Humans have always dreamed of creating intelligent machines. When programmable computers were developed, people wanted to make them intelligent.
The real challenge of ARTIFICIAL intelligence lies in formally describing problems and tasks.
One approach discussed in this book, which allows a computer to gain knowledge from experience and make sense of the world based on a hierarchical conceptual system, is called AI deep learning.
Early knowledge-based ai methods raised the question of whether Fred with a razor is a human being.
The performance of simple machine learning algorithms is largely dependent on the representation of data. However, it is not easy to obtain data characteristics for many tasks. The use of machine learning methods to discover ways to represent themselves is called “presentation learning.” The classic example is autoencoders, where the goal is to preserve as much information as possible from the data that passes through the encoder and decoder.
When designing or learning features, the goal is usually to isolate variation factors that can explain the data.
Learning the correct representation of data is one perspective of deep learning.
There are two ways to measure the depth of a model, one is the number of sequential instructions executed, and the other is to describe the depth of graphs to which concepts are associated.