Page 123 - Artificial Intellegence_v2.0_Class_11
P. 123
How Does a Neural Network Work?
The node will assign a number called "weight" to each of the incoming connections. When the network is operational,
the node gets various data elements, varied numbers on each of its connections, and multiplies them by the appropriate
weights.
To learn, an artificial neural network must first understand what it has done incorrectly and what it is doing correctly.
This is referred to as feedback. Feedback is how we learn what is correct and incorrect, and it is also what an artificial
neural network requires to learn. This is where the similarities to the human brain begin to emerge.
For example, when learning to play tennis, you learn that if you hit the ball too hard, it will go out of the court and
you will lose a point; if you don't hit the ball hard enough, it will not go over the net; but if you hit it perfectly, it will
go onto the opposite side of the court and you will win a point. This is a famous example of feedback in which you
either lose or gain a point. Neural networks learn in the same way that the brain does, using a feedback mechanism
known as back-propagation (also called "backprop").
This is when you compare the network's output to the expected output and modify the weights of the connections
between the neurons in the network based on the difference between the outputs. This is accomplished by moving
backward from the output units through the hidden neurons to the input neurons. Back-propagation causes the
network to learn over time by shrinking the gap between the output and the intended output until the two exactly
match, at which point the neural network learns the proper output.
Hidden layers
Input layer
Output layer
Difference in
desired values
Backprop output layer
Experiential Learning
Video Session
Scan the QR code OR visit the following link to watch the video: Neural Networks
explained in 7 minutes
https://www.youtube.com/watch?v=vpOLiDyhNUA
What you learned from this video?
Introduction to AI 121

