Basic skills are always good, and a good foundation determines the quality and height of the superstructure.
Starting from today, I will continue to serialize some basics of deep learning, including concepts, mathematical principles, and codes. There is indeed no hot topic to discuss recently.
Let’s first look at the comparison between machine learning and deep learning:
"Data and features determine the upper limit of machine learning, while models and algorithms only approach this upper limit." One of the essential differences between machine learning and deep learning is feature engineering, and feature engineering is the most important factor in determining the quality of the final result. One of the factors; the top description of the above picture is the process of machine learning. If a computer understands that the input information is a car, then humans need to do a lot of feature engineering work, which is to extract some characteristic information of this thing and then hand it over to the machine. To recognize and process, the lower part of the above picture is the field of deep learning. The most important thing in the field of deep learning is that it can automatically extract the characteristics of the data and process it.
The second biggest difference between machine learning and deep learning is the amount of data (of course data quality is more critical to a certain extent and is beyond the scope of this lesson). The amount of data required by the two is basically several times or even dozens of times. A hundredfold relationship. Due to the amount of data, the computing power required for deep learning and machine learning are not of the same order of magnitude.
So how does deep learning implement feature extraction? This leads to