Artificial Neural Network


Artificial intelligence is process of making computers program that can think like human, act like human and solve critical problems like human does through scientific calculations.

Now, lets assume if we are trying to mimick human brain and its behavior then we must know the detail human brain functions and components involved in solving critical problems.

Well we do not need to research of human in detail for learning Artificial neural networks however having concept is plus point for you to start with.

Human has neurons in the brain which is certain value or expression. As these neurons are connected with certain relationship. This relationship helps us to link the knowledge of one field to another field and learn to compare.

Similarly, Artificial Neural network contsins certain nodes representng value which are connected to others nodesthrough relationship/edges or links.

Each neuron are the computing element which have input from other nodes and output to another nodes.

How the connection or selection of neuron is possible, are there all connection of neuron from all the nurons?
– The selection of new coputing neurons depend on the activation function. An activation function calculate the total sum of input to a neuron and compare it with threshold value. If sum is less than threshold then a neuron has a output of -1 and if the sum is greater or equals to threshold then output of an neuron will be +1.

Since this activation function uses sign like -/+ to determine the output of and neuron so this is called sign function.

Above are the activation function of an neuron.

Source: Deep Learning on Medium