CHAPTER_2

HOME  /  LESSONS  /  CHAPTER_2

Chapter_2
An Artificial Neural Network that Memorizes Object Names

Step 1.  Run Program2Z.exe


Step 2.  Click the “Initialize” button (Button ①) to initialize the artificial neural network. When this button is clicked, all weights of the artificial network are initialized with random numbers.


Step 3.  Click on the “Convergence Rate” button, (Button ②) repeatedly to set it to 0.1. Each time you click, one of the four predefined values will appear.

Step 4.  Click the “Repeat” button (Button ③). With each click, this neural network will repeat learning 500 times. Click it four times to repeat the learning for 2000 times.


Step 5. Drag the drone photo and place it on the center of the screen. Check the value calculated by the trained artificial neural network in the Output column. The result will be like Drone – 0.8, Car – 0.1, etc. (④)

Results


After 1,000 iterations, Drone is input to the artificial neural network. 

It recalls 80% as Drone, 10% as Car, and 10% as Bicycle. If the neural network is trained for 2,000 iterations, its output is 100% for Drone and 0% for all the others.

KamelConsulting, LLC.

CEO. Byung S. Moon | Tel. +82-10-6627-4306 | byung.s.moon@gmail.com ㅣ Biz License 314-86-26086

2F, DICIA ,20 Daedeok-daero 512, Daejeon, S. Korea


2022 Kamel Consulting, LLC