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
   118   119   120   121   122   123   124   125   126   127   128