learning.
All neurons do not perform same work. Each is designed to perform some function like one neuron may fire when seeing any image. One neuron may fire when listening some music etc.
Next article is about
This is how our brain works.
Similarly, an Artificial Neuron takes some inputs, then some activation function performs on given input and gives output.
Mc Culloch(neuroscientist) and Pitts(Logician) in 1943 proposed a highly computational model of neuron which do all the processing of inputs and gives only boolean output means either 0 or 1. We will cover this in our next article.
And question is, Does boolean outputs are possible in today's world? So which new model is then invented and all those details in next article.
Largest collection of up-to-date tutorials to learn programming languages. We are focused on easy learning. Massive collection of interview questions one may need for preparation.