Artificaial Inteligence won’t need huge data by using this basic mathematics concept
Artificaila inteligence requires large amount of data to detect pattern in order to perform any required task. This deep learning concept is used in varity of fields.
The researchers at MIT found that distilling the most relevant information describing handwritten numbers in a dataset (known as MNIST) and puting them together greatly reduced the number of characters their AI system needed to learn to recognize letters in a new dataset. Instead of just showing it the number 3 thousands of times, they trained it to recognize that the target was a number that looked somewhat like the digit 8.
This way, it is trained to recognize numbers in a new way. Machine learning (kNN) transform this idea to graphical approach. So database is represented in form of XY co-ordinates. Thus, AI system is easily trained to place dots on a graph on the correct side of a line they had drawn without the need for a large dataset.
The system still requires a large amount of dataset to start the initial process.This is drawback of this method.
Category: AI